GeHx (x=0-3) and GenHx (n=2-7) in flash pyrolysis of GeH4

Article abstract of DOI:10.1016/S0009-2614(01)01385-9

Gas-phase thermal decomposition of germane, GeH₄, has been studied using a flash pyrolysis technique (～ 20 μs reaction time), coupled with a vacuum-ultraviolet (VUV) photoionization time-of-flight mass spectrometer (TOFMS). Pyrolysis temperatures ranged from 685 to 1500 K, and resulted in the formation of GeH₂ above 900 K and GeH_x (x = 0, 1, 3) above 1100 K. In addition, Ge_n (n=2-7) and a small amount of highly unsaturated hydrogenated Ge_nH_x (n=2-7) clusters were observed above 1100 K.

Full text of DOI:10.1016/S0009-2614(01)01385-9

10 January 2002
Chemical Physics Letters 351 (2002) 171–177
www.elsevier.com/locate/cplett
GeHx (x ¼ 0–3) and GenHx (n ¼ 2–7) in flash pyrolysis
of GeH4
*
Steven D. Chambreau, Jingsong Zhang
Department of Chemistry, University of California, Riverside, CA 92521, USA
Received 16 October 2001; in final form 30 October 2001
Abstract
Gas-phase thermal decomposition of germane, GeH4, has been studied using a flash pyrolysis technique (ꢀ 20 ls
reaction time), coupled with a vacuum-ultraviolet (VUV) photoionization time-of-flight mass spectrometer (TOFMS).
Pyrolysis temperatures ranged from 685 to 1500 K, and resulted in the formation of GeH2 above 900 K and GeHx
ðx ¼ 0; 1; 3Þ above 1100 K. In addition, Gen (n ¼ 2–7) and a small amount of highly unsaturated hydrogenated GenHx
(n ¼ 2–7) clusters were observed above 1100 K. Ó 2002 Elsevier Science B.V. All rights reserved.
1. Introduction
and atmospheric CVD have also been studied
[11,12]. UHV deposition essentially eliminates the
contribution of gas-phase reactions and can be
considered exclusively a surface process, while in
higher-pressure deposition processes (10À5 Torr to
1 atm), gas-phase reactions must be taken into
consideration.
The morphology, Si and Ge content, particle
size and smoothness [13], and free carrier mobili-
ties [6,14] of Si1ÀxGex (amorphous and polycrys-
talline) films grown using SiH4 or Si2H6 and GeH4
precursors have been well characterized at low Ge
levels (x < 0:35). The growth mechanisms in-
volved, however, are not well understood. In low
temperature film deposition (T ¼ 400–600 °C),
growth rates increased with increasing GeH4 par-
tial pressure, which was attributed to the lower
activation energy of hydrogen desorption from Ge
surface sites versus Si sites [15]. In this temperature
regime, the increase in the Ge:Si ratio did not in-
crease linearly with the increase in GeH4 : SiH4
GeH4 is an important precursor in chemical
vapor deposition (CVD) of Si1ÀxGex thin films.
Due to their improved thin-film transistor (TFT)
characteristics over Si films and the lower thermal
budget for their deposition, the polycrystalline
Si1ÀxGex alloys (poly-SiGe) are a promising re-
placement for poly-Si in heterojunction bipolar
transistors, MOSFETs, and pixel drives for active
matrix liquid crystal displays (AMLCD) [1].
While many CVD processes of Si1ÀxGex film
growth are carried out under high to ultra-high
vacuum (UHV) conditions [2–6] which are costly
and limit production, reduced pressure [1,7–10]
* Corresponding author. Also at: Air Pollution Research
Center, University of California, Riverside, CA 92521, USA.
Fax: +1-909-787-4713.
E-mail address: jingsong.zhang@ucr.edu (J. Zhang).
0009-2614/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved.
PII: S0009-2614(01)01385-9

172
S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
Table 1
Heats of reaction, DH0, and activation energies, Ea, for thermal decomposition reactions of germane
This worka
Dasb
Ruscicc
Newmand
Botintseve
GeH4 ! GeH2 þ H2
GeH2 ! Ge þ H2
DH0
Ea
DH0
Ea
DH0
DH0
DH0
DH0
DH0
36.0
P 35:1
50–54.3
17:5 ꢁ 10
52:1 ꢁ 3:1
P 15:5
6 35:1
87:0 ꢁ 5:0
28.2
6 28:9ð26:4Þ
GeH4 ! GeH3 þ H
GeH3 ! GeH þ H2
GeH3 ! GeH2 þ H
GeH2 ! GeH þ H
GeH ! Ge þ H
83.1
21.6
56.6
68.7
63.2
85.2
< 85:5ð82 ꢁ 2Þ
6 31:4
> 56:4ð59Þ
< 68:9ð66Þ
> 53:7ð63Þ
56.9
69.2
63.2
45 ꢁ 10
44:5 ꢁ 6:5
All values are in kcal/mol.
a DH0 (0 K) values are calculated at G3 level of theory with geometries and zero-point energies (ZPEs) at QCISD/cc-pvTZ level of
theory. See [25] for more information of G3 theory and calculations.
b Ref. [27]. No ZPE correction.
c Ref. [22]. More probable values are in parentheses.
d Ref. [20].
e Ref. [21].
ratio. Instead, the Ge:Si ratio leveled off and
GeH4 ! GeH2 þ H2
GeH2 ! Ge þ H2
The activation energy of reaction (4) has been
determined to be 50–54 kcal/mol, while H atom
loss from GeH4 was considered unlikely due to the
large H3Ge–H bond dissociation energy of
82 ꢁ 2 kcal=mol [20–22]. Energetics of the relevant
reactions are listed in Table 1 and shown in Fig. 1.
In this Letter, direct experimental evidence of
the formation of GeH2 at T > 900 K and GeHx
ð4Þ
ð5Þ
amorphous Si1ÀxGex films were limited to
x <ꢀ 0:35. The growth rate was not limited by
adsorption of growth species, and was instead at-
tributed to other factors such as surface nucleation
or gas-phase kinetics [15,16].
Understanding homogeneous thermal decom-
position of the precursors may help explain the
limitations of amorphous Si1ÀxGex growth and
allow for increased Ge incorporation into amor-
phous Si1ÀxGex films. The flash pyrolysis technique
is able to isolate and detect the initial reactive in-
termediates and primary pyrolysis products
formed upon gas-phase thermal decomposition
[17–19], thus revealing the mechanisms involved in
pyrolysis of the precursors.
The homogeneous thermal decompositions of
SiH4 and Si2H6 have been studied extensively. Our
recent experimental work on thermal decomposi-
tion pathways for silane and disilane have con-
firmed the following initiation steps [19]:
SiH4 ! SiH2 þ H2
Si2H6 ! SiH4 þ SiH2
Si2H6 ! Si2H4 þ H2
The thermal decomposition of GeH4 has been
studied experimentally [20,21], and similar initial
dissociation mechanisms have been proposed:
ð1Þ
ð2Þ
ð3Þ
Fig. 1. Energy diagram for the homogeneous thermal decom-
position of GeH4. Energies are those calculated in this work.
Indicated activation energies, Ea, are from [21]. See Table 1 for
more details.

S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
173
3. Results and discussion
ðx ¼ 0; 1; 3Þ species at T > 1100 K is presented.
These results are in good qualitative agreement
with the literature. Also reported here is the first
observation of production of gas-phase Gen (n ¼ 2–
7) and highly unsaturated hydrogenated GenHx
(n ¼ 2–7) clusters in GeH4 pyrolysis above T ¼ 1100
K. In general, the thermal decomposition and small
cluster growth involved in pyrolysis of GeH4 be-
have similarly to those of silane and disilane.
Ionization potentials (IPs) of Ge and GeHn
(n ¼ 2–4) have been determined experimentally
[22,24]. The adiabatic IP of GeH4 is 6 10:53 eV,
slightly above the 10.49 eV available from the
VUV source, thus the cold GeH4 in the molecular
beam should not be detectable by 118.2 nm VUV
photoionization. The experimental IPs for Ge,
GeH2, and GeH3 are 7.90, 9.25, and 7.94 eV, re-
spectively. The IP of GeH was calculated by us-
ing CCSD(T,Full)/avg-cc-pvTZ//QCISD//cc-pvTZ+
ZPE//QCISD/cc-pvTZ method [25], and was de-
termined to be 7.71 eV, close to previous theoret-
ical values [26,27]. Thus, all GeHx (x ¼ 0–3) species
can be ionized by a single photon at 118.2 nm. The
appearance energies of Geþ (with 2H2), GeHþ
(with H2 þ H), GeHþ (with H2), and GeHþ (with
2. Experimental
The pyrolysis experiment and the experimential
setup have been described previously [17–19]. The
pyrolysis nozzle was based on that of Chen and co-
workers [17]. Flash pyrolysis was carried out by
flowing a gas mixture of 1% germane (99.997+%,
Aldrich Chemicals) in argon at a total backing
pressure of 1.3 atm through a short, heated SiC
tube (10 mm heated length, 2 mm o.d., 1 mm i.d.,
Carborundum). The residence time of the gas
species in the heated zone of the SiC tube was on
the order of ꢀ 20 ls [17]. The precursor and
thermal decomposition products exited the heated
tube and supersonically expanded into the vacuum
chamber (held at < 5 ꢂ 10À4 Torr with the mo-
lecular beam on), and then passed through a
skimmer into the photoionization region of a Wi-
ley–McLaren type linear time-of-flight mass spec-
trometer (TOFMS) [18,19,23]. The pyrolysis
products were ionized by single VUV photon
ionization at k ¼ 118:2 nm (10.49 eV). TOF mass
spectra were recorded on a multi-channel scaler
(EG&G, Turbo-MCS), and were accumulated
with 2000 laser shots. The 118.2 nm radiation was
produced by frequency tripling the 355 nm output
from an Nd:YAG laser in a Xe cell. The VUV
radiation was focused by an MgF2 lens through a
small aperture into the photoionization zone,
while the fundamental 355 nm UV beam diverged
in this region, thereby minimizing multi-photon
ionization (MPI) and the amount of UV radiation
scattered in the ionization zone. When the tripling
medium was removed, no ion signal was observed
at all pyrolysis temperatures, indicating negligible
MPI or secondary electron-impact ionization (EI)
from the fundamental UV radiation.
2
3
H) from ionization fragmentation of GeH4 are
10.7, 11.3, 10.8, and 10.8 eV, respectively [22,28].
Thus, photoionization fragmentation of GeH4
should not occur at 10.49 eV (118.2 nm). Note that
the appearance energy of GeHþ (with H2) from
ionization fragmentation of GeH3 radical is
6 9:0 eV [22].
The results of GeH4 pyrolysis and germanium
cluster production are shown in Figs. 2–4. The
small peaks between m=e 70 and m=e 76 at low
pyrolysis temperatures are attributed to a minor
Fig. 2. Mass spectra of pyrolysis of germane (1% in Ar) be-
tween 685 and 1500 K, detected using 10.49 eV photoioniza-
tion/TOFMS. Baselines of the mass spectra are offset for clarity.

174
S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
above T ꢀ 1100 K, as indicated by the higher mass
peaks between 140 and 550 (Fig. 4).
The small mass peaks at m=e 70–76 present
below 800 K, before significant thermal decom-
position takes place in the ꢀ 20 ls time scale, are
due to a minor amount of EI and/or photoion-
ization fragmentation of the GeH4 precursors. The
EI is caused by a small number of photoelectrons
resulting from scattered light in the photoioniza-
tion region. Every effort has been made to mini-
mize the contribution from EI, such as by
minimizing the 118.2 nm spot size and by masking
the photoionization region from 355 nm radiation,
but it cannot be eliminated completely without
losing the photoionization signal as well. A mass
spectrum for the EI fragmentation pattern of
GeH4 was modeled by using the monoisotopic EI
mass spectrum of GeH4 [28,29] and the relative Ge
isotopic abundances. The calculated spectrum,
however, is not similar to our observed mass
spectra below 800 K. This might be due to different
temperatures of GeH4 and electron-energy-de-
pendent EI fragmentation [30,31] (with a range of
electron energy from 0 to 300 eV and a most
probable energy of 150 eV due to the location of
the molecular beam in this experiment). Alterna-
tively, photoionization fragmentation of GeH4 by
both the VUV and the fundamental UV photons
(at a total photon energy ꢀ 14 eV) could occur,
although the mass pattern modeled based on
photoionization efficiency curves of GeH4 [22] is
different from the observed as well.
Fig. 3. Enlarged view of mass spectra for pyrolysis of germane
(685–1500 K) for the monomer. The mass spectra are offset for
clarity.
The contribution from EI and/or photoioniza-
tion fragmentation, which is small and constant up
to 800 K, is subtracted out as described below. If
the extent of EI is assumed to be nearly constant at
all pyrolysis temperatures (i.e., the products and
precursors were efficiently cooled upon supersonic
expansion), the mass spectra at low pyrolysis
temperatures (T 6 800 K, before significant ther-
mal dissociation) can be used as a baseline and
subtracted from the higher temperature spectra to
eliminate the EI contribution. If supersonic cool-
ing is less efficient, ion fragmentation would be
expected to increase with temperature as the in-
ternal energy of the products increases. If this were
the case, the resulting peaks, after background
subtraction, would represent the upper limit to
Fig. 4. Enlarged view of mass spectra for GenHx (n ¼ 2–7)
formation upon flash pyrolysis of germane (1% in Ar) between
1160 and 1500 K. Baselines of the TOF mass spectra are offset
for clarity.
amount of EI of GeH4 precursors (discussed be-
low), and these peaks remain nearly constant up to
800 K. Above 900 K, the peaks between m=e 70
and 79 begin to grow, and steadily increase up to
T ¼ 1500 K, indicating thermal decomposition of
GeH4 (Fig. 3). Contributions to these mass peaks
from each GeHx (x ¼ 0–3) species are deconvo-
luted by the method discussed below. In addition,
Gen and GenHx (n ¼ 2–7) clusters are produced

S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
175
photoionization products. However, it is generally
believed that the supersonic cooling is efficient and
the temperature of the molecular beam is 6 100 K
[17]. This is largely consistent with the nearly
constant peak intensities up to T ¼ 800 K.
average values are listed in Table 2. As absolute
photoionization cross-sections of the various
GeHx (x ¼ 0–3) species at 118.2 nm are not known,
no correction is made for the photoionization ef-
ficiencies. Nevertheless, the deconvolution results
starting from Ge and GeH and from GeH2 and
GeH3 are similar, implying that this correction is
not very significant.
Experimentally, evaluating the pyrolysis prod-
ucts of germane is more complicated than silane
due to the existence of five stable isotopes of ger-
manium with similar isotopic abundances: 70Ge
(20.5%), 72Ge (27.4%), 73Ge (7.8%), 74Ge (36.5%),
and 76Ge (7.8%), while silicon is primarily present
as 28Si ðꢀ 92%Þ. Mass peaks in the TOF mass
spectra are first fitted with Gaussian peaks in order
to determine the experimental intensities of each
mass peak, and the baseline contributions from EI
fragmentation are removed. The contributions of
the GeHx (x ¼ 0–3) products are then deconvo-
luted in a process using the relative abundances of
Ge isotopes. As the m=e 70 peak is exclusively due
to 70Ge, the contributions of 72Ge, 73Ge, 74Ge, and
76Ge to m=e 72, 73, 74, and 76 peaks can be cal-
culated from the m=e 70 peak using isotope
abundances. Similarly, the contributions of GeH
species can be calculated using m=e 71, which is
exclusively 70GeH, and isotope abundances. The
Ge and GeH components are then subtracted from
the experimental peaks, leaving the contributions
from GeH2 and GeH3. The remaining intensity at
m=e 72 is exclusively due to 70GeH2, which can be
used to determine all isotope contributions of
GeH2. Another iteration of this process effectively
separates the contributions of all GeHx (x ¼ 0–3)
species to the experimental spectra. A similar de-
convolution process is performed by starting with
m=e peaks 78 and 79, which result from 76GeH2
(with no or little contribution from parent 74GeH4)
and 76GeH3 exclusively. The results from both
calculations are in reasonable agreement, and the
At 900 K, the initial pyrolysis products are
GeH2 and H2 (reaction (4)). At ꢀ 1100 K, Ge is
formed as per reaction (5), and it becomes the
predominant product at 1500 K. The large in-
crease of Ge implies efficient subsequent decom-
position of GeH2 to Ge and H2 [although an
intersystem crossing is needed in going from
1
3
~
GeH2ðX A1Þ to Geð PÞ]. These two steps are the
lowest energy pathway to Ge product (Table 1 and
Fig. 1). Also above ꢀ 1100 K, the higher energy
channel is identified
GeH4 ! GeH3 þ H
A small amount of GeH product is also observed
ð6Þ
above ꢀ 1100 K, presumably due to:
GeH2 ! GeH þ H
ð7Þ
ð8Þ
GeH3 ! GeH þ H2
Some GeHþ ions could be produced from
photoionization fragmentation of GeH3, as the
appearance energy of GeHþðGeH3Þ is 6 9:0 eV
and the relative photoion yield of GeHþ=GeH3þ
from GeH3 at 118.2 nm is ꢀ 0:2 [22]. The GeH3
and GeH intensities reported in Table 2 have not
been corrected for this relatively minor ion frag-
mentation process, and the GeH3 and GeH values
at a given temperature can be considered as min-
imum and maximum values, respectively. Other
ion fragmentation processes, such as GeHþ2 from
Table 2
Calculated relative intensities of GeH4 pyrolysis products, averaged from both procedures of deconvolution (see text for more details)
Temperature (K)
GeH3
GeH2
GeH
Ge
900
1020
1160
1320
1500
0
5
2
6
0
1
0
3
14
4
20
32
52
6
15
58
16
29
46
146
All values have been corrected for baseline as described in the text. Relative errors are typically within ꢁ25% of the listed values.

176
S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
Table 3
Total intensity of both Ge2 and Ge2Hx clusters and estimated Ge2Hx : Ge2 ratio versus temperature
Temperature (K)
Relative intensity of Ge2 þ Ge2Hx
Ge2Hx : Ge2 ratio
1160
1320
1500
55
148
343
0.17
0.10
0.18
Intensity of m=e 140 peak and Ge isotope abundances are used to obtain the contribution exclusively from the Ge2 cluster.
GeH3, were not observed by Ruscic et al. [22]. In
summary, the predominance of the Ge product
implies that reactions (4) and (5), rather than the
higher energetic pathways (reactions (6)–(8)), are
the main initial steps in pyrolysis of GeH4. This
conclusion is consistent with the literature [20,21].
At temperatures above ꢀ 1100 K, the forma-
tion of GenHx (n ¼ 2–7) clusters was detected (Figs.
2 and 4). As before, the presence of five germa-
nium isotopes greatly complicates determining the
extent of hydrogenation of these clusters. How-
ever, similar to the deconvolution process of GeHx
(x ¼ 0–3), by using the intensity of m=e 140 peak
(due to only 70Ge2) and Ge isotope abundances,
the contribution exclusively from the Ge2 cluster
and thus the Ge2Hx : Ge2 ratio can be calculated,
and the results are presented in Table 3. These
results indicate that most of the Ge2Hx clusters are
pure Ge2 with only a small amount of highly un-
saturated hydrogenated Ge2Hx species. The low
intensity of m=e 141 peak (due to 70Ge2H exclu-
sively) also suggests that Ge2H cluster is minimal.
Determining the GenHx : Gen ratio for GenHx
ðn P 3Þ is more difficult, and due to lack of reso-
lution of these larger clusters, the extent of hy-
drogenation in these species is not estimated.
Nevertheless, the mass peak envelopes of GenHx
ðn P 3Þ indicate predominant contributions from
bare Gen ðn P 3Þ clusters (Fig. 4). This behavior in
GeH4 pyrolysis is similar to the pyrolysis of both
SiH4 and Si2H6 [32]. The silicon clusters formed
upon pyrolysis of the SiH4 and Si2H6 precursors
were found to be largely bare Sin clusters and a
small amount of highly unsaturated SinHx clusters.
For example, Si6Hx clusters were predominantly
bare (x ¼ 0), but had a small fraction which were
mono- or di-hydrogenated species ðx ¼ 1; 2Þ. With
smaller Ge–H bond strengths compared to those
for Si–H, complete dehydrogenation of GeH4 to
Ge seems to be more efficient, also a lower hy-
drogen content is expected in the GenHx clusters.
The GenHx clusters could be produced by efficient
dehydrogenation of polygermanes that might be
formed by insertion of GeH2 to GeH4, similar to
the pyrolysis of SiH4 [19,32]. In addition, the Gen
clusters could be formed directly from condensa-
tion of Ge atoms generated in the pyrolysis of
GeH4. Indeed, a strong correlation between the
concentrations of reactive species GeH2 and Ge
and the particle production has been observed in
RF discharges of GeH4 [33].
4. Conclusion
The formation of GeHx (x ¼ 0–3) species upon
flash pyrolysis of germane was directly observed
by VUV–TOFMS. The initial primary product
was GeH2, and at higher temperatures Ge, GeH,
and GeH3 were also formed, with atomic Ge being
the predominant product. These observations are
in good agreement with previous studies on ger-
mane pyrolysis, and provide direct experimental
evidence to support the thermal decomposition
mechanisms, shown in reactions (4)–(8) and Fig. 1.
Subsequent reaction of pyrolysis products re-
sulted in the formation of GenHx (n ¼ 2–7) clusters
at temperatures above 1100 K. Although the exact
extent of hydrogenation of these clusters is difficult
to define, they are largely pure Gen clusters, with
only a small amount of highly unsaturated GenHx
species.
Acknowledgements
We thank Liming Wang for the calculations of
GeH ionization energy and reaction energetics.

S.D. Chambreau, J. Zhang / Chemical Physics Letters 351 (2002) 171–177
177
[14] K. Ismail, J.O. Chu, B.S. Meyerson, Appl. Phys. Lett. 64
(1994) 3124.
Financial support from National Science Foun-
dation (CHE-0111635), UC Regents’ Faculty
Fellowships and Faculty Development Award, the
Alfred P. Sloan Foundation and the Camille and
Henry Dreyfus Foundation are gratefully ac-
knowledged.
[15] P.M. Garone, J.C. Sturm, P.V. Schwartz, S.A. Schwarz,
B.J. Wilkens, Appl. Phys. Lett. 56 (1990) 1275.
[16] J. Olivares, J. Sangrador, A. Rodriguez, T. Rodriguez, J.
De Phys. Iv 9 (1999) 321.
[17] D.W. Kohn, H. Clauberg, P. Chen, Rev. Sci. Instr. 63
(1992) 4003.
[18] S.D. Chambreau, J.S. Zhang, J.C. Traeger, T.H. Morton,
Int. J. Mass Spectrom. 199 (2000) 17.
References
[19] S.D. Chambreau, J. Zhang, Chem. Phys. Lett. 343 (2001)
482.
[20] C.G. Newman, J. Dzarnoski, M.A. Ring, H.E. O’Neal, Int.
J. Chem. Kin. XII (1980) 661.
[1] J.W. Kim, M.K. Ryu, K.B. Kim, S.J. Kim, J. Electrochem.
Soc. 143 (1996) 363.
[21] V.N. Botintsev, I.S. Zaslonko, V.S. Mikhyev, V.N. Smir-
nov, Kinetica I Kataliz 26 (1985) 1297.
[2] V.A. Tolomasov, L.K. Orlov, S.P. Svetlov, R.A. Rubtsova,
A.D. Gudkova, A.V. Kornaukhov, A.V. Potapov, Y.N.
Drozdov, Crystallogr. Rep. 43 (1998) 493.
[3] Y. Koide, S. Zaima, K. Itoh, N. Ohshima, Y. Yasuda, J.
Appl. Phys. 68 (1990) 2164.
[22] B. Ruscic, M. Schwarz, J. Berkowitz, J. Chem. Phys. 92
(1990) 1865.
[23] D.M. Lubman, R.M. Jordan, Rev. Scientific Instr. 56
(1985) 373.
[4] M.E. Jones, S.E. Roadman, A.M. Lam, G. Eres, J.R.
Engstrom, J. Chem. Phys. 105 (1996) 7140.
[5] F. Hirose, H. Sakamoto, J. Vac. Sci. Technol. A 16 (1998)
2974.
[24] C.M. Brown, S.G. Tilford, M.L. Ginter, J. Opt. Soc. Am.
67 (1977) 584.
[25] L.A. Curtiss, P.C. Redfern, V. Rassolov, G. Kedziora, J.A.
Pople, J. Chem. Phys. 114 (2001) 9287.
[6] Y.H. Xie, D. Monroe, E.A. Fitzgerald, P.J. Silverman,
F.A. Thiel, G.P. Watson, Appl. Phys. Lett. 63 (1993)
2263.
[26] R.C. Binning, L.A. Curtiss, J. Chem. Phys. 92 (1990) 1860.
[27] K.K. Das, K.J. Balasubramanian, J. Chem. Phys. 93 (1990)
5883.
[7] M. Cao, A. Wang, K.C. Saraswat, J. Electrochem. Soc. 142
(1995) 1566.
[28] F.E. Saalfeld, H.J. Svec, Inorg. Chem. 2 (1963) 46.
[29] I.L. Agafonov, G.G. Devyatykh, I.A. Frolov, N.V. Larin,
Zh. Fiz. Khim. 36 (1962) 1367.
[8] S. Ito, T. Nakamura, S. Nishikawa, Appl. Phys. Lett. 69
(1996) 1098.
[30] M. Probst, H. Deutsch, K. Becker, T.D. Mark, Int. J. Mass
Spectrom. 206 (2001) 13.
[9] S.M. Jang, R. Reif, Appl. Phys. Lett. 59 (1991) 3162.
[10] R. Schutz, J. Murota, T. Maeda, R. Kircher, K. Yokoo, S.
Ono, Appl. Phys. Lett. 61 (1992) 2674.
[31] M.A. Ali, Y.K. Kim, W. Hwang, N.M. Weinberger, M.E.
Rudd, J. Chem. Phys. 106 (1997) 9602.
[11] H. Kuhne, T. Morgenstern, P. Zaumseil, D. Kruger, E.
Bugiel, G. Ritter, Thin Solid Films 222 (1992) 34.
[12] D. Angermeier, W.S. Kuhn, R. Druihle, D. Ballutaud, R.
Triboulet, J. Electrochem. Soc. 144 (1997) 694.
[13] T.J. King, K.C. Saraswat, J. Electrochem. Soc. 141 (1994)
2235.
[32] S.D. Chambreau, J. Zhang, J. Phys. Chem. (2001)
(submitted).
[33] H. Kawasaki, J. Kida, K. Sakamoto, T. Fukuzawa, M.
Shiratani, Y. Watanabe, Jpn. J. Appl. Phys. Pt. 2 37 (1998)
L475.