Low-temperature LPCVD of polycrystalline gex si1-x films with high germanium content

Article abstract of DOI:10.1149/1.2177006

A study of low-pressure chemical vapor deposition (LPCVD) of Gex Si1-x (x<0.65) films was carried out. The films were deposited on thermally oxidized silicon wafers at deposition temperatures of 430-480°C. Pure Ge H4 and Si H4 (or Si2 H6) gas sources were used. It has been found that the reactive sticking probability ratio between Ge H4 and Si H4 varies between 11 and 3.5, depending on both the deposition temperature and film composition. An increase of germanium content from 40 to 60 atom % caused a sharp increase of deposition rate. The polycrystalline layers mainly contained (110)- and (111)-oriented grains. The (111)-oriented grains were overgrown by the (110)-oriented grains at 430°C after a film thickness of about 100 nm. Reducing the total pressure below 10 Pa led to a dominance of (111)-oriented grains. A relative Ge H4 depletion in the downstream direction influenced both the composition and crystallinity of the layers, and the deposition rate. To minimize the depletion effect, Si2 H6 has been used instead of Si H4. No improvement has been obtained but the depletion from Si2 H6 occurred instead of the depletion from Ge H4. A proper mixture of Si H4, Si2 H6, and Ge H4 can be suggested to compensate the depletion effects.

Full text of DOI:10.1149/1.2177006

Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
G363
0013-4651/2006/153͑5͒/G363/9/$20.00 © The Electrochemical Society
Low-Temperature LPCVD of Polycrystalline GexSi1−x Films
with High Germanium Content
,z
*
Alexey Kovalgin and Jisk Holleman
MESAϩ Research Institute, Chair of Semiconductor Components, University of Twente, 7500 AE Enschede,
The Netherlands
A study of low-pressure chemical vapor deposition ͑LPCVD͒ of GexSi1−x ͑x Ͻ 0.65͒ films was carried out. The films were
deposited on thermally oxidized silicon wafers at deposition temperatures of 430–480°C. Pure GeH4 and SiH4 ͑or Si2H6͒ gas
sources were used. It has been found that the reactive sticking probability ratio between GeH4 and SiH4 varies between 11 and 3.5,
depending on both the deposition temperature and film composition. An increase of germanium content from 40 to 60 atom %
caused a sharp increase of deposition rate. The polycrystalline layers mainly contained ͑110͒- and ͑111͒-oriented grains. The
͑111͒-oriented grains were overgrown by the ͑110͒-oriented grains at 430°C after a film thickness of about 100 nm. Reducing the
total pressure below 10 Pa led to a dominance of ͑111͒-oriented grains. A relative GeH4 depletion in the downstream direction
influenced both the composition and crystallinity of the layers, and the deposition rate. To minimize the depletion effect, Si2H6 has
been used instead of SiH4. No improvement has been obtained but the depletion from Si2H6 occurred instead of the depletion from
GeH4. A proper mixture of SiH4, Si2H6, and GeH4 can be suggested to compensate the depletion effects.
© 2006 The Electrochemical Society. ͓DOI: 10.1149/1.2177006͔ All rights reserved.
Manuscript submitted May 31, 2005; revised manuscript received September 30, 2005. Available electronically March 8, 2006.
distribution, a fast diffusion along the grain boundaries is necessary,
which is most rapid in material with columnar grains. It has been
noticed for poly-Si that maximizing the ͑110͒ texture is important
for applications requiring rapid diffusion.30 Third, a higher reactivity
of GeH4 compared to SiH4 leads to a depletion of the gas mixture
from GeH4 in downstream direction of the reactor. This gradual
change in the gas-phase composition ͑i.e., decrease of the
GeH4/SiH4 partial pressure ratio͒ causes the decrease of germanium
content in the films grown at different positions in a long wafer
boat.31 Film properties ͑texture, morphology, etc.͒ strongly depend
on film composition and, therefore, are influenced by the depletion
from germane.32 In addition, the deposition kinetics of poly-
GexSi1−x films is strongly influenced by film composition, especially
for the films containing a high fraction of germanium,33 and up to
date, a study of such a deposition at lower temperatures down to
430°C is lacking.
In this work, a study of LPCVD of GexSi1−x films with germa-
nium content up to 65 atom % has been carried out. The growth
kinetics, composition, and texture of GexSi1−x ͑x = 0.3–0.65͒ films
have been studied vs deposition parameters. Pure GeH4 and SiH4
gas sources were mainly used for the experiments. Pure GeH4 and
Si2H6 gas sources have also been investigated as potential precur-
sors for deposition of GexSi1−x layers. For both precursor systems,
the distributions of deposition rate and film composition along the
wafer boat in a batch-type reactor have been obtained. Optimum
deposition conditions resulting in high-quality, ͑110͒-oriented poly-
GexSi1−x layers are proposed.
Polycrystalline germanium–silicon ͑GexSi1−x͒ is being investi-
gated for several applications in integrated circuits. It has been pro-
posed as a promising alternative to the conventional gate material
͑polycrystalline silicon͒ because it can partially mitigate the gate
depletion problem in deep submicrometer complementary metal ox-
ide semiconductors ͑CMOS͒.1-5 It has a smaller bandgap with re-
spect to silicon,6 lower activation temperature for the same percent-
age of activated dopants,7,8 and lower diffusivity for certain
dopants.9,10 A lower resistivity for p-type layers can be achieved,
because GexSi1−x can incorporate more B than pure Si due to strain
compensation.11,12 It is further considered as an attractive material
for deep submicrometer CMOS gate electrodes due to its better
work function of p-type doped GexSi1−x compared to poly-Si and
high compatibility to CMOS processing.13-19 One can manipulate
the bandgap of the material by varying the germanium content, add-
ing a degree of freedom in the design of a MOS transistor. Nearly all
changes in bandgap are translated into a band offset at the valence
band side. This makes the Fermi level of the material in case of
p-type doping moving toward the mid-gap direction of silicon.
Germanium–silicon heterostructures on silicon have been studied
extensively in the sense of integration of strained GeSi into the
current CMOS process.20 Such heterostructures are expected to play
a key role in realizing silicon-based quantum electronics. Further-
more, polycrystalline GexSi1−x alloys have been recently investi-
gated as a microstructural material21-25 to replace oxide as the sac-
rificial layer in microelectromechanical systems ͑MEMS͒
technology. Finally, silicon–germanium nanodots embedded in an
insulator have potential application for room-temperature operation
of single-electron transistor memories and nonvolatile memory
devices.26-28 Embedding of GeSi nanodots in the SiO2 matrix results
in longer retention time compared to Si.29
Experimental
A 25-nm-thick SiO2 layer was grown by means of thermal oxi-
dation at 950°C in a dry-O2 ambient on ͑100͒-oriented n-type Si
wafers with a resistivity of 2–5 ⍀ cm. The deposition experiments
were started within 30 min after the oxide growth. To avoid the low
nucleation rate of GexSi1−x nuclei on the SiO2 surface, a few-
nanometer-thick amorphous silicon layer was predeposited using the
same LPCVD system. The deposition of GexSi1−x films followed
immediately, without vacuum break.
A hotwall horizontal LPCVD system has been used for the ex-
periments. The one-meter-length furnace contained three indepen-
dent resistive heaters to ensure a close-to-flat ͑ 2°C͒ temperature
distribution in the 30-cm central part of the furnace, where the wafer
boat with 70 wafers was situated. The space between the wafers was
0.5 cm. High-purity SiH4 ͑or Si2H6͒ and GeH4 were mixed in the
gas line before introduction into the reaction tube. The gas flows
were varied between 0–37 and 0–88 sccm for GeH4 and SiH4, re-
spectively. The total pressure ranged between 3 and 200 Pa. When
Polycrystalline GexSi1−x can be grown from a mixture of SiH4
͑or Si2H6͒ and GeH4 as reactive gases in a conventional low-
pressure chemical vapor deposition ͑LPCVD͒ system. Several major
problems make the growth of polycrystalline GexSi1−x more difficult
than poly-Si growth. One problem that arises is the low nucleation
rate of GexSi1−x nuclei on the SiO2 surface. The incubation time, i.e.,
the time where constant deposition rate is reached, depends on depo-
sition conditions. Such a point of constant growth can only be
reached when the nuclei layer is closed. This time depends on nucle-
ation rate and growth rate of the nuclei and varies from a few sec-
onds up to tens of minutes. The second important problem corre-
sponds to the texture of the films. To obtain a uniform dopant
*
Electrochemical Society Active Member.
z E-mail: a.y.kovalgin@utwente.nl
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

G364
Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
using Si2H6, the upper limit for the gas flow did not exceed
10 sccm. If not mentioned in the text, no extra dilution of the reac-
tive gases with, e.g., N2 or H2 was done.
The thickness of the deposited GexSi1−x films ranged between 30
and 300 nm. The film thickness measurements were carried out by
means of a DEKTAK 3030 profile meter on an etched step. The step
was etched by using an HNO3/HF-based etchant for poly-Si with
the measured selectivity of about 10 to the underlying SiO2 layer.
Considering the average overetch time of 30 s ͑i.e., up to 1.2 nm of
SiO2 can be removed͒, the inaccuracy of the thickness measure-
ments did not exceed 4% for a 30-nm-thick film and 0.4% for a
300-nm-thick film.
The germanium fraction in the layers was measured using an
energy-dispersive X-ray analyzer ͑EDX͒ calibrated with standard
pure silicon and pure germanium samples. The layer texture was
obtained with a Philips X-ray diffractometer ͑XRD͒, irradiating the
Cu K␣ line of 1.5406 Å. Based on the ͑220͒-XRD peak position, it
was possible to measure the germanium content in a thin ͑down to
50 nm͒ GexSi1−x layer.
Figure 1. Dependence of germanium content on GeH4 fraction in the gas
phase for GexSi1−x films deposited at 430°C. Total pressures: ͑1͒ 65 and ͑2͒
20 Pa; GeH4 flow rates: ͑1͒ 9 and ͑2͒ 37 sccm.
Ϸ exp͑−EA,GeH /kT͒/exp͑−EA,SiH /kT͒. Because EA,GeH and EA,SiH
Germanium Content and Reactive Sticking Probability Ratio
4
do not necessar4ily have the same4value, the reactive st4icking prob-
Germanium content is an important characteristic of GexSi1−x
films because it determines the valence band position and therefore
the electrical properties of the material. According to Ref. 34, the
germanium fraction x in GexSi1−x alloy can be calculated as
ability ratio ␥GeH /␥SiH should depend on temperature. The activa-
tion energies for4 chemisorption are low though, e.g., for SiH4
a
4
value of 0.15 eV was proposed34 as an average of values obtained
from literature.35-40 The activation energy for GeH4 chemisorption is
expected to be even lower because of the weaker Ge–H bond and
thus, easier dissociation. Such low values for the activation energies
let us conclude that the temperature dependence for the reactive
sticking probability ratio should not be very strong. It has already
been demonstrated in Ref. 34 that for a deposition temperature in the
range of 550–650°C, the composition of the GexSi1−x alloy as a
function of gas composition for the condition of undepleted growth
had little dependence on temperature. This was also in agreement
with the results published by Kuiper and Lathouwers41 and Robbins
et al.42
kGeH
4
MFGeH
4 kSiH
4
x =
͓1͔
kGeH
4
MFSiH + MFGeH
4 kSiH
4
4
where MFGeH and MFSiH are the mass flow-rates of GeH4 and
4
4
SiH4, respectively, and kGeH and kSiH are the adsorption rate con-
4
stants ͑i.e., adsorption probabilities͒ t4o free surface sites. The ad-
sorption rate constants are related to the reactive sticking probabil-
ties ͑␥GeH and ␥SiH ͒ to free surface sites through
In this work, the deposition experiments were carried out at
much lower temperature and for films with much higher germanium
content than elsewhere. Comparison of the available data makes it
possible to investigate the sticking probability ratio not only within
a large temperature interval but also over the extended range of film
composition. Figure 2 plots the ␥GeH /␥SiH ratio, calculated using
4
4
kT
kGeH = ␥GeH
͓2͔
ͱ
ͱ
4
4
2␲MGeH
4
4
4
kT
Eq. 1-3, vs germanium content in the films. To calculate the ratio,
the measured composition of the layer X is put into Eq. 1, which
leads with known mass flows MFGeH and MFSiH to the ratio of
kSiH = ␥SiH
͓3͔
4
4
2␲MSiH
4
4
4
and ␥SiH = ␥0 exp͑−EA,SiH /kT͒, where k is the Bolzmann’s con-
stant, MG4eH and MSiH are4the molecular masses of GeH4 or SiH4,
reaction rate constants kGeH /kSiH . According to Eq. 2 and 3, the
4
latter is proportional to the 4ratio of reactive sticking probabilities.
Two observations can be made here. First, in the entire temperature
range studied ͑430–650°C͒, the composition of the GexSi1−x films in
principle changes with temperature at a fixed ratio of the source
gases. For the films having more than 50 atom % germanium such
changes are less pronounced than for the films with, e.g., less than
30% germanium. Second, the ratio of sticking probabilities is af-
fected by the film composition at a fixed deposition temperature.
This could be due to the different activity of Ge sites compared to Si
sites. For example, if surface germanium atoms are chemically more
active than surface silicon atoms, silane prefers adsorption on ger-
manium sites with respect to silicon sites. In this case, the higher the
concentration of germanium sites, the higher the relative adsorption
of silane and the lower the sticking probability ratio.
respectively4, ␥0 is the 4pre-exponential factor, and EA,SiH is the ac-
4
tivation energy for chemisorption of SiH4. A similar definition holds
for ␥GeH with the remark that the pre-exponential factor should not
4
be the same due to the different reactivities of GeH4 and SiH4. Note
that ␥GeH and ␥SiH refer here to the reactive sticking probability to
4
4
free surface sites.
In accordance with Eq. 1, the germane fraction in the gas phase
crucially influences the germanium content in the films. The germa-
nium content, as presented in Fig. 1, was obtained for the first mea-
suring wafer in the wafer boat, which corresponded to the condition
of undepleted growth at a temperature of 430°C. Due to the signifi-
cantly higher adsorption rate constant for GeH4 with respect to that
for SiH4,34 GeH4 adsorbs more intensively compared to SiH4, i.e., a
30% gas-phase share of germane results in about 60 atom % of
germanium in the films. From Eq. 1, this corresponds to a ratio of
the adsorption rate constants kGeH /kSiH of 3.6, and is in close
For a fixed film composition of, e.g., 30%, the lowest deposition
temperature of 430°C ensures a sticking probability ratio as high as
11 compared to the much lower value of about 5 obtained at
550–650°C. The consequence is that at fixed gas-flow ratios, the
deposited film becomes germanium-rich with decreasing the tem-
perature from 550 to 430°C. This can be explained by assuming that
4
4
agreement to the ratio of 3:1 calculated by Holleman et al.34 for a
deposition temperature in the range of 550–650°C.
The adsorption rate constant ratio is expected to depend
on the ratio of the exponential terms for ␥ which contain the acti-
vation energy for SiH4 and GeH4 chemisorption. So ␥GeH /␥SiH
the activation energy for chemisorption of SiH4, EA,SiH , is larger
4
then EA,GeH for GeH4, as was suggested earlier in this paper. This
4
4
4
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
G365
Table I. The activation energy EA,GeSi calculated for different
deposition conditions and varying film composition x. The activa-
tion energy has been extracted from the slope of the Arrhenius
plot in the deposition temperature range between 430 and 480°C.
Total
GeH4
SiH4
GeH4
Ge
EA,GeSi Amorphous
Pressure Pressure Pressure flow contenta energy
or
͑Pa͒
͑Pa͒
͑Pa͒ ͑sccm͒ ͑atom %͒ ͑eV͒
crystalline
9
11
20
1
1
5
9
9
8
10
15
21
91
9
9
9
9
9
9
9
37–42
31–32
48–55
49–57
34–38
33–35
34–35
1.5
1.5
1.0
0.8
Cryst.
Cryst.
Cryst.
Cryst.
30
100
150
200
1.3 Cryst./Amorph.
1.1
0.7
14
18
136
182
Amorph.
Amorph.
a For each indicated range of Ge content in the films, the highest con-
tent corresponds to the lowest deposition temperature ͑i.e., 430°C͒.
Figure 2. Reactive sticking probability ratio ␥GeH /␥SiH vs germanium con-
4
4
tent in GexSi1−x films obtained for a broad range of deposition conditions.
͑ᮀ͒ Total pressure 0.65 or 0.2 mbar; GeH4 flow rate varies from
6 to 37 sccm and SiH4 flow rate ranges between 20 and 86 sccm such that
the GeH4 fraction in the gas phase changes from 7 to 65%. ͑᭺͒ Total pres-
sure 0.09–0.3 mbar; GeH4 flow rate 9 sccm; SiH4 flow rate 86, 60, and
21 sccm. ͑᭝͒ Total pressure 0.09–0.3 mbar; GeH4 flow rate 9 sccm; SiH4
flow rate 86, 60, 26, and 21 sccm. ͑छ͒ Total pressure 1 mbar, GeH4 flow
rate varies in the range 2–30 sccm, SiH4 flow rate ranges between 40 and
100 sccm.
*
*
␪
/␪ = ͑␪*Ge/␪ ͒exp͑− ⌬E/kT͒
͓6͔
H
H
*Si
Ge
Si
where ⌬E is the difference in bond energy of a hydrogen atom
adsorbed to a Si site ͑i.e., HS*i͒ and a hydrogen atom adsorbed to a
Ge site ͑i.e., H*Ge͒. ␪*Ge/␪ represents a surface ratio between Ge
*Si
and Si sites. ⌬E was calculated to be 0.3 eV.34
For pure silicon grown from SiH4 at a low deposition tempera-
ture of 480°C, the deposition rate rSi is below one nanometer per
min. For a lower temperature of 430°C, the deposition rate de-
creases so much that it becomes impossible to detect the layer after
half an hour of processing. At such low temperatures the surface is
covered with hydrogen and its desorption is the rate-limiting step.34
As the presence of germanium sites enhances the hydrogen desorp-
tion, deposition of pure germanium at the lowest temperature of
430°C occurs without the limitations given by hydrogen desorption.
At such low temperatures the deposition kinetics of poly-GexSi1−x
films therefore shifts from the hydrogen-desorption limiting regime
to adsorption/supply-limiting regime. In other words, rGeSi is mostly
would explain that there is a general trend in Fig. 2 toward higher
ratios for lower temperatures. When these values also depend on the
sites to which they bond, it can explain the observation of lower
ratios at higher Ge contents.
It has been found that the total pressure did not affect signifi-
cantly the germanium content and therefore ␥GeH /␥SiH ratio for
4
both the polycrystalline and amorphous GexSi1−x films. 4The same
held for the transition region from the amorphous to polycrystalline
phase. The measured points deviated within a few atomic percent for
different total/partial pressures.
determined by ␪ for the films having low x values, and by kGeH
and kSiH for the films with high x. Although the deposition kinetics
*
4
4
of poly-GexSi1−x films is strongly influenced by film composition,
there is a lack of publications studying this dependence at such a
low temperature.
On Deposition Kinetics
Poly-GexSi1−x deposition using SiH4 and GeH4 as source gases is
dominated by Langmuir–Hinshelwood heterogeneous decomposi-
tion at low partial pressures. As proposed by Gates et al.35 for the
decomposition sequence of SiH4, the surface decomposition of AH4
͑A = Ge,Si͒ can initially occur with the formation of AH*3 and H*.34
Here the sign * means that the species are bound to the surface sites.
Such a scheme requires a dual site, i.e., two free surface sites at
neighboring positions. The next step is the decomposition of AH3*
resulting in the formation of AH*2 and H*. The deposition rates rSi
and rGe of Si and Ge, respectively, are according to this decompo-
sition sequence given by
The overall activation energy EA,GeSi ͑i.e., a superposition of the
activation energies used in Eq. 2-6͒ can experimentally be extracted
from the slope of the Arrhenius plot using the set of rGeSi Ϸ exp͑
−EA,GeSi/kT͒, where rGeSi = rGe + rSi is the GeSi growth rate. For
the activation energy for hydrogen desorption from pure Si ͑x = 0͒,
values in the range of 1.8–1.95 eV have been proposed.34,35,39,40,45,46
For hydrogen desorption from pure Ge ͑x = 1͒ surfaces the activa-
tion energies between 1.35 and 1.45 eV have been calculated.34,47
As discussed above, the activation energy for GeH4 chemisorption
should be lower than that for SiH4, i.e., lower than 0.15 eV. So
depending on both x and deposition temperature, one can find values
between 1.95 and Ͻ0.15 eV for the growth activation energy. For
the hydrogen-desorption-limited polysilicon deposition in the tem-
perature range 550–650°C, the activation energy ranging between
1.6 and 1.7 eV is normally extracted from the Arrhenius plot.30,48
Such a lower value compared to 1.95 eV can be caused, e.g., by
insignificant reagent supply-related effects ͑i.e., diffusion and deple-
tion͒ in the reactor.
2
rSi = kSiH ͓SiH4͔␪
͓4͔
͓5͔
*
4
2
*
rGe = kGeH ͓GeH4͔␪
4
where ͓SiH4͔ ͑or ͓GeH4͔͒ is the concentration of SiH4 ͑or GeH4͒
The values for the activation energy, calculated in this work for
the GexSi1−x ͑x Ϸ 0.3–0.6͒ films, are listed in Table I. These values
confirm qualitatively that the growth regime moves from hydrogen-
desorption-limited growth toward chemisorption-limited growth
with increasing x. Namely, the highest EA,GeSi of about 1.5 eV ͑i.e.,
hydrogen desorption limitation͒ was calculated for the polycrystal-
line material with x varying between 0.3 and 0.4. For the higher x
molecules, and ␪ is the fraction of free surface sites. Admission of
*
GeH4 into the reaction chamber enhances the hydrogen desorption
due to catalytic action of adsorbed germanium in this process43,44
and therefore increases the fraction of free surface sites. When the
pre-exponential factors of the desorption rate constants of H2 from
Si and Ge sites are equal then the distribution of adsorbed hydrogen
34
H* over Ge sites ͑␪ ͒ and over Si sites ͑␪ ͒ is given by
*
Ge
*
H
H
Si
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

G366
Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
Figure 3. Dependence of deposition rate on germanium content x for
GexSi1−x films deposited at 430°C. Total pressures: ͑᭡͒ 9, ͑᭿͒ 20, and ͑ࡗ͒
65 Pa; GeH4 flow rates: ͑᭡, ࡗ͒ 9 and ͑᭿͒ 13–37 sccm; SiH4 flow rates: ͑᭡,
ࡗ͒ 60–21 and ͑᭿͒ 88–0 sccm.
Figure 4. Plots of deposition rate vs total pressure for GexSi1−x films depos-
ited at 430°C. GeH4 flow rate: ͑1͒ 9 and ͑2͒ 37 sccm; SiH4 flow rate: ͑1͒ 21
and ͑2͒ 86 sccm. No dilution with other gases.
to a homogeneous polymer formation are probably unlikely in this
system under the pressure range studied, because these reactions
would cause inhomogeneous and rough surfaces, which was not the
case. The films were reflecting with roughness better then 6 nm and
the uniformity was better than 1.5%.
= 0.48–0.57, the activation energy decreased to 1.0 and further to
0.8 eV. In our opinion, for x beyond 0.5 the slope is no longer
related to the activation energy for hydrogen desorption. In this
light, the deposition kinetics shifts to adsorption/supply-limiting re-
gime because the surface intensively looses its hydrogen coverage.
Based on Eq. 6 and taking the lower desorption energy of H from
Ge sites compared to H from Si sites into account it can be argued
that the fraction of free sites is mostly composed of Ge sites. The
number of Ge–Ge sites is simply proportional to x2. Therefore,
based on Eq. 4 and 5 one may expect a quadratic increase of growth
on the value of x. From Fig. 3 it can be seen that for x below 0.5, the
deposition rate can be described by a quadratic dependence in a
satisfactory manner. However, the rate enhancement seems to be
more dramatic with increasing x beyond 0.5. From the data fit
shown in Fig. 3, one can notice that the presence of triple ͑i.e.,
rGeSi Ϸ x3͒ and quadruple ͑i.e., rGeSi Ϸ x4͒ sites can better describe
the experimental dependence for x Ͼ 0.5. To explain this let us as-
sume that in addition to the decomposition scheme in the above
requiring two free surface sites, the heterogeneous surface decom-
position of AH4 ͑A = Ge,Si͒ can better occur on three free surface
Two observations can be made in Fig. 4. First, the difference
between the two curves gradually increases with increasing pres-
sure. This means that the deposition at the lower gas flows ͑i.e.,
curve 1͒ is suffering more from the precursor depletion effect. The
shape of curve 1 allows for better distinguishing the zero-order part
when the surface is fully covered with hydrogen. The increase of the
deposition rate due to gas-phase reactions is clearly observed at
pressures above 100 Pa. The second observation is that for the curve
measured at the higher gas flows ͑i.e., curve 2͒, the enhancement of
the deposition rate due to possible gas-phase reactions occurs at
much lower pressures ͑i.e., lower than 50 Pa͒ compared to curve 1.
The latter confirms that curve 1 is rather affected by the depletion
effect and the actual precursor partial pressure is therefore lower
than that calculated on the basis of the gas flows and the total pres-
sure.
As mentioned, with increasing the total pressure above approxi-
mately 100 Pa the material becomes amorphous. The deposition at
such pressures leads to lower activation energies, as it can be con-
cluded from Table I. The activation energy decreases to 1.3 eV with
respect to 1.5 eV for the similar-composed polycrystalline material,
and further to about 0.7 eV. This is because the formation of gas-
phase products together with the expected depletion effects start to
govern the overall activation energy.
sites at neighboring positions: one site will be occupied by AH*2 and
two sites by H*. Such an adsorption of two hydrogen atoms at
neighboring positions should enhance the following desorption of
H2 due to the decreasing role of surface migration. For the chemi-
sorption on four free surface sites at neighboring positions, this ef-
fect should even be more pronounced.
When the surface gradually becomes free of H* with increasing
x, it also enhances the surface migration of the Ge and Si adatoms.
They can now move to positions of the lowest energy, which is the
crystalline form rather than the amorphous form. The point corre-
sponding to 100% germanium in Fig. 3 ͑no hydrogen coverage ex-
pected͒ probably suffered from the precursor depletion effect due to
the high deposition rate.
The dependencies of deposition rate on total gas pressure ͑Ptot͒
for a SiH4–GeH4 system allow distinguishing several parts in the
curves, where the film formation is governed by various mecha-
nisms ͑Fig. 4͒. The dependencies can be explained in terms of a
Langmuir–Hinshelwood model being linear at low pressure and zero
order at high pressure when the surface is fully covered with hydro-
gen. A further increase of total pressure enhances the growth rate
and formation of the amorphous material because of the appearance
of highly active species like GeHx or SiHx due to gas-phase
reactions.49 These species can easily insert or react with adsorbed
hydrogen and by doing so increase the deposition rate. This means
that the hydrogen desorption becomes no longer the rate-limiting
step. Gas-phase reactions do not immediately cause the film to be
inhomogeneous because no high-order polysilanes but low-order
molecules like Si2H6 appear. High-order gas-phase reactions leading
Film Texture and Morphology
There are several important parameters determining the orienta-
tion of GexSi1−x crystallites to be grown.50 Dependence on deposi-
tion temperature refers to surface mobility and hydrogen desorption.
Gas-flow ratio governs film composition, which certainly influences
texture from a hydrogen desorption point of view. Deposition pres-
sure affects the nucleation process because of the collision fre-
quency with the surface. High pressures cause formation of the
amorphous material due to gas-phase reactions.
It appears that poly-GexSi1−x films deposited at 430°C typically
contain both ͑110͒- and ͑111͒-oriented grains. In some spectra a
weak ͑311͒ peak can be observed. Both silicon and germanium have
a diamond lattice with the lattice constant of 5.42 and 5.65 Å,
respectively.51 Using Vergard’s law the GexSi1−x lattice can be taken
as a linear interpolation between those of Ge and Si. Figure 5 con-
firms this linear behavior of the lattice constant and has a practical
use for an express analysis of film composition based on the peak
position in the XRD spectra.
It has been found that the volume fraction of ͑111͒-oriented
grains strongly depends on film thickness, for fixed deposition con-
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
G367
Table II. Conditions during the cycled deposition of Ge0.59Si0.41
films at 430°C, 9 sccm GeH4, and 21 sccm SiH4.
Volume
percent
Step
Film
Step
duration Pressure
Ge
thickness of ͑110͒
Sample number
͑min͒
͑Pa͒
͑atom %͒
͑nm͒
grains
A
1
2
3
1
1
1
5
90
5
¯
23
24
¯
¯
1
5
1
90
59
88
40
B
1
2
3
1
1
1
90
5
90
5
Figure 5. Correlation between the ͑220͒ peak position in the XRD spectra of
poly-GexSi1−x films and germanium content in those films.
4
1
¯
23
24
¯
¯
90
5
ditions. Growth of the ͑111͒ grains is completely stopped beyond a
film thickness of about 100 nm ͑Fig. 6͒. This result can be explained
in terms of a higher vertical growth rate of ͑110͒-oriented grains
compared to ͑111͒-oriented grains, causing the so-called evolution-
ary selection.52,53 Eventually, the ͑111͒ grains are overgrown by the
͑110͒ grains. This leads to a blockage of the ͑111͒ grains by the
͑110͒ grains. The difference in growth rate can be explained, e.g., by
dissimilar adsorption–desorption for various crystal planes.54,55 All
crystals during their evolution will finally be bound by the thermo-
dynamically most stable plane, which is the ͑111͒ plane. The growth
rate in the ͑111͒ direction is the lowest compared to the growth rates
in the other possible directions. As all the existing crystal planes
form different angles with the ͑111͒ plane, these angles together
with the relative growth rates in those directions determine the ori-
entation of the surviving crystals.
For equivalently thick films, the total pressure influences the vol-
ume fraction of ͑111͒-oriented crystallites: a decrease of total pres-
sure causes an increase of ͑111͒ grain fraction. This probably arises
from nucleation matters. Two series of experiments have been car-
ried out to investigate the pressure-dependent texture formation.
Namely, the deposition processes were performed in a cycled re-
gime, where during a deposition experiment the pressure was alter-
nated from a high to low value or vice versa ͑see Table II for de-
tails͒. Within a series, the equivalent cycles consisted of low- and
high-pressure steps. The order of the steps within the cycles was
inverted for the two series. The XRD spectra of the films nucleated
at the low-pressure conditions exhibited an increase, sometimes dra-
matic, in the volume fraction of ͑111͒-oriented grains ͑see sample A
in Table II and in Fig. 7͒. In contrast to this, the nucleation step,
which is the first step during deposition where the growth starts on
the SiO2 surface, occurring at the higher pressure led to a dominance
of ͑110͒ grains ͑see sample B͒. One should bear in mind that a thin
1
1
59
90
80
layer of amorphous silicon was always predeposited onto the oxide
surface before starting the deposition of poly-GexSi1−x
.
To explain the pressure-dependent nucleation, we have devel-
oped a nucleation model33 which was intended for a qualitative de-
scription only. This very simple model is based on a purely statisti-
cal approach and uses the diamond lattice features, without taking
into account the quantum-mechanical nature of electrons that form
interatomic bonds. Therefore, no total energy calculation ͑i.e., tight-
binding approach͒ is carried out. Instead, the model only treats the
collision frequency of the molecules with the surface leading to the
adsorption, which is pressure-dependent. The surface migration is
not taken into account because of the low deposition temperature.
Because poly-GexSi1−x has a diamond lattice, various distances be-
tween adjacent atoms characterize the atomic planes facing from
͑111͒- and ͑110͒-oriented grains. Using elementary geometry rules
one can simply derive the following: for a two-dimensional ͑110͒
nucleus the shortest distance between the adjacent species in the
plane is smaller with respect to that for a two-dimensional ͑111͒
nucleus. Although during the growth the particles are randomly ad-
sorbed, the average distance between the adjacent particles adsorbed
on the surface ͑e.g., inter atomic distance͒ decreases with increasing
pressure. This is simply because the density of the adsorbed species
increases. In principle, this distance does not necessarily determine
the final grain orientation. However, we speculate that the probabil-
ity of formation of the closer-packed clusters ͓i.e., ͑110͒ pre-grains
Figure 6. Volume fraction of ͑110͒-oriented grains vs layer thickness for
GexSi1−x films deposited at 430°C. GeH4 flow rate 9 sccm, SiH4 flow rate
21 sccm, and total pressure 5 Pa.
Figure 7. XRD spectra of the films obtained after cycled deposition per-
formed at 430°C. GeH4 flow rate 9 sccm, and SiH4 flow rate 21 sccm ͑see
Table II for the exact conditions͒.
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

G368
Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
Figure 8. Percentage of certain oriented nuclei vs number of adsorption acts
per time unit.
Figure 9. Measured distributions in the downstream direction for germa-
nium content at 430°C, GeH4 flow rate 37 sccm and SiH4 flow rate 86 sccm:
͑1͒ Ptot = 20 Pa, no gas dilution; ͑4͒ two-step deposition performed during
2 min at Ptot = 53 Pa without gas dilution and 30 min at Ptot = 40 Pa with a
350-sccm dilution by N2. The symbols correspond to the same symbols as in
Fig. 10.
in our example͔ increases with decreasing the interatomic distance.
In such a simple model, no bond-formation issues are considered.
We have realized a computer code which analyzes distances be-
tween the surface species after a number of randomized adsorption
acts. The relative amount of certain nucleated grains is defined as a
fraction of the species adsorbed within a proper diameter. Figure 8
shows the result of such a simulation. One can notice that the domi-
nance of the far-adsorbed species ͓i.e., ͑111͒-oriented grains in our
approximation͔ disappears at higher pressures. Contrary to this, low-
ering the pressure increases the appearance probability for the far-
adsorbed species ruling the formation of ͑111͒ nuclei.
were kept at 6 and 14 Pa for GeH4 and SiH4, respectively. An ex-
ception was made for curve ͑4͒, where the partial pressures were
two times lower compared to the other curves. The symbols in Fig.
10 correspond to the same data series represented in Fig. 9. The
poly-GexSi1−x ͑x = 0.57–0.61͒ films were formed at a deposition
temperature of 430°C, GeH4 gas flow of 37 sccm, and SiH4 gas
flow of 86 sccm.
The film morphology has been studied in detail in our earlier
paper.32 The root-mean-square surface roughness Rrms strongly de-
pended on both total pressure and film thickness. The tendency of
the Rrms behavior exhibits apparent growth of the roughness with
increasing thickness. The latter is attributed to an increase of the
deviation in both the grain lateral sizes and grain heights, which is
confirmed by atomic force microscopy ͑AFM͒ measurements. The
behavior of Rrms vs total pressure exhibits a similar trend, i.e., some
grains become wider and higher with increasing total pressure. Ac-
cording to the scanning electron microscopy ͑SEM͒/AFM measure-
ments carried out in Ref. 32, the surface was composed of clusters
having rather pronounced curvature, sometimes circular in shape if
observed form the top. For circular-shaped clusters, a minimum of
roughness can be achieved for surfaces packed with the smallest
clusters having identical sizes. Therefore, flat amorphous films will
have a minimum roughness. If the clusters are big enough and,
furthermore, have different sizes, the surface becomes rougher. Our
best 45–55-nm-thick poly-GexSi1−x ͑x Ϸ 0.6͒ samples deposited at
a pressure of 20–100 Pa have a roughness of about 1 nm. It appears
that the ͑111͒ grains have a smaller size compared to the ͑110͒
grains. The difference is about a factor of 2–3. As mentioned, an
increase of the deviation in the grain diameters leads to the forma-
tion of a rougher layer. Because at total pressures below 10 Pa we
observe the transition from a ͑110͒-to ͑111͒-dominant layer, this
leads to a roughening ͑Rrms = 2 nm͒ of the surface.32
In Fig. 10, series ͑1͒ was deposited at a low total pressure of
20 Pa, without dilution with nitrogen. The second series was ob-
tained at an increased pressure of 150 Pa; additionally, a nitrogen
flow of 800 sccm was introduced into the chamber. As-chosen con-
ditions enabled keeping the precursor partial pressures constant at
6 Pa for GeH4 and 14 Pa for SiH4. By dilution in N2 the self-
dilution effect due to the conversion of one molecule of SiH4 or
GeH4 into two molecules of H2 was reduced. This should result in
flatter profiles. One can only notice a small difference between the
slopes corresponding to series ͑1͒ and series ͑2͒. Series ͑2͒ even
shows a little stronger decay of the growth rate in contrast to the
expectation. The difference, however, is within the measurement
error.
It can be calculated based on Fig. 9 and 10 and based on the total
area of wafers and tube wall that the total conversion of GeH4 and
GeH4/SiH4 Precursor System: A Relative GeH4 Depletion
in the Downstream Direction
As the high sticking probability ratio results in a relative GeH4
depletion from the gas mixture, this further causes changing
germane-to-silane partial pressure ratio along the precursor-drift di-
rection in batch-type reactors. The change of the partial pressure
ratio leads to a decay of germanium content in the layers, as shown
in Fig. 9. It can be seen that, between the 33.5-cm slot and the
61.5-cm slot, the germanium content decreases within 3–4 atom %.
Figure 10 shows the deposition rate distributions along the wafer
boat. The same as in Fig. 9, these experiments represent films de-
posited with and without nitrogen dilution. In spite of the differences
in the total pressure for curves ͑1͒, ͑2͒, and ͑3͒, the partial pressures
Figure 10. Measured distributions of deposition rate of poly-GexSi1−x ͑x
= 0.57–0.61͒ films in the downstream direction at 430°C, GeH4 flow rate
37 sccm and SiH4 flow rate 86 sccm: ͑1͒ Ptot = 20 Pa, no gas dilution; ͑2͒
Ptot = 150 Pa, dilution with 800 sccm of N2; ͑3͒ two-step deposition per-
formed during 1 min at Ptot = 100 Pa without gas dilution and 29 min at
Ptot = 150 Pa with a 800-sccm dilution by N2; and ͑4͒ two-step deposition
performed during 2 min at Ptot = 53 Pa without gas dilution and 30 min at
Ptot = 40 Pa with a 350-sccm dilution by N2. ͑ࡗ, b͒ Same as in Fig. 9.
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
G369
Figure 11. Measured distributions of deposition rate of poly-GexSi1−x ͑x
= 0.42–1.0͒ films in the downstream direction at 430°C and Ptot 20 Pa. GeH4
flow rate: ͑1͒ 13, ͑2͒ 19, ͑3͒ 31, ͑4͒ 37, and ͑5͒ 37 sccm; SiH4 flow rate: ͑1͒
88, ͑2͒ 88, ͑3͒ 88, ͑4͒ 86, and ͑5͒ 0 sccm; average germanium content in the
films: ͑1͒ 42, ͑2͒ 51, ͑3͒ 58, ͑4͒ 60, and ͑5͒ 100 atom %.
Figure 12. Distribution of deposition rate along the wafer boat. Conditions:
temperature 430°C, total pressure 20 Pa, Si2H6 gas-phase fraction 14%.
to free surface sites between GeH4 and SiH4 is in the range from 5
to 6 ͑Fig. 2͒. Because Si2H6 has a higher sticking probability com-
pared to SiH4 and Si2H6 produces two Si atoms per decomposed
molecule of disilane, one can expect an improvement of the unifor-
mity for germanium content distribution in the downstream direction
if silane is substituted by disilane. There is also a possibility that
disilane has a higher sticking probability than germane. In the latter
case the germanium content distribution is reversed. From
literature34,49,56 it is known that the ratio of sticking probabilities
between Si2H6 and SiH4 is in the range 10–40 and between GeH4
SiH4 is only 5 and 1.5%, respectively. The decay of the growth rate
therefore has to be attributed to the small decay of Ge content in the
layers along the gas-flow direction, because Ge strongly influences
the growth rate ͑see Fig. 3͒. The decay of the deposition rate in the
flow direction is much more noticeable for the films grown at higher
GeH4/SiH4 ratio, resulting in higher germanium content, as illus-
trated in Fig. 11. The formation of such germanium-rich films en-
hances downstream depletion and therefore leads to more significant
drop of the deposition rate. Based on our empirical data, we devel-
oped a gas consumption model intended for the practical needs to
predict the downstream distributions for both the deposition rate and
germanium content at diverse gas flows and total pressures.31,33 The
model exhibits the feasibility for simulations for more than 60 wa-
fers placed downstream of the reactor. The model confirms that in
particular, GeH4 consumption determines the decay of deposition
rate.
Based on our results we therefore conclude that total pressure
does not play an important role in the range studied, whereas the
partial pressures, in particular of GeH4, are important from the
depletion point of view. Because of the large amount of wafers
placed in the wafer boat and the limited annulus between the wafers
and the tube wall, one might expect a total pressure drop in the
downstream direction due to the viscosity of gases. However, the
pressure drop measured by a pair of calibrated pressure gauges
͑baratrons͒ was within the measurement error, i.e., less than 1 Pa.
Apart from film composition, the depletion also influences other
film properties such as crystallinity or morphology. Because the
deposition rate gradually decreases in the downstream direction, this
leads to the formation of thinner layers. Therefore, one can expect a
change of the film crystallinity in that direction because of the
thickness-dependent behavior. We measured a decrease of the frac-
tion of ͑110͒-oriented grains, roughly from 90 to 60 vol %. One
should bear in mind that the partial pressures were kept constant.
Several experiments on dilution with hydrogen were carried out.
As hydrogen is a product of the precursor decomposition, it prob-
ably influences the deposition process either through the self-
dilution effect ͑i.e., one molecule of SiH4 or GeH4 forms two mol-
ecules of H2͒ or due to the interaction of H2 with free surface sites
and suppressing the rate of hydrogen desorption. To investigate this,
H2 was introduced into the reaction chamber. The precursor partial
pressures were kept constant. No hydrogen influence on deposition
rate was noticed.
and SiH4 is about
5
at 625°C. So
␥
4/␥GeH /␥Si H
SiH
4 2 6
= 1:5:10 ... 40. When this also holds for 430°C one already may
expect the reverse behavior.
Figure 12 represents the deposition-rate distribution along the
wafer boat. The values of deposition rate are comparable to those
obtained for the SiH4–GeH4 system. This confirms that the deposi-
tion process is limited by surface hydrogen desorption at the tem-
peratures studied. From Fig. 13, it appears that the sticking prob-
ability for Si2H6 is higher with respect to the sticking probability for
GeH4. A 14% fraction of disilane in the gas phase supplies about 50
atom % of silicon in the films ͑see the data corresponding to the
highest flows͒. Thus, the Si/Ge ratio in the gas phase is almost
doubled in the layers. The ratio of sticking probabilities can be cal-
culated as ␥GeH /␥Si H = 0.40. Therefore, for 430°C this results in
4
␥
/␥GeH /␥Si H =216:5 ... 6:15 for x Ϸ 0.5 and shows similar
SiH
4
6
trend as in4liter2ature.
The germanium content distributions, shown in Fig. 13, exhibit
opposite trends than those obtained for the SiH4–GeH4 system;
namely, the germanium content increases with distance from the
furnace entrance. In this light, a relative depletion from Si2H6 takes
place in the downstream direction. An interesting observation is that
A Comparison between GeH4/SiH4 and GeH4/Si2H6
Precursor Systems
As mentioned previously for polycrystalline GexSi1−x ͑x Ϸ 0.6͒
films deposited at 430°C the ratio of reactive sticking probabilities
Figure 13. Distribution of germanium content along the wafer boat. For
deposition conditions see Fig. 12.
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

G370
Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
deposition-rate distribution can be converted from a descending or-
der into ascending order. To explain this, let us draw attention to Fig.
3, where we observed a rapid increase of the deposition rate vs
germanium content in the range between 50 and 65 atom %. As the
points marked with triangles in both Fig. 12 and 13 fall in this
specific range, the depletion effect suppressing the deposition rate is
overcompensated by the growth enhancement at high germanium
contents. This results in greater growth rates in the gas-flow direc-
tion.
Regarding film texture, no significant dissimilarities were ob-
served between the two precursor systems studied. For example,
deposition from a Si2H6–GeH4 mixture at a total pressure of 20 Pa,
temperature of 430°C, Si2H6 flow of 4 sccm, and GeH4 flow of
56 sccm resulted in a 250-nm-thick poly-Ge0.67Si0.33 film with a
volume ratio of 4 between ͑110͒- and ͑111͒-oriented grains. This
resembled very much the films formed in the GeH4–SiH4 system.
In comparison with our research, the activation energy evaluated
in Ref. 57 for a Si2H6–GeH4 system decreased from 1.5 eV for
poly-Si to 1 eV for poly-Ge0.31Si0.69. The decrease of the activation
energy was also explained in terms of hydrogen desorption. One can
notice that for similar-composed films with x Ϸ 0.3, the activation
energy obtained in Ref. 57 was lower than that in our work, which
was probably due to using disilane. Disilane is easier to dissociate
with the formation of SiHx radicals, which are chemically active and
can better react with adsorbed hydrogen at such low temperatures.
For example, we measured the deposition rate of amorphous silicon
for a pure Si2H6 source at 430°C, which was about 0.8 nm/min.
However, for a pure SiH4 source the deposition was impossible to
detect after half an hour. Another reason for obtaining the lower
activation energy was that the process in Ref. 57 has probably suf-
fered from gas-supply-limiting conditions, which would have caused
visible lowering of the slope of the Arrhenius plot, especially at
higher temperatures. The latter conclusion was made because of the
low gas-flow rates ͑0–5 sccm GeH4 and 5 sccm Si2H6͒, high depo-
sition rates of about 10 nm/min, and a 40-cm-long hot part of the
furnace before the first measuring wafer. Additionally, a GeH4/Si2H6
ratio of 1 resulted in deposition of Ge0.3Si0.7 layers, which was not
the case for our study and probably indicated a strong depletion
from Si2H6. According to our investigation, amorphous Ge0.5Si0.5
films can be deposited at a gas-flow ratio of about 6.
Figure 14. XRD spectra of polycrystalline Ge0.6Si0.4 films with a low frac-
tion of amorphous phase after deposition at 430°C and after a 30-min post-
annealing at 800°C in a dry N2 ambient.
crystalline-amorphous phases and grain boundaries. As we already
discussed, germanium content is a few percent lower in the poly-
crystalline material with respect to the amorphous material. It there-
fore appears that under the conditions studied, annealing selectively
increases the germanium fraction in the ͑110͒-grain bulk due to the
diffusion of germanium probably from the amorphous phase and/or
grain boundaries. This is consistent with the statement in Ref. 30 on
page 78 that the ͑110͒ texture is more favorable for fast diffusion of
dopants.
Conclusions
In this work, an intensive study including deposition and charac-
terization of polycrystalline GexSi1−x films with x up to 0.65 was
carried out for low deposition temperatures varying between 430
and 480°C. The experiments were performed using pure GeH4/SiH4
and GeH4/Si2H6 gas sources. It has been measured, for the first time
to the best of our knowledge, that the reactive sticking probability
ratio between GeH4 and SiH4 varies between 11 and 3.5 depending
on both the deposition temperature and film composition. The depo-
sition rate was a sensitive measure of the film composition: we
observed a rapid increase of the deposition rate with the germanium
content beyond 50 atom % due to the enhancement of hydrogen
desorption. This makes precise control of the deposition rate and
therefore film thickness difficult for high fractions of germanium.
A relative GeH4 depletion from the source mixture led to a
gradual change of the partial pressure ratio between the precursors
along the precursor-drift direction. This consequently influenced
both the composition and crystallinity of the layers, and the deposi-
tion rate. Higher values of the deposition rate caused more notice-
able depletion effects. No influence of extra hydrogen on deposition
kinetics was observed. To investigate the feasibility of minimizing
the depletion effects, pure GeH4 and Si2H6 gases were investigated
for the role of the potential sources. No improvement was obtained
but depletion from Si2H6 occurred due to a higher sticking probabil-
ity for Si2H6 compared to GeH4. A proper mixture of SiH4, Si2H6,
and GeH4, where both effects compensate each other, can be sug-
gested for a better uniformity over the wafer boat.
Film Annealing
High-temperature annealing steps are included in standard
CMOS processing. Because the deposition temperature employed in
this work is much lower than usual annealing temperatures, signifi-
cant impact on film texture can be expected after annealing. To
investigate this, a 30-min postanneal at 800°C in a dry-N2 ambient
was applied to some samples. Germanium content and texture were
compared before and after the annealing step. It has been found that
for the layers with a low fraction of amorphous phase, annealing
only slightly influenced film crystallinity. Briefly, the ͑110͒ grains
normally decreased in size ͑a 12% decrease was observed for ap-
proximately 60-nm grains͒. The ͑111͒ grains however mostly in-
creased in size within 10% of the initial value of about 30 nm. The
volume fraction of ͑110͒ grains with respect to ͑111͒ grains de-
creased from 70–80 to 60–70%. However, annealing of the material
with a high fraction of amorphous phase resulted in the crystalliza-
tion with a clear dominance of ͑111͒ grains over ͑110͒ grains. In the
latter case, the grain size increased from 15–40 to 40–80 nm for
͑110͒ grains and from 5–10 nm to 50–100 nm for ͑111͒ grains.
The germanium content, measured by taking the ͑220͒ peak po-
sition in the XRD spectra into account, always slightly increased
after the annealing step ͑see Fig. 14 and Fig. 5͒. However, the po-
sition of the ͑111͒ peak in the same spectra did not show a measur-
able shift. Also, the EDX measurements did not exhibit any increase
of the germanium content. It is known that XRD provides the infor-
mation relevant to a crystalline material, without analyzing the
amorphous phase and grain boundaries, whereas EDX measures the
The polycrystalline layers mainly contain ͑111͒- and ͑110͒-
oriented grains with its ratio depending on both deposition pressure
and film thickness. The ͑111͒ grains are completely overgrown with
the ͑110͒ grains beyond a film thickness of about 100 nm due to the
evolutionary selection. Raising the total pressure above 100–200 Pa
͑no dilution with an inert gas͒ causes an enhanced formation of the
amorphous phase. Reducing the total pressure below 10 Pa led to a
dominance of ͑111͒-oriented grains in the layers due to nucleation
matters. The nucleation model was successfully introduced in order
to explain the pressure-dependent nucleus formation.
A 30-min postannealing at 800°C in a dry-N2 ambient did not
crucially influence the texture of polycrystalline Ge0.6Si0.4 films with
a low amount of the amorphous phase. For the layers with a high
average composition without making
a separation between
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 153 ͑5͒ G363-G371 ͑2006͒
G371
23. S. Sedky, P. Fiorini, M. Caymax, S. Loteti, K. Baert, L. Hermans, and R. Mertens,
J. Microelectromech. Syst., 7, 365 ͑1998͒.
amount of the amorphous phase, annealing leads to crystallization,
resulting in a dominance of ͑111͒-oriented grains having an in-
creased diameter.
24. A. E. Franke, Y. Jiao, M. T. Wu, T.-J. King, and R. T. Howe, in Technical Digest of
the 2000 Solid-State Sensor and Actuator Workshop, p. 18 ͑2000͒.
25. Y.-C. Jeon, T.-J. King, and R. T. Howe, J. Electrochem. Soc., 150, H1 ͑2003͒.
26. K. Yano, T. Ishii, T. Hashimoto, T. Kobayashi, F. Murai, and K. Seki, IEEE Trans.
Electron Devices, 41, 1628 ͑1994͒.
27. Y. Takahashi, M. Nagase, H. Namatsu, K. Kurihara, K. Iwadate, Y. Nakajima, S.
Horiguchi, K. Murase, and M. Tabe, in Tech. Dig. - Int. Electron Devices Meet.,
1994, 938.
Acknowledgments
This work was partially financially supported by Philips Re-
search Laboratories in Eindhoven, The Netherlands. We would like
to thank A. M. Otter of the MESA+ Clean Room of the University of
Twente for the EDX and SEM measurements, E. H. J. Ruiter, A. A.
I. Aarnink, G. Boom, A. Kooy, and S. Kruger for general technical
support. We also thank C. J. van der Poel and Y. Ponomarev of
Philips Research for their helpful discussions.
28. D.-W. Kim, S. Hwang, T. F. Edgar, and S. Banerjee, J. Electrochem. Soc., 150,
G240 ͑2003͒.
29. Y.-C. King, T.-J. King, and C. Hu, in Tech. Dig. - Int. Electron Devices Meet.,
1998, 115.
30. T. I. Kamins, Polvcrystalline Silicon for Integrated Circuits and Displays, Kluwer
Academic Publishers, Norwell, MA ͑1998͒.
31. A. Y. Kovalgin and J. Holleman, in Thin Film Transistor Technologies VI, Y. Kuo,
Editor, PV 2003-23, p. 216, The Electrochemical Society Proceedings Series, Pen-
nington, NJ ͑2003͒.
32. A. Kovalgin and J. Holleman, J. Phys. IV, 11, 47 ͑2001͒.
33. A. Y. Kovalgin, J. Holleman, C. Salm, and P. H. Woerlee, in Thin Film Transistor
Technologies V, Y. Kuo, Editor, PV 2000-31, p. 269, The Electrochemical Society
Proceedings Series, Pennington, NJ ͑2001͒.
34. J. Holleman, A. E. T. Kuiper, and J. F. Verweij, J. Electrochem. Soc., 140, 1717
͑1993͒.
35. S. M. Gates, C. M. Greenlief, S. K. Kulkarni, and H. H. Sawin, J. Vac. Sci. Technol.
A, 8, 2965 ͑1990͒.
36. T. J. Donahue and R. Reif, J. Electrochem. Soc., 133, 1691 ͑1986͒.
37. R. J. Buss, P. Ho, W. G. Breiland, and M. E. Coltrin, J. Appl. Phys., 63, 2808
͑1988͒.
38. J. H. Comfort and R. Reif, J. Electrochem. Soc., 136, 2386 ͑1989͒.
39. M. Liehr, C. M. Greenlief, S. R. Kasi, and M. Offenberg, Appl. Phys. Lett., 56, 629
͑1990͒.
40. M. Liehr, C. M. Greenlief, M. Offenberg, and S. R. Kasi, J. Vac. Sci. Technol. A, 8,
2960 ͑1990͒.
The University of Twente assisted in meeting the publication costs of this
article.
References
1. C. L. Huang, N. D. Arora, A. I. Nasr, and D. A. Bell, IEEE Electron Device Lett.,
29, 1208 ͑1993͒.
2. K. F. Schuegraf, C. C. King, and C. Hu, in VLSI TSA, Proceedings of Technical
Papers, p. 86 ͑1993͒.
3. B. Ricco, R. Versari, and D. Esseni, IEEE Trans. Electron Devices, 17, 103 ͑1996͒.
4. W. Lo, H. Lin, W.-J. Hsia, C. Yates, V. Hornback, J. Elmer, W. Catabay, M.
Mirabedini, V. Gopinath, E. Li, D. Pachura, J. Lin, L. Duong, S. Prasad, M. Mat-
sunaga, and T. Tsuda, J. Electrochem. Soc., 152, G110 ͑2005͒.
5. H. Lin, W. Lo, S. Gu, V. Hornback, J. Elmer, and W. Catabay, J. Electrochem. Soc.,
152, G126 ͑2005͒.
6. T. J. King, J. P. McVittie, K. C. Saraswat, and J. R. Pfiester, IEEE Trans. Electron
Devices, 41, 228 ͑1994͒.
7. T. J. King, J. R. Pfiester, J. D. Schott, J. P. McVittie, and K. C. Saraswat, Tech. Dig.
- Int. Electron Devices Meet., 1990, 153.
8. Y. V. Ponomarev, C. Salm, J. Schmitz, P. H. Woerlee, and D. J. Gravesteijn, in
41. A. E. T. Kuiper and E. G. C. Lathouwers, J. Electrochem. Soc., 139, 2594 ͑1992͒.
42. D. J. Robbins, J. L. Glasper, A. G. Cullis, and W. Y. Leong, J. Appl. Phys., 69,
3729 ͑1991͒.
VLSI TSA, Proceedings of Technical Papers, p. 311 ͑1997͒.
9. C. Salm, D. T. van Veen, D. J. Gravesteijn, J. Holleman, and P. H. Woerlee, J.
Electrochem. Soc., 144, 3665 ͑1997͒.
43. B. S. Meyerson, K. J. Uram, and F. K. LeGoues, Appl. Phys. Lett., 53, 2555
͑1988͒.
44. J. E. Crowell, G. Lu, and B. M. H. Ning, in Mater. Res. Soc. Symp. Proc., 204, 253
͑1991͒.
10. S. K. Kang, J. J. Kim, D. H. Ko, H. B. Kang, C. W. Yang, T. H. Ahn, I. S. Yeo, T.
W. Lee, and Y. H. Lee, J. Electrochem. Soc., 150, G167 ͑2003͒.
11. B. Tillack, P. Zaumseil, G. Morgenstern, D. Krüger, B. Dietrich, and G. Ritter, J.
Cryst. Growth, 157, 181 ͑1995͒.
12. C. Isheden, H. H. Radamson, E. Suvar, P.-E. Hellström, and M. Östling, J. Elec-
45. G. Schülze and M. Henzler, Surf. Sci., 124, 336 ͑1983͒.
46. K. Sinniah, M. G. Sherman, L. B. Lewis, W. H. Weinberg, J. T. Yates, Jr., and K.
C. Janda, Phys. Rev. Lett., 62, 567 ͑1989͒.
trochem. Soc., 151, C365 ͑2004͒.
13. P. M. Garone, J. C. Sturm, and P. V. Schwartz, Appl. Phys. Lett., 56, 1275 ͑1990͒.
14. B. S. Meyerson, IBM J. Res. Dev., 34, 806 ͑1990͒.
47. L. Surnev and M. Tikhov, Surf. Sci., 138, 40 ͑1984͒.
48. J. H. Klootwijk, Ph.D. Thesis, University of Twente, Enschede ͑1997͒.
49. J. Holleman and J. F. Verweij, J. Electrochem. Soc., 140, 2089 ͑1993͒.
50. C. Salm, Ph.D. Thesis, University of Twente, Enschede ͑1997͒.
51. H. Stör und W. Klemm, Z. Anorg. Allg. Chem., 241, 305 ͑1939͒.
52. A. van der Drift, Philips Res. Rep., 22, 267 ͑1967͒.
53. A. J. Dammers and S. Radelaar, Textures Microstruct., 14–18, 757 ͑1991͒.
54. J. G. E. Gardeniers, M. M. W. Mooren, M. H. J. M. de Croon, and L. J. Giling, J.
Cryst. Growth, 102, 233 ͑1990͒.
15. M. Racaneli and D. W. Greve, Appl. Phys. Lett., 56, 2524 ͑1990͒.
16. S.-M. Jang and R. Reif, Appl. Phys. Lett., 59, 3162 ͑1991͒.
17. S.-M. Jang and R. Reif, Appl. Phys. Lett., 60, 707 ͑1992͒.
18. N. M. Russel and W. G. Breiland, J. Appl. Phys., 73, 3525 ͑1993͒.
19. Y. V. Ponomarev, C. Salm, J. Schmitz, P. H. Woerlee, P. A. Stolk, and D. J.
Gravesteijn, Tech. Dig. - Int. Electron Devices Meet., 1997, 536.
20. C. H. Huang, A. Chin, and W. J. Chen, J. Electrochem. Soc., 149, G209 ͑2002͒.
21. A. E. Franke, D. Bilic, D. T. Chang, P. T. Jones, T.-J. King, R. T. Howe, and G. C.
Johnson, in Technical Digest of the 10th International Conference on Solid-State
Sensors and Actuators, Transducers 99, IEEE, p. 530 ͑1999͒.
22. S. Sedky, P. Fiorini, M. Caymax, A. Verbist, and C. Baert, in Technical Digest of
the 1997 International Conference on Solid-State Sensors and Actuators, Transduc-
ers, IEEE, p. 237 ͑1997͒.
55. J. G. E. Gardeniers, L. J. Giling, F. de Jong, and J. P. van der Eerden, J. Cryst.
Growth, 104, 727 ͑1990͒.
56. C. R. Kleijn, J. Electrochem. Soc., 138, 2190 ͑1991͒.
57. J.-W. Kim, M.-K. Ryu, K.-B. Kim, and S.-J. Kim, J. Electrochem. Soc., 143, 363
͑1996͒.
Downloaded on 2015-05-03 to IP 141.210.2.78 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).