Ether-like Si-Ge hydrides for applications in synthesis of nanostructured semiconductors and dielectrics

Article abstract of DOI:10.1039/b908280h

Hydrolysis reactions of silyl-germyl triflates are used to produce ether-like Si-Ge hydride compounds including H₃SiOSiH₃ and the previously unknown O(SiH₂GeH₃)₂. The structural, energetic and vibrati

Full text of DOI:10.1039/b908280h

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
Ether-like Si–Ge hydrides for applications in synthesis of nanostructured
semiconductors and dielectrics
Jesse B. Tice,a Change Weng,a John Tolle,a Vijay R. D’Costa,b Rachna Singh,b Jose Menendez,b
John Kouvetakisa and Andrew V. G. Chizmeshya*a
Received 27th April 2009, Accepted 1st June 2009
First published as an Advance Article on the web 15th July 2009
DOI: 10.1039/b908280h
Hydrolysis reactions of silyl-germyl triflates are used to produce ether-like Si–Ge hydride compounds
including H3SiOSiH3 and the previously unknown O(SiH2GeH3)2. The structural, energetic and
vibrational properties of the latter were investigated by experimental and quantum chemical simulation
methods. A combined Raman, infrared and theoretical analysis indicated that the compound consists
of an equal mixture of linear and gauche isomers in analogy to the butane-like H3GeSiH2SiH2GeH3
with an exceedingly small torsional barrier of ~0.2 kcal mol-1. This is also corroborated by
thermochemistry simulations which indicate that the energy difference between the isomers is less than
1 kcal mol-1. Proof-of-principle depositions of O(SiH2GeH3)2 at 500 ◦C on Si(100) yielded nearly
stoichiometric Si2Ge2O materials, closely reflecting the composition of the molecular core. A complete
characterization of the film by RBS, XTEM, Raman and IR ellipsometry revealed the presence of
Si0.30Ge0.70 quantum dots embedded within an amorphous matrix of Si–Ge–O suboxide, as required for
the fabrication of high performance nonvolatile memory devices. The use of readily available starting
materials coupled with facile purification and high yields also makes the above molecular approach an
attractive synthesis route to H3SiOSiH3 with industrial applications in the formation of Si–O–N high-k
gate materials in high-mobility SiGe based transistors.
H3SiOSiH3 + 2 NH3 → Si2N2O (s) + 6 H2
(1)
Introduction
The Si2N2O phase can be viewed as an assembly of Si–O–Si
building blocks linked together at the Si sites by trigonal nitrogen
centers. In our above synthesis approach the molecular Si–O–Si
cores of H3SiOSiH3 deliver both the compositional and bonding
configuration at the nanoscale required to form the desired solid-
state material. In contrast to the formation of conventional silicon
Silicon dioxide (SiO2) is the most prominently used gate dielectric
for microelectronics devices incorporating elemental silicon and
Si-rich Si1-xGex alloy semiconductors. However, as CMOS dimen-
sional scaling has progressed, the sizes of Si-based transistor struc-
tures have beenconsiderablyreduced requiringalternate dielectrics
to ensure device reliability and the intended performance gains
expected on account of Moore’s Law. Within the last ten years
silicon oxynitride (SiON) layers with thickness of <5 nm were
introduced to replace conventional SiO2 in its role as a gate
insulator.1–3 These materials offer substantial improvements over
the oxide counterpart including fewer interface defects and the
ability to act as barriers to conventional B, P and As dopants
diffusing from the polycrystalline Si gate into the substrate.4 The
stoichiometric Si2N2O analog has a higher dielectric constant than
either SiO2 or SiON and it is more likely to form a chemically
robust, ordered structure adjacent to the Si interface.5–7 Therefore
this material could be used to further mitigate current leakage
and suppress dopant penetration while maintaining high gate
capacitance over the reduced length scales anticipated in modern
devices. In recent work we presented a preliminary account of the
formation of Si2N2O films at low temperatures via reactions of
disiloxane H3SiOSiH3 and a large excess of NH3 according to the
reaction shown by eqn (1).8
2,3
oxynitride gate dielectrics via reactions of NH3 with SiH4 (or
chlorosilane derivatives), the above process has the advantage
to readily guarantee reproducibility of composition on a large
production scale because precise control of the NH3 activity is not
required in the reaction medium. This is particularly attractive
from a processing perspective and it is primarily due to the
fact that disiloxane represents the limiting reagent while nitrogen
derived from NH3 is incorporated only to the degree required to
achieve the compound stoichiometry. This dramatically reduces
the complexity of the process, since precise conditions of the
reaction are no longer critical and the problem is reduced to
the simplest possible interaction involving the precursor and an
ambient of ammonia.
Recently we have shown that H3SiOSiH3 can also be used
as a single source to fabricate light emitting films comprised
of Si nanoparticles embedded in an amorphous SiO2 matrix.9
The latter may have potential application in optoelectronics
including flat panel displays as well as charge storage devices
such as nonvolatile memory. In our prior work amorphous
films of SiOx (x < 2) were first deposited at 750–850 ◦C on
Si(100) substrates via the unimolecular decomposition of the
compound. This was followed by rapid thermal annealing to
crystallize the excess silicon and generate distinct Si nanoparticles
aDepartment of Chemistry and Biochemistry, Arizona State University,
Tempe, AZ 85287–1604, USA. E-mail: chizmesh@asu.edu; Tel: (480) 965–
6072
bDepartment of Physics, Arizona State University, Tempe, AZ 85287–1604,
USA
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 6773–6782 | 6773
©

View Article Online
with tunable sizes (1 and 10 nm), discrete shapes, and uniform
distributions throughout the amorphous SiO2 matrix. The resul-
tant nanostructures exhibit room-temperature photoluminescence
(PL) and the emission energy increases with decreasing crystal
size in accordance with quantum confinement concepts. Impor-
tant benefits of optically active silicon nanostructures enclosed
in amorphous SiO2 include improved reliability, chemical and
electrical stability, full compatibility with existing complementary
metal–oxide–semiconductor techniques, and the possibility of
generating strong emission across the entire range of visible
light.
Results and discussion
Synthesis of H3SiOSiH3
Despite the common occurrence and widespread applications of
organic disiloxanes a practical synthesis of the fully inorganic
H3SiOSiH3 analog remains elusive. Several synthetic routes to
H3SiOSiH3 have been previously reported, the most notable
involving reactions of SiH3I as the source of SiH3 and irradiation
of SiH4 in the presence of HNO3.14–16 These approaches have
not been demonstrated to be practical for routine synthesis
because they afford very limited yields and utilize highly reactive
and unstable starting materials making isolation and handling
difficult particularly in large-scale preparations. To circumvent
these problems, for the initial growth studies using this compound,
we developed an alternative synthetic route involving a single-step
reduction of Cl3SiOSiCl3 with LiGaH4, as described here for the
first time by eqn (2).
In the above examples the common highlight is the use
of a completely inorganic (C–H free) and volatile H3SiOSiH3
precursor with built-in Si–O–Si atomic arrangements that allow
stoichiometric control at the atomic level ensuring formation
of highly homogeneous materials with controllable quality and
composition leading to the desired properties. In view of the
success in using siloxane to directly prepare functional materials
based on Si, our present work seeks to expand this technology
concept by replacing the –SiH3 groups with –SiH2GeH3 groups in
the ether-like backbone to yield O(SiH2GeH3)2, thereby enabling
new applications in the area of Si–Ge photonics and dielectrics.
For example the highest performance nonvolatile flash-memory
devices are fabricated using Si1-xGex alloy nanocrystals, or so
called quantum dots, embedded in amorphous high-k dielectric
media including SiON and SiO2.10–13 The writing/erasing speed
of these devices was found to be faster and the retention times of
the encoded information significantly longer by several orders of
magnitude compared to more conventional device architectures.
The SiGe quantum dots are typically grown by CVD reactions
of SiH4 and GeH4 via a self assembly mechanism producing
a random distribution of individual grains at 600 ◦C. This is
followed by a thermal oxidation of the layers at 800 ◦C to
generate the target nanostructures and the corresponding oxide
matrix yielding areal densities and nominal dot sizes in the range
of 1011 cm-2 and 5–10 nm, respectively. It is widely reported
that the performance of flash memories incorporating these
materials can be engineered by tuning the composition of the
alloy and optimizing the shape and size of the corresponding
nanocrystals.
In this study we explore the possibility of producing Si1-xGex
alloy quantum dots surrounded by a dielectric matrix using
the simplest single-source precursor O(SiH2GeH3)2, akin to the
prior work based on H3SiOSiH3. In the following we discuss
the synthesis and fundamental properties of O(SiH2GeH3)2 and
its use in proof-of-concept depositions of Si2Ge2O materials
which closely reflect the composition of the parent molecule. We
show that the as-deposited films comprise of SiGe quantum dots
embedded within an amorphous matrix of Si–Ge–O suboxide,
as required for the envisioned applications, without the need
for additional processing steps. The compound is conveniently
obtained via simple hydrolysis of readily available and fairly
inexpensive silyl-germyl triflate derivatives. The success of this
approach prompted us to reconsider the synthesis of H3SiOSiH3
ultimately leading to the development of a practical process that
can be readily scaled up to the industrial scale. The compound is
easily separated from the byproducts in pure form and isolated in
nearly quantitative, multigram yields, indicating that the process
is amenable to large-scale production.
2 Cl3SiOSiCl3 + 3 LiGaH4 → 3 H3SiOSiH3
+ 3 LiCl + 3 GaCl3
(2)
This approach provided yields ranging up to 25% for small
laboratory-level preparations (see Experimental section) but we
found that scaling up the process was not economically or
logistically feasible and thus prohibitive, particularly from an
industrial perspective. For example, the Cl3SiOSiCl3 reactant is
not readily available and the LiGaH4 counterpart is expensive to
manufacture due to the high cost of the GaCl3 starting material.
Furthermore, LiGaH4 once produced is also difficult to isolate and
maintain in pure form owing to its instability at room temperature.
Surprisingly, our attempts of similar reduction reactions involving
the ubiquitous LiAlH4 analog yielded no evidence of the target
product, thus the use of LiGaH4 as the reducing agent was
absolutely critical to the synthesis of H3SiOSiH3. Another major
drawback of the above approach was the propensity of the LiGaH4
to decompose during the course of the experiment through side-
reactions generating large H2 over-pressures even at low reaction
temperatures in the vicinity of -78 ◦C. This problem necessitated
small scale preparations, and extremely delicate/stringent control
of the reaction conditions such as temperature, pressure and
solvent-reactant concentrations, yielding in turn a limited amount
of the compound in the range of 1–2 g, which is adequate for CVD
growth on small laboratory scale but not suitable for industrial
processing.
In this study we find that the general approach developed
for the synthesis of O(SiH2GeH3)2 (as described in subsequent
sections) represents a viable large-scale preparation of H3SiOSiH3
for applications in cost-effective fabrication of dielectrics. The
compound is produced routinely and reproducibly via reactions
of H2O with silyl triflate (H3SiOSO2CF3) as described by eqn (3).
The latter is produced quantitatively by reacting phenylsilane with
triflic acid (HOSO2CF3) via elimination of C6H6. The reaction
mechanism leading to H3SiOSiH3 could initially involve formation
of unstable silanol (H3SiOH) as a first step shown by eqn (4). This
intermediate would then react with another silyl triflate molecule
via condensation to form disiloxane while regenerating the parent
triflic acid (see eqn (5)) which can in turn be recycled to produce
more starting material H3SiOSO2CF3 (eqn (6)), thereby reducing
overall process cost (emphasized in bold in eqn (3) and (6)).
6774 | Dalton Trans., 2009, 6773–6782
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
The reactions of H3SiOSO2CF3 were performed using an excess
of the H2O and the gaseous H3SiOSiH3 product was distilled and
purified to obtain yields as high as 90–95%. IR and NMR were
used as the primary means of characterization indicating that
H3SiOSiH3 is devoid of impurities, including unreacted water. To
further explore the above mechanism we conducted an analogous
reaction involving methanol and silyl triflate, which yielded methyl
siloxane17 in accordance with eqn (7).
H3SiOSO2CF3 + CH3OH → H3SiOCH3 + HOSO2CF3
(7)
Using an excess of methanol the above reaction produces highly
pure methyl siloxane (H3SiOCH3) as confirmed by gas phase
IR. The high volatility of the product allows facile separation
from unreacted methanol via fractional distillation to isolate the
compound in 63% yield.
Synthesis of O(SiH2GeH3)2. The above hydrolysis approach
was initially developed to synthesize the new O(SiH2GeH3)2
(digermyldisiloxane) compound by combining H3GeSiH2SO3CF3,
(the germylsilyl analog of the silyl triflate) with an excess of
water. The digermyldisiloxane was isolated as a colorless, py-
rophoric liquid with a vapor pressure of 10 Torr at 23 ◦C. Its
molecular structure was elucidated by gas chromatography, mass
spectrometry (GC-MS), infrared spectroscopy (FTIR), Raman,
multinuclear NMR and quantum chemical simulations. The mass
spectrum revealed the parent ion as an isotopic envelope centered
at 223 amu with a fragmentation pattern consistent with the
O(SiH2GeH3)2 molecular structure. The 1H and 29Si NMR spectra
showed a triplet at 3.05 ppm corresponding to the GeH3 terminal
ligands, a quartet at 5.09 associated with the SiH2 protons, and
Fig. 1 Comparison of the simulated and experimental spectrum for
O(SiH2GeH3)2. The optimized spectrum labeled “THEORY” contains a
mixture of 55% linear and 45% gauche isomers.
1
singlet at -21.1 ppm due to the 29Si. A 2D H COSY spectrum
revealed crosspeaks between the SiH2 at 5.09 ppm and the GeH3
at 3.05 ppm indicating the direct connectivity between the Ge
and Si atoms in the –SiH2GeH3 fragment. A 1H–29Si HMQC
showed that the SiH2 atoms (5.09 ppm) are directly bound to the Si
atom (-21.1 ppm) which further confirms the proposed molecular
structure. In relation to the butane-like H3GeSiH2SiH2GeH3
analog, the SiH2 and 29Si resonances are shifted downfield to
ppm values of 3.29 and -105, respectively, due to the presence
of the highly electronegative oxygen atom. The IR spectra
obtained from gaseous samples showed the Si–H, Ge–H and Si–O
vibrational modes at 2153, 2069 and 1078 cm-1, respectively, as
expected for the proposed molecular structure. Fig. 1 compares
the experimental spectrum with a simulated pattern indicating an
excellent agreement between the principal features. The simulated
spectrum was obtained by first calculating the individual spectra
of the normal and gauche conformations shown in Fig. 2 and then
adjusting their linear combination to best fit the experimental data
as described in the following section. This fitting procedure was
first developed in our prior work involving the closely related
butane-like H3GeSiH2SiH2GeH3 where it was used to estimate
the proportions of the linear and gauche isomers in the observed
gas spectrum.18 We note in passing that the NMR spectra of
this class of hydride compounds only provides direct evidence of
Fig. 2 Structural models of the “normal” and “gauche” isomers of the
digermyl-disiloxane compound.
local bonding linkages, and not the geometric conformations of
the molecular backbones. Accordingly, application of the fitting
scheme to the IR spectrum of O(SiH2GeH3)2 indicates that the
compound exists as a nearly equal admixture (55 and 45%) of
normal and gauche isomers. Thus this result suggests that the
relative stabilities of the isomers might be comparable at room
temperature.
A similar analysis of the Raman spectra obtained from a solu-
tion of H3GeSiH2SiH2GeH3 in C6H6 yielded a virtually identical
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 6773–6782 | 6775
©

View Article Online
Table 1 Unscaled frequencies (in cm-1) and mode assignments for the N (normal) and G (gauche) isomers. The italic entries correspond to modes with
a common frequency and vibrational character. The largest frequency difference between the isomers is found in the antisymmetric Si–O stretches near
1080 cm-1
n-O(SiH2GeH3)2
g-O(SiH2GeH3)2
Primary mode assignments
N1
341
G1
G2
G2¢
G3
G4
341
349
362
470
772
Antisymmetric Si-Ge stretching
Out-of-phase GeH3 & SiH2 wagging ^ to backbone
In-phase GeH3 & SiH2 wagging ^ to backbone
In-phase GeH3 rocking ꢀ to backbone
In-phase GeH3 wagging
N2
N3
N4
362
470
770
N5
N6
803
869
G5
G6
797
872
Out-of-phase GeH3 wagging
In-phase GeH2 & SiH2 wagging
N7
N8
N9
N10
N11
N12
N13
N14
N15
891
900
918
957
G7
G8
G9
G10
G11
G12
G13
G14
G15
G16
893
900
915
960
Out-of-phase GeH2 & SiH2 wagging
In-phase SiH2 wagging & “GeH2” scissors
Out-of-phase SiH2 wagging & “GeH2” scissors
In-phase SiH2 scissors
Out-of-phase SiH2 scissors
Antisymmetric Si–O stretching
In- and out-of-phase symmetric GeH3 stretching
In- and out-of-phase asymmetric GeH3 stretching
In- and out-of-phase asymmetric SiH2 stretching
Isolated Si–H stretching
973
968
1075
2107
2122
2210
1089
2113
~ 2127
~ 2210
2221
Table 2 Assignment of the observed Raman peaks of the O(SiH2GeH3)2
compound. The frequencies in italic correspond to totally symmetric
modes
Frequency/cm-1
Assignment
344/355
465
607
Si–Ge stretch
GeH3 wag
Si-O stretch
748
775
877
937
2062
2150
SiH2 twisting relative to backbone
H wag ꢀ to backbone
GeH2 scissors and wag ^ to backbone
In-phase SiH2 wag
Ge–H stretch
Si–H stretch
mixture a gauche (44%) and normal (56%) isomers in the liquid
state. A representative depolarized spectrum after subtraction of
the solvent is compared in Fig. 3 with the theoretical admixture.
Note that the high frequency Raman shifts are dominated by
the Ge–H and Si–H stretches at 2069 and 2153 cm-1, in good
agreement with the IR frequencies. The low-energy spectral
range is very rich and shows a strong doublet at 350 cm-1 that
corresponds to Si–Ge stretches, not seen in the IR spectrum. The
350, 2069 and 2153 cm-1 peaks have depolarization ratios much
3
smaller than , revealing their totally symmetric character. A more
complete assignment of modes is given in Table 2 and discussed in
detail in the following section.
4
Fig.
3
Experimental and theoretical Raman spectrum for the
O(SiH2GeH3)2 compound. The theoretical data were corrected as dis-
cussed in the text.
Quantum chemical simulations of O(SiH2GeH3)2. In this sec-
tion we describe the theoretical analysis of the vibrational spectra
and elucidate the thermochemistry of both the normal and gauche
O(SiH2GeH3)2 isomers. All quantum chemical simulations were
conducted using density functional theory at the B3LYP level
and a standard 6-311N++G(3df,3pd) basis set as implemented
in the Gaussian03 code.19 Tight convergence conditions were
employed in all of the structural optimizations with no symmetry
constraints applied to either molecule. Our results indicate that
both normal and gauche isomers are asymmetrical tops possessing
rotational constants A, B and C (in GHz) of (2.46, 0.33, 0.29)
and (4.14, 0.28, 0.27), respectively, while their corresponding
optimized geometries exhibited C2v and C1 symmetry. In both
cases the spectrum of vibrational frequencies was positive definite,
indicating that the equilibrium structures are both dynamically
stable. The Si–Ge, Si–H and Ge–H bond lengths (2.392, 1.484 and
˚
1.538, respectively) were identical to within ~ 0.002 A, and closely
matched those that we reported in our earlier work on the butane-
like H3GeSiH2SiH2GeH3 isomers.18 This is also true of the hydride
bond angles, for which we find very typical◦tetrahedral values such
as ∠HSiO ~109.4 0.2◦, ∠HGeSi ~110.7 , ∠HSiH ~108.1o and
∠HGeH ~108.1◦. The Si–O bond lengths of the isomers were also
6776 | Dalton Trans., 2009, 6773–6782
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Fig. 4 Calculated infrared vibrational spectra of the normal (top panels) and gauche (bottom panels) O(SiH2GeH3)2 isomers. The mode assignments
and unscaled frequencies corresponding to the modes labeled Ni (normal) and Gi (gauche), are listed in Table 1.
˚
Our analysis closely follows that employed in our work on the
n- and g-isomers of the H3GeSiH2SiH2GeH3 compound. In the
latter case, we found that the most symmetric vibrations in the
normal isomer were split and/or shifted in the gauche counterpart.
This is likely due to the strong coupling of these modes through the
butane-like molecular backbone. In the ether-like O(SiH2GeH3)2
analog we find that the non-skeletal vibrations occur at essentially
the same frequencies (to within 2–5 cm-1) in both isomers, but
with quite different relative intensities in general (see Table 1).
For example, the GeH3 wagging modes (N5 and G5) are much
weaker in the lower symmetry gauche isomer than in the normal
counterpart.
The simulated low-frequency spectra (Fig. 4) were next used
to estimate the relative admixture g of the isomers (100g = %
normal and 100(1 - g ) = % gauche) in the observed spectrum of the
O(SiH2GeH3)2 compound. A best fit is obtained by simultaneously
varying the admixture parameter g , the intensity of the simulated
spectrum relative to that observed b, and the global frequency
scale factor h which produces the best match in peak positions over
the entire low-frequency range (300–1300 cm-1). The optimization
is carried out numerically by minimizing the following objective
function (see eqn (8) below):
found to be identical (and equal to 1.645 A), however, the Si–O–Si
angle in the gauche (147.1◦) was found to be about 3◦ smaller
than in its normal counterpart (150.1◦). We note that the above
structural parameters are very similar to those typically observed
˚
in condensed oxide phases such a-quartz in which bSiO ~1.61 A
and ∠SiOSi ~144–146◦.
Fig. 4 shows plots of the calculated spectra of n-O(SiH2GeH3)2
and g-O(SiH2GeH3)2 (top and bottoms panels, respectively)
showing the high- and low-frequency regions individually. All of
the principal vibrational features are labeled Ni (normal) and Gi
(gauche) and the corresponding normal mode assignments are
provided in Table 1. We note that with a few exceptions, such as
modes 2 and 14–16, the label numbers reflect distinct vibrational
modes shared by the two molecular isomers.
The high-frequency (>2000 cm-1) spectral features in both
isomers originate from the Si–H and Ge–H stretches centered
approximately at 2100 and 2200 cm-1, respectively. The sym-
metric Ge–H stretches are found near 2110 cm-1 (labeled N13
and G13, respectively). The higher intensity asymmetrical Ge–H
counterparts occur near 2124 cm-1, however, in the lower sym-
metry gauche conformer these are split over a narrow range
of frequencies centered on the values indicated by tildes. The
symmetric and asymmetric SiH2 stretches virtually coincide in
the normal isomer (feature N15 near 2210 cm-1), while the
corresponding modes in the gauche isomer are again split into
an envelope of multiplets centered on the same frequency. The
only distinct feature associated with the gauche isomer is a set of
lone Si–H stretching modes near 2221 cm-1 (feature G16).
The weak low-frequency Si–Ge skeletal modes {N1, G1} occur
at the same frequency (341 cm-1) in both isomers. In contrast,
the strong asymmetric Si–O skeletal vibrations {N12, G12} exhibit
the largest frequency difference (~15 cm-1) between the two
isomers, and therefore play a key role (see below) in distinguishing
between the two conformations. The low-frequency non-skeletal
vibrational structure, also shown in Fig. 4, is considerably more
complex and involves symmetric and antisymmetric Si–H/Ge–H
wagging, rocking and scissoring vibrations, both in- and out-of-
phase in relation to the oxygen center.
2
⎡
⎤
O(b,g ,h) =
Eexp (w )−b g I (hw ) +(1−g ) I (hw )
⎥
(
)
⎦
k
n
k
g
k
∑
⎢
⎣
k
(8)
yielding values g = 0.55, b ~1.4 and h = 0.9948, which were then
employed to create the synthetic spectrum labeled as “Theory”
in Fig. 1 above. We note that the frequency scale factor ~0.995
obtained here for the ether-like compounds is similar to the value
of ~0.990 obtained in our prior work on butane-like SiGe hydrides
using the B3LYP/6-311++g(3df,3pd) level of theory. The best fit
admixture implies nearly equal fractions of the two isomers (55%
normal and 45% gauche) suggesting that their thermodynamic
stability is comparable. A preliminary estimate of their relative
stability is afforded by the static-molecule electronic energies
at the B3LYP/6-311++g(3df,3pd) level of theory, which yielded
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 6773–6782 | 6777
©

View Article Online
identical energies to within a few milliHartree (E0(normal) =
-4814.430220Eh and E0(gauche) = -4814.430147Eh). A corre-
sponding thermochemical estimate was obtained using the well-
known CBS-QB3 level of theory, which reproduces experimental
heats of formation to within a few kcal mol-1. CODATA atomic
heats of formation were combined with the molecular enthalpy
estimates from CBS-QB3 to obtain standard heats of formation of
-62.03 and -62.20 kcal mol-1 for the normal and gauche isomers,
respectively, corroborating the existence of equal isomeric propor-
tions suggested by the independent spectral fitting procedure.
The analysis of the Raman spectrum proceeds along the same
lines as the discussion of the infrared absorption. In Table 2 we
show the assignment of the main Raman lines. The theoretical
spectrum in Fig. 3 was corrected by the same scale factors as
those used above in the IR calculations (0.979 for frequencies
above 1500 cm-1 and 0.995 for frequencies below 1500 cm-1). The
theoretical Raman intensities were normalized to the 876 cm-1
line for the top panel and to the 2062 cm-1 mode for the lower
panel of Fig. 3. The observed doublet near 350 cm-1 can be used
to independently determine the normal/gauche admixture. Each
of the isomers produces two Si–Ge Raman peaks, corresponding
to in-phase and out-of-phase stretches of the two Si–Ge bonds.
However, the predicted Raman intensity is much higher for the
in-phase vibration, so that the Raman spectrum of a single isomer
should be essentially dominated by one peak. The observation of a
doublet is naturally explained by the combined contribution from
the in-phase Si–Ge stretches in the normal and gauche isomers,
respectively, which according to the theoretical simulations are
separated by 7.5 cm-1. A fit of the experimental Raman spectra
using the theoretically predicted Raman intensities for the two
modes yields a gauche isomer concentration of 44%, in remarkable
agreement with the IR analysis, above. Further spectroscopic
evidence for the coexistence of the two isomers is obtained from
the high-energy Raman spectrum in the region of Ge–H stretches,
which experimentally is represented by a single, relatively broad
peak but is clearly split into two main peaks in the simulation of
the normal isomer. The combination of the two isomers produces
a complex broader structure, as seen in Fig. 3, which explains the
experimental observation if the peaks are further broadened by
effects such as isotopic disorder, particularly in the case of the Ge
atoms.
To further elucidate the relative stabilities of the isomers we also
calculated the potential energy surface (PES) of the O(SiH2GeH3)2
molecule as a function of the Ge–Si–Si–Ge backbone torsion
angle. This is similar in spirit to our earlier calculations for
the butane-like GeSiSiGe hydride in which the linear (normal)
conformation corresponds to a torsion angle of 180◦ while the
gauche modification occurs at ~66◦. In the present work we
modified the Z-matrix to explicitly incorporate the Ge–Si–Si–Ge
torsion angle, and then optimized all remaining structural degrees
of the freedom, including the bond lengths and band angles
involving the oxygen atom. Our results indicate that the optimized
position of the oxygen along the PES curve is always found to be
such that the repulsion between its lone-pairs and the remaining
“SiH2GeH3” framework is minimal. Fig. 5 compares the torsional
PES of O(SiH2GeH3)2 with that of H3GeSiH2SiH2GeH3. We note
that while the energy-torsion profiles are plotted on a common
horizontal axis, the actual torsion angle of O(SiH2GeH3)2 is 180◦
out-of-phase with that of H3GeSiH2SiH2GeH3.18 This is consistent
Fig. 5 Relaxed PES as a function of the molecular backbone torsion
angle for O(SiH2GeH3)2 (solid curve, A) and the butane-like hydride
H3GeSiH2SiH2GeH3 (dotted curve, B) (ref. 18). In both cases the plots
show that the gauche (g) isomer is metastable compared to its linear (n)
counterpart, but only slightly so in the ether-like compound. Note the
different vertical energy scales for compounds A and B, and that the
torsion angle for the butane-like molecule is 180◦ out-of-phase with that
of the ether-like compound (see ref. 18).
with the convention that the “normal” conformation is the one in
which all backbone atoms are co-planar. Thus, in Fig. 5 the butane-
like torsion is plotted as “t - 180◦” to allow the PES profiles of
the two molecules to be easily compared. The results show that
the global minimum corresponds to the “normal” configurations,
while in both molecules the “gauche” configurations are local
minima. Using the DFT B3LYP/6-311G++(3df,3pd) description
we obtain DEn–g ~0.4 kcal mol-1 for the H3GeSiH2SiH2GeH3.
However, a significantly smaller value of DEn–g ~0.05 kcal mol-1
is computed for O(SiH2GeH3)2 indicating that the forward and
reverse barriers are of comparable magnitude (~0.2 kcal mol-1).
On statistical grounds this implies that the populations of normal
and gauche isomers are expected to be the same, which is consistent
with the experimental comparison of the vibrational spectra with
the theoretical ~ 50/50 mixture proposed above.
Formation of SiGe nanocrystals in amorphous Si–Ge–O matrix.
Deposition studies of O(SiH2GeH3)2 was conducted on an Si(100)
substrate using a molecular beam epitaxy (GS-MBE) chamber
maintained at a base pressure of 3 ¥ 10-10 Torr. The substrates
were first outgassed at 650 ◦C under UHV to remove surface
contaminants and then flashed at 1050 to 1200 ◦C for 20 s to
desorb the native oxide from their surface. Gaseous O(SiH2GeH3)2
was introduced onto the substrate surface at 500 ◦C through
an injection manifold using a high precision needle valve. This
arrangement allowed a constant flow to be maintained yielding a
growth pressure of 5 ¥ 10-5 Torr. The duration of the depositions
was typically up to 90 min, producing layers that exhibited a visual
appearance indistinguishable from that of the underlying Si wafer.
The bulk elemental content of the samples was determined
by simulations of the Rutherford backscattering spectra (RBS).
The Si and Ge concentrations were measured at 2 MeV (see
Fig. 6) and a resonance nuclear reaction at 3.05 MeV was used
to establish the precise oxygen content. For all films grown at
500 ◦C the composition was found to be 35–37% Si, 45–43%
Ge and 20% O closely matching the Si2Ge2O composition of
the O(SiH2GeH3)2 precursor. Transmission electron microscopy
(XTEM) observations of representative films showed the presence
of fairly uniform layers exhibiting undulated surfaces, amorphous
6778 | Dalton Trans., 2009, 6773–6782
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Fig. 8 Raman spectrum of the Si–Ge–O material showing the charac-
teristic Ge–Ge, Si–Ge and Si–Si vibrational bands observed in crystalline
Si–Ge alloys.
Fig. 6 Representative 2 MeV RBS spectrum of a ~250 nm thick SixGeyOz
film on Si(100), yielding a composition of 35% Si, 45% Ge and 20%
O. Thick gray and thin black lines represent measured data and simulated
fit, respectively.
diethoxymethane-like units via stoichiometric elimination of the
well-known digermylsilane SiH2(GeH3)2 compound according to
eqn (9), and shown schematically in Fig. 9.
interfaces and average thicknesses in the range ~200 and 400 nm.
The latter values are consistent with RBS estimates of the thickness
and correspond to samples grown for 45 and 90 min at a growth
rate of 3–4 nm min-1. The high-resolution micrographs revealed the
formation of well-defined and well-separated crystalline nanopar-
ticles evenly distributed within an amorphous matrix with an
average size of 4–5 nm (see Fig. 7). Using selected area diffraction
patterns of the bulk layer and that of the Si substrate as an
internal calibration standard, the unit cell parameter of the cubic
2 O(SiH2GeH3)2 → SiH2(OSiH2GeH3)2 + SiH2(GeH3)2
(9)
The process can be envisioned as an isodesmic reaction sequence
in which the bond enthalpy lost in breaking the Si–O and Si–Ge
bonds is recovered when the diethoxymethane-like and digermyl-
silane product are formed. The steps involved in an actual reaction
process likely involve discrete scission reactions in which individual
Si–Ge (~300 kJ mol-1) and Si–O (~370 kJ mol-1) bonds, are broken
in turn. In this scenario, the hypothetical SiH2(OSiH2GeH3)2
species polymerizes on the hot surface to form an amorphous
Si–Ge–O suboxide while the SiH2(GeH3)2 dissociates to produce
Si0.33Ge0.66 alloy crystallites, with a composition very close to the
experimental observation. In previous work we have carried out
detailed deposition studies of SiH2(GeH3)2 on Si(100) over a
temperature range from 400–600 ◦C. At temperatures less than
450 ◦C we found that the compound produces monocrystalline
and perfectly uniform Si0.33Ge0.66 layers with atomically smooth
surfaces, while for T > 450 ◦C we invariably formed arrays of
self-assembled and perfectly coherent quantum dots possessing
˚
structure of these grains was determined to be 5.58 A. This value
corresponds to a Si1-xGex alloy with composition x = 0.7 on the
basis of Vegard’s Law, which assumes linear interpolation between
the lattice constants of elemental Si and Ge vs. atomic content. The
presence of crystalline dots with Si0.30Ge0.70 composition was also
corroborated by Raman scattering experiments. Fig. 8 contains
an overlay of representative spectra obtained for several films
exhibiting the typical peak lineshape of crystalline Si1-xGex alloys,
consisting of Ge–Ge, Si–Ge and Si–Si bands. An analysis based
on the known compositional dependence of the Raman bands20
yielded x ~0.64–0.73, in excellent agreement with the diffraction
measurements.
The observed composition of the SiGe nanocrystals suggests a
specific reaction path for the assembly and final composition of
the “nanoparticle + matrix” film shown in the XTEM micrograph
of Fig. 7. Here we envision an oxygen condensation mechanism
in which two or more precursor molecules react to form larger
a composition of Si0.33Ge0.66 close to that observed (Si0.30Ge0.70)
in this study. In view of these observations the outcome of
our present O(SiH2GeH3)2 depositions carried out under similar
temperature/pressure conditions is perhaps not surprising—the
common feature being the strong driving force to produce
Fig. 7 (a) XTEM micrograph showing nanocrystals of Si0.30Ge0.70 embedded in amorphous Si–Ge–O suboxide matrix. (b) The image on the right is an
enlarged view of an individual crystallite showing the (111) lattice planes. (c) SAED pattern of the nanocrystals (rings) and the Si in (110) projection.
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 6773–6782 | 6779
©

View Article Online
Fig. 10 Imaginary part of the dielectric function of the Si–Ge–O material
showing several absorption features in the spectral range corresponding to
IR-active vibrations in SiO2 and GeO2 glasses.
lineshape to the asymmetric stretch in SiO2, and there is clearly a
peak below 300 cm-1 (which unfortunately is beyond the spectral
range of our ellipsometer) which can be assigned to the rocking
O motion in a Ge–O–Ge bond. A complete study would require
detailed vibrational calculations, but it is already apparent from
this brief analysis that the IR spectrum is consistent with our
conclusion above that the oxide matrix contains both Ge and Si
atoms.
Fig. 9 Condensation reaction described by eqn (9) shown here as
an isodesmic sequence, including a possible transition state structure
suggested by preliminary quantum chemical calculations (Si, Ge, O and H
atoms are drawn as gold, blue, red and white spheres, respectively).
Si0.33Ge0.66 quantum dots. In this case the nano-structures are
intergrown within the dielectric medium comprised of polymerized
SiH2(OSiH2GeH3)2 as suggested by the reaction in eqn (9).
Conclusions
Finally, we note that the reaction depicted by eqn (9) implies a
composition of Si3Ge2O2 (or Si1.5GeO) for the amorphous matrix
and Si0.33Ge0.66 for the nanocrystals, with the two components
occurring in a 1 : 1 ratio. The lower density of nanocrystals
potentially occurring in our actual films can be accounted for
by assuming that some fraction of the O(SiH2GeH3)2 molecules
react at the surface to form a stoichiometric suboxide with formula
Si2Ge2O via release of 5 moles of H2. Thus, the combined action
of the latter direct reaction with the condensation reaction in eqn
(9) would produce 2 + M moles of suboxide with composition
In this study we have reported the synthesis of ether-like SiGe
hydrides including H3SiOSiH3 and O(SiH2GeH3)2 using hydrolysis
reactions involving silyl-germyl triflates. In the case of H3SiOSiH3
the overall process represents a new and practical synthetic
route utilizing convenient reaction conditions. The compound is
obtained for the first time in nearly quantitative yields and is
readily purified to achieve semiconductor grade material for fur-
ther applications in CVD fabrication of Si–O–N microelectronic
components. The second, previously unknown O(SiH2GeH3)2
was characterized experimentally using FTIR, Raman, mass
spectroscopy and NMR, and studied in detail using state-of-the-
art first principles quantum chemical simulation methods. Theory
predicts that the compound exists as two energetically identical
isomers, closely related to the linear and gauche conformations
observed in butane-like molecules, separated by a very small tor-
sional barrier of ~0.2 kcal mol-1. The observed vibrational spectra
are found to correspond to a mixture of 55% linear and 45% gauche
isomers, in accord with the thermodynamic expectations. The
O(SiH2GeH3)2 molecule was then employed to deposit Si2Ge2O
materials on Si(100) at 500 ◦C. The composition, structure and
nano-morphology of the films was characterized by RBS, XTEM,
Raman and IR ellipsometry, which revealed the presence of
Ge-rich Si0.30Ge0.70 quantum dots (diameter ~5 nm) embedded
within an amorphous matrix of Si–Ge–O suboxide possessing a
composition between Si1.5GeO and Si1.75Ge1.5O depending on the
precise density of nanoparticles present. The strategic synthesis of
the new O(SiH2GeH3)2 compound, and its subsequent deposition
on Si(100) corroborates the notion that the precursor molecule
confers its composition to the film via complete incorporation
of its core. Collectively, our work represents a potentially useful
for every mole of SiGe2 (Si0.33Ge0.66). For
Si⎛
⎞Ge⎛
⎞O
3+2M
2+M
2+2M
⎜
⎝
2+M
⎟
⎟
⎟
⎠
⎟
⎟
⎟
⎠
⎜
⎜
⎜
⎜
⎜
⎝
example, M = 0 corresponds to a 1 : 1 mixture of Si3Ge2O2
(Si1.5GeO) “matrix” and SiGe2 nanocrystals, while M = 2 yields
a film with a 4 : 1 ratio of Si7Ge6O4 (Si1.75Ge1.5O) “matrix” and
SiGe2 nanocrystals.
To test these ideas we have carried out far-infrared studies
of the vibrational properties of the samples using spectroscopic
ellipsometry. Fig. 10 shows the imaginary part of the infrared
dielectric function of a typical film. The IR response of Si1-xGex
crystals is very weak, so that the observed signal can be assigned
to the amorphous matrix. In amorphous SiO2 and GeO2 the
asymmetric stretch, symmetric stretch and rocking movements
of the O atom in the Si–O–Si and Ge–O–Ge bonds produce
three distinct absorption bands, which appear at 1082, 800 and
455 cm-1 in SiO2 and 857, 556 and 278 cm-1 in GeO2, respectively.21
The spectrum in Fig. 10 shows at least six different absorption
bands, suggesting different local environments for the O atoms, as
expected for an oxide matrix incorporating Ge and Si atoms. Some
of the observed peaks can be directly related to the SiO2 and GeO2
spectra. For example, the 1030 cm-1 band is close in frequency and
6780 | Dalton Trans., 2009, 6773–6782
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
framework for the development of new Si–Ge–O based dielectric
materials, beginning at the molecular level.
H3SiSO3CF3 (2.90 g, 16.0 mmol) at -35 ◦C and 10 Torr generating
an immediate increase in pressure to 90 Torr. The volatiles were
immediately passed through traps at -78 and -196 ◦C. The
reaction flask was slowly warmed to 23 ◦C with continuous
Experimental
◦
pumping through -78 and -196 C traps for 1 h. Gas-phase IR
indicated the presence of unreacted methanol in the -78 ◦C trap,
while the -196 ◦C trap contained pure CH3OSiH3 (0.62 g, 63%
yield).
General considerations
All manipulations were carried out under inert conditions using
standard high vacuum line and drybox techniques. Dry, air-free
solvents were distilled from either anhydrous CaCl2 or sodium
benzophenone ketyl prior to use. The NMR spectra were collected
on an Inova 500 MHz spectrometer. Samples were dissolved
in deuterated benzene, and all nuclei were referenced either
to the signal of TMS [Si(CH3)4] or the residual solvent peak
as indicated below. Gaseous infrared spectra were obtained in
10-cm cells fitted with KBr windows. Gas chromatography mass
spectrometry (GC-MS) data were obtained using a JEOL JMS-
GC Mate II spectrometer. Lithium tetrahydroaluminate, phenyl
trichlorosilane, trifluoromethane sulfonic acid, lithium hydride
(Aldrich), gallium trichloride and hexachlorodisiloxane (Alfa
Aesar), and electronic grade germane gas (donated by Voltaix,
Inc.) were used as received. Deionized water was thoroughly
degassed to remove all of the dissolved atmospheric oxygen prior
to use. Phenylsilane, lithium tetrahydridoogallane and germylsilyl
trifluoromethanesulfonate were prepared according to literature
procedures.22–24 All starting materials were checked by NMR
spectroscopy to verify their purities. Potassium germyl (KGeH3)
was synthesized in monoglyme by reaction of gaseous GeH4 with a
finely dispersed sodium-potassium (80% K) alloy. Note: the Teflon
bar that was used to stir the GeH4/Na-K/monoglyme solution
was encapsulated in glass owing to the high reactivity of Teflon
with Na–K alloys. Caution: all reactions should be conducted
with uttermost care. Any silane byproducts are pyrophoric and
potentially explosive.
Synthesis of O(SiH2GeH3)2. Distilled water was thoroughly
degassed (0.40 g, 22.2 mmol) and then slowly added to a 100 mL
flask containing H3GeSiH2SO3CF3 (3.21 g, 12.6 mmol) at -35 ◦C
under reduced pressure (~2 Torr). The reaction flask was slowly
warmed to 23 ◦C while stirring for 1 h. The volatiles were
fractionally distilled under dynamic vacuum through U-traps held
at -30 and -196 ◦C. The latter was empty, while the former
contained pure O(SiH2GeH3)2 as a clear, colorless liquid (1.0 g,
~70% yield). Vapor pressure: ~10 Torr (23 ◦C). IR (gas cm-1): 2153
(vs), 2069 (vs), 1148 (vw), 1078 (s), 953 (m), 886 (s), 863 (s), 791 (vs),
763 (vs), 466 (w), 356 (w). 1H NMR (500 MHz, C6D6): d 3.05 (t,
6H, GeH3), d 5.09 (q, 4H, SiH2). 29Si NMR: d -21.1. GC-MS: m/z
isotopic envelopes centered at 223 (M+), 193 (SiOGe2Hx+), 176
(SiGe2Hx+), 150 (Si2OGeHx+), 119 (SiOGeHx+), 102 (SiGeHx+),
75 (GeHx+), 31 (SiHx+), 16 (O+). The Raman measurements on
the benzene solutions were measured at room temperature in
the back-scattering configuration using a 532 nm laser line. The
incident power was ~2 mW and the light was focused on the
surface of the cuvette using a 10¥ objective. The scattered light
was analyzed with a single-state 0.5 m monochromator equipped
with 1800 lines/mm gratings and detected with a liquid-nitrogen
cooled CCD detector. The spectral resolution is approximately
4 cm-1. Similar experimental conditions were used for film
measurements.
Ellipsometric measurements. Spectroscopic ellipsometry mea-
surements were carried out at room temperature using an Infrared
Variable Angle Spectroscopic Ellipsometer manufactured by
J. A. Woollam Co. The IR-VASE system is based on a Fourier-
Transform Infrared Spectrometer and covers the 0.04 eV to 0.62 eV
range. The measurements were performed at three angles of
incidence (55, 65 and 75 ◦). The Si–Ge–O samples were modeled
as a three-layer system consisting of a Si substrate, Si–Ge–O film
and a surface layer. The dielectric function of the Si substrate, with
a resistivity of 0.01–0.02 ohm cm, was measured separately and
used in tabulated form. The surface was modeled as a SiO2 layer
and the optical constants from Woollam software database were
used in tabulated form. The IR dielectric function of the Si–Ge–O
film was described using an optical dispersion model consisting
of several Lorentzian oscillators and a parametric oscillator. The
Lorentzian line shapes describe the various vibrational modes,
whereas the parametric oscillator describes the lowest direct gap
in the material. All adjustable parameters in the model, including
the thicknesses of the surface and film layers are fitted using
a proprietary Marquardt–Levenberg algorithm provided by the
ellipsometer’s manufacturer. A point by point fit is then used to
extract the dielectric function of the material. Agreement of this
dielectric function with the optical dispersion model guarantees
Kramers–Kronig consistency between the real and imaginary
parts of the experimental dielectric function.
Synthesis of H3SiOSiH3 via reduction of Cl3SiOSiCl3.
A
solution of Cl3SiOSiCl3 (8.83 g, 31 mmol) in butyl ether (40 mL)
was added slowly under reduced pressure (10 Torr) to a mixture of
LiGaH4 (3.75 g, 56.4 mmol) in 100 mL of butyl ether with stirring
at -50 ◦C. The procedure involved the addition of several drops of
Cl3SiOSiCl3/butyl ether to the reaction mixture with subsequent
trapping of any volatiles through -78 and -196 ◦C traps under
dynamic vacuum. Once all the Cl3SiOSiCl3 had been added, the
reaction flask was allowed to slowly warm to room temperature
with continuous trapping over the course of 2 h. Gas-phase IR
revealed the presence of pure H3SiOSiH3 (4.3–8 mmol, 15–25%
yield) in the -196 ◦C trap, and butyl ether in the -78 ◦C trap.
Synthesis of H3SiOSiH3via reactions of triflates. Distilled
water was thoroughly degassed (1.4 g, 78 mmol) and then slowly
added to a 100 mL flask containing H3SiSO3CF3 (23.04 g,
128 mmol) held at -35 ◦C and 10 Torr pressure. An immediate
increase in pressure to 30 Torr was ◦observed and ◦the volatiles were
passed through traps held at -78 C and -196 C. The reaction
flask was then slowly warmed to room temperature under dynamic
pumping through -78 ◦C and -196 ◦C traps for 1 h. Gas-phase IR
◦
indicated the presence pure (SiH3)2O in the -196 C trap (4.5 g,
~90% yield), while the -78 ◦C contained the HSO3CF3 byproduct.
Synthesis of CH3OSiH3. Degassed CH3OH (0.52 g,
16.3 mmol) was slowly added to a 100 mL flask containing
This journal is
The Royal Society of Chemistry 2009
Dalton Trans., 2009, 6773–6782 | 6781
©

View Article Online
15 J. E. Drake, B. M. Glavincevski and R. T. Hemmings, Can. J. Chem.,
1980, 58, 2161.
Acknowledgements
16 W. Kriner, A. G. Mc Diarmid and E. C. Evans, J. Am. Chem. Soc.,
1958, 80, 1546.
17 (a) B. Sternbach and A. G. MacDiarmid, J. Am. Chem. Soc., 1961, 83,
3384; (b) C. Glidwell, D. W. J. Rankin, A. G. Robiette, G. M. Sheldrick,
B. Beagley and S. Craddock, J. Chem. Soc. A, 1970, 315.
The work was supported by AFOSR MURI (FA9550-06-01-
0442), DOE (DE-FG36-08GO18003) and NSF (IIP 075049)
18 A. V. G. Chizmeshya, C. J. Ritter, C. Hu, J. Tolle, R. A. Nieman, I. S. T.
Tsong and J. Kouvetakis, J. Am. Chem. Soc., 2006, 128, 6919.
19 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb,
J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C.
Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci,
M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M.
Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E.
Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo,
R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi,
C. Pomelli, J. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P.
Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D.
Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K.
Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S.
Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y.
Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. G. Johnson,
W. Chen, M. W. Wong, C. Gonzalez and J. A. Pople, GAUSSIAN 03
(Revision B.04), Gaussian, Inc., Wallingford, CT, 2004.
20 J. Mene´ndez, in Raman Scattering in Materials Science, edited by
W. H. Weber and R. Merlin, Springer, Berlin, 2000, vol. 42, p. 55.
21 F. L. Galeener and G. Lucovsky, Phys. Rev. Lett., 1976, 37,
1474.
22 J. Zech and H. Schmidbaur, Chem. Ber., 1990, 123, 2087.
23 A. E. Shirk and D. F. Shriver, Inorg. Synth., 1977, 17, 45.
24 C. J. Ritter, C. Hu, A. V. G. Chizmeshya, J. Tolle, D. Klewer, I. S. T.
Tsong and J. Kouvetakis, J. Am. Chem. Soc., 2005, 127, 9855.
References
1 K. J. Plucinski, I. V. Kasperczyk and B. Sahraouni, Semicond. Sci.
Technol., 2001, 16, 467.
2 E. P. Gusev, H.-C. Lu, E. L. Garfunkel, T. Gustafsson and M. L. Green,
IBM J. Res. Dev., 1999, 43, 265.
3 M. L. Green, E. P. Gusev, R. Degraeve and E. L. Garfunkel, J. Appl.
Phys., 2001, 90, 2057.
4 K. A. Ellis and R. A. Buhrman, Appl. Phys. Lett., 1999, 74, 967.
5 Y.-N. Xu and W. Y. Ching, Phys. Rev. B, 1995, 51, 17379.
6 Q. Tong, J. Wang, Z. Li and Y. Zhou, J. Eur. Ceram. Soc., 2007, 27,
4767.
7 H. Hillert, S. Jonsson and B. Sundman, Z. Metallkd, 1992, 83, 648.
8 L. Torrison, J. Tolle, J. Kouvetakis, S. K. Dey, D. Gu, I. S. T. Tsong and
P. A. Crozier, Mater. Sci. Eng. B, 2003, 97, 54.
9 L. Torrison, J. Tolle, David J. Smith, C. Poweleit, J. Menendez, M. M.
Mitan, T. L. Alford and J. Kouvetakis, J. Appl. Phys., 2002, 92, 7475.
10 C. H. Kao, C. S. Lai, M. C. Tsai, K. M. Fan, C. H. Lee and C. S. Huang,
Electrochem. Solid State Lett., 2008, 11(4), K44.
11 Y. Liu, S. Dey, S. Tang, D. Q. Kelly, J. Sankar and S. K. Banerjee, IEEE
Trans. Electron Devices, 2006, 53, 2598.
12 A. C. Prieto, A. Torres, J. Jimenez, A. Rodriguez, J. Sangrador and T.
Rodriguez, J. Mater Sci: Mater. Electron, 2008, 19, 155.
13 S. H. Huang, Y. A. Ma, X. J. Wang and F. Liu, Nanotechnology, 2003,
14, 25.
14 R. Varma, A. K. Ray and B. K. Sahay, Inorg. Nucl. Chem. Lett., 1969,
5, 497.
6782 | Dalton Trans., 2009, 6773–6782
This journal is
The Royal Society of Chemistry 2009
©