Molecular beam study of possible CVD intermediates from Group-14 organometallic precursors

Article abstract of DOI:10.1016/S0009-2614(00)00039-7

Neutral products were examined via photoionization mass spectrometry following high-voltage discharge on a variety of volatile organometallic compounds containing Group-14 elements in helium. HV discharge on organosilicon precursors in the presence of nitrogen produced Si₂N as the nearly exclusive product; M₂N analogs were observed for M=Ge, Sn precursors but did not show the same remarkable propensity for formation. With an organogermane precursor, GeOH, GeNO₂, and GeNO₃ also showed significant propensities to form. Finally, toluene - the all carbon analog of phenylsilane - reacted with oxygen copresent in the discharge to give C₂O₂⁺ as the only observed cationic product following photoionization.

Full text of DOI:10.1016/S0009-2614(00)00039-7

25 February 2000
Ž
.
Chemical Physics Letters 318 2000 448–453
www.elsevier.nlrlocatercplett
Molecular beam study of possible CVD intermediates from
Group-14 organometallic precursors
)
Mark Sulkes , Lawrence C. Baldwin, Mark J. Fink
Department of Chemistry, Tulane UniÕersity, New Orleans, LA 70118, USA
Received 22 October 1999; in final form 4 January 2000
Abstract
Neutral products were examined via photoionization mass spectrometry following high-voltage discharge on a variety of
volatile organometallic compounds containing Group-14 elements in helium. HV discharge on organosilicon precursors in
the presence of nitrogen produced Si2N as the nearly exclusive product; M2N analogs were observed for MsGe, Sn
precursors but did not show the same remarkable propensity for formation. With an organogermane precursor, GeOH,
GeNO2, and GeNO3 also showed significant propensities to form. Finally, toluene – the all carbon analog of phenylsilane –
reacted with oxygen copresent in the discharge to give C2Oq2 as the only observed cationic product following photoioniza-
tion. q 2000 Elsevier Science B.V. All rights reserved.
1. Introduction
We are in the process of carrying out photolysis of
Ž
.
CD3 3SiN3 to help confirm the foregoing assign-
While studying mechanistic photochemistry of
organosilicon species using molecular beam methods
ment. Following photolysis of small silicon-contain-
ing precursors in the absence of nitrogen, another
peak consistently appeared at 68 amu, very probably
attributable to Si2C. The foregoing species presum-
ably formed via bimolecular combinations of frag-
ments following photolysis. It is apparently the case
that Si2 N and perhaps Si2C have a much greater
propensity for formation than other small silicon
species. Any such quasi-‘magic number’ species
would possibly be significant in interstellar contexts.
In addition, species such as these could possibly be
Ž
photolysis of precursors followed by photoioniza-
.
tionrmass spectra to characterize products , several
unanticipated but striking results presented them-
w
x
selves 1,2 . The 193 nm photolysis of small silicon-
containing precursors always led to large product
peaks at 70 amu in situations when nitrogen atoms
were available, either from the precursor photolysis
itself, as in the case of azidotrimethylsilane,
Ž
.
CH3 3SiN3, or as an adduct in the carrier gas. These
Ž
.
circumstances, as well as the 71 and 72 amu isotopic
useful in chemical vapor deposition CVD pro-
satellite intensities, suggested that the 70 amu peak
cesses. To study ‘favored species’ formation, it ap-
w x
Ž
.
in these instances could be identified as Si2 N 3 .
peared that high-voltage HV discharge, rather than
laser photolysis, of precursors would be a more
versatile approach. The results from preliminary
studies of silicon, germanium, and tin species are
herein presented.
)
Corresponding author. Fax: q1-504-865-5576; e-mail:
cm06acf@mailhost.tcs.tulane.edu
0009-2614r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S0009-2614 00 00039-7

(
)
M. Sulkes et al.rChemical Physics Letters 318 2000 448–453
449
w x
2. Experimental method
nm third-harmonic pulses from a Nd:YAG laser 4 .
The one-photon energy of 10.5 eV should have been
sufficient to photoionize any species present. Time-
of-flight mass measurements in a reflectron configu-
ration ensued following photoionization. If the gas
from the tripling cell was evacuated, such that only
355 nm pulses were incident on the gas beam, the
product mass peaks under observation disappeared.
The mass spectrometer could measure mass peaks up
to ;600 amu. In general, it was found that only
negligible product mass peaks appeared above the
After expansion through the orifice of a General
Valve Series 9 pulsed solenoid nozzle, the carrier gas
Ž
.
He, ;100 psi with seeded precursor molecules
entered a short flow channel where the discharge
occurred, fragmenting the precursor molecules. Sub-
sequently, collisions in the flow channel removed
excess energy from fragments to the carrier gas. A
substantial fraction of neutral species were ultimately
formed, some following additional bimolecular
chemistry. It is the neutral species that were studied
in our experiments.
Ž
parent masses of precursors studied but note the
.
results below for discharges on toluene .
The polycarbonate flow extension had a length of
8.5 mm and axial channel diameter of 1.5 mm, with
opposed tip HV discharge electrodes, nearly touch-
ing in the axial center of the channel, placed 4 mm
down from the top of the flow extension. Stainless-
steel sewing needles, ds0.032Y, were used for the
HV electrodes. The sewing needles were held within
Nylon 6-32 screws that allowed for tip spacing ad-
justments, with a typical tip spacing of ;100 mm.
Negative HV pulses ;500 ns in width and typically
3–5 kV in amplitude were applied between the pins
at 20 Hz to induce breakdown in the gas pulses.
Under ideal conditions a pulsed discharge was sus-
tainable for hours. In practice, however, fouling of
the pins eventually occurred, either impairing dis-
charge or shorting out the pin tips if carbonaceous
impurities were present. The pins were cleaned and
regapped every one to two days, resulting in slightly
different tip gaps in different experiments. Similar
experimental results were obtained where 0.058Y
tungsten wires with sharpened tips were used instead
of sewing needles. Since, however, the SS sewing
needles showed little tip deterioration after weeks of
use, it was convenient to continue using sewing
needles. HV pulses occurred ;600 ms after the
pulse applied to the gas solenoid and ;60 ms
before the 118 nm photoionization pulses further
downstream, producing HV breakdown near the front
edge of the gas pulse; HV pulse timing needed to be
refined within 10–20 ms for optimum precursor
fragmentation. With badly adjusted timing no dis-
charge breakdown would result.
Helium was used as the carrier gas in all the
experiments, typically with a backing pressure of
100 psig. All the organometallic species that were
used were liquids; helium gas flowed over small
sample reservoirs at room temperature prior to the
Ž
gas expansions. In instances where other gases N2 ,
.
O2 , CO, N2O were co-added, they were present as
5% or 10% fractions in helium.
Compared to laser photolysis of precursors, HV
discharge presented several attractive features for
these and more extended studies. There was a high
level of precursor fragmentation; any neutral prod-
ucts that preferentially formed via subsequent bi-
molecular chemistry could be identified in the prod-
uct mass spectra. Also, from a practical view, HV
discharge implementations can readily be made on a
larger size scale, in which significant amounts of
useful precursors could potentially be produced.
3. Results and discussion
3.1. Organosilicon precursors
The high propensity for the formation of Si2 N is
shown in the HV discharge results carried out on
Ž
.
phenylsilane, PhSiH3 Fig. 1 . With no discharge the
only significant peak following 118 nm photoioniza-
tion was the parent mass peak of phenylsilane at 108
amu. A discharge in pure helium produced a number
of fragments, the most prominent being the atomic
silicon peak at 28 amu. If nitrogen was added to the
helium carrier gas, the only visible product peak was
Si2 N at 70 amu. There was still some unfragmented
parent peak in the mass scan of Fig. 1, unlike the one
The gas emerging from the flow channel was
skimmed and subjected to 118 nm photoionization
pulses generated by frequency tripling ;20 mJ 355

(
)
450
M. Sulkes et al.rChemical Physics Letters 318 2000 448–453
Ž
.
Fig. 1. High-voltage discharge of phenylsilane PhSiH3 in helium with and without co-added N2 , CO, N2O, and O2.
for pure helium. The pin electrode spacing was
CO added, a species that is isoelectronic with N2. In
this instance, however, it is evident that no quasi-
Ž
.
changed slightly electrodes removed and cleaned in
each scan, thereby altering the discharge conditions.
To qualify the remarkable ‘favored product’ results
following N2 addition, the scan above it in Fig. 1
shows HV discharge with a corresponding amount of
Ž
‘magic number’ species like Si2 N formed note the
.
significant peak for atomic silicon that remains .
Similar results were obtained following HV dis-
Ž
.
charge on azidotrimethylsilane in helium Fig. 2 .
Ž
.
Fig. 2. High-voltage discharge of azidotrimethylsilane Me3SiN3 in helium.

(
)
M. Sulkes et al.rChemical Physics Letters 318 2000 448–453
451
Ž
Here, no co-added nitrogen was necessary to produce
a large peak at 70 amu, with isotopic satellites at 71
and 72 amu. Note also the presence of a discernible
peak at 68 amu, due to Si2C. Hence at least to some
extent Si2C formation is competitive with Si2 N for-
mation. No other product or fragment peaks within at
least an order of magnitude of response were dis-
cernible; all the atomic silicon has reacted to form
the favored products.
Finally, HV discharge on phenylsilane in the pres-
ence of oxygen resulted in another significant prod-
uct, as is seen in Fig. 1. While it is clear from the
mass spectra that the formula of the 45 amu peak
must be SiOH, the actual bonding arrangement is
unknown at this juncture. Note that under the experi-
mental conditions SiOH may not be a ‘magic num-
ber’ species, as evidenced by the continued presence
of Si2 dimer mass peaks. With N2O addition no
produced GeEt4 and GeEt3 cations each peak has
.
the germanium isotopic distribution signature but no
atomic germanium. Upon HV discharge in helium, a
significant germanium peak appeared. Addition of
N2 caused a Ge2 N peak to appear, in analogy to
Si2 N. However, Ge2 N did not show the strong
relative propensity for formation of Si2 N, as evi-
denced by a large atomic germanium peak that re-
mained. Both GeOH and GeNO2 appeared as well,
due to residual oxygen present. Upon O2 addition
GeOH competitively removed the atomic germanium
peaks. Initially, strong GeNO2 and GeNO3 peaks
were also evident, but the latter two disappeared
after prolonged flow of O2rHe. With co-added N2O,
GeOH, GeNO2 , and GeNO3 peaks also appeared, to
the exclusion of any discernible atomic germanium
peak. All three of these species evidently showed a
high propensity for formation as evidenced by the
absence of a significant atomic germanium peak. The
bonding arrangements of the foregoing species is
unknown at this point.
Ž
prominent 70 amu peak appeared one is evident at
.
higher detector sensitivity , presumably because the
nitrogen atoms are competitively bound by oxygen.
3.2. Organogermanium precursors
3.3. Organostannane precursors
HV discharge experiments were carried out on
Discharge experiments were done using tetram-
Ž
.
Ž
.
tetraethylgermane, GeEt4, in helium Fig. 3 . The
118 nm photoionization of GeEt4 without discharge
ethylstannane, Me4Sn, as a precursor Fig. 4 . A
discharge in pure He produced an enhanced tin peak.
Ž
.
Fig. 3. High-voltage discharge of tetraethylgermane GeEt4 in helium with and without the presence of N2O, O2 , and N2.

(
)
452
M. Sulkes et al.rChemical Physics Letters 318 2000 448–453
Fig. 4. High-voltage discharge of tetramethylstannane in helium with and without co-added O2.
With O2 co-added, there was limited formation of
SnOH, though most of the atomic tin peak remained.
tions for Si2 N, identical HV discharge experiments
were carried out using toluene as a precursor. Toluene
is a direct analog of the phenylsilane used in the
previously described silicon experiments. It turned
out that negligible amounts of C2 N were observed,
but the results are quite striking, as can be seen for
Ž
.
With N2 addition not shown there was extremely
limited formation of Sn2 N. Evidently Sn is too large
to form stable species like silicon and germanium.
Ž
.
3.4. Toluene and perdeuteriotoluene
deuterotoluene Fig. 5 .
With N2 addition the only discernible product is
at 114 amu. Using undeuterated toluene, analogous
results were obtained, except the product peak was at
To investigate the extent to which C2 N would
form in conditions analogous to the formation condi-
Ž
.
Fig. 5. High-voltage discharge of toluene-d8 C7 D8 in helium with co-added N2 and O2.

(
)
M. Sulkes et al.rChemical Physics Letters 318 2000 448–453
453
105 amu. The detected product cationic peak must
have the formula C8 Dq9 rC8 Hq9 . No nitrogen-con-
taining species formed competitively with the fore-
going species. In electron impact mass spectra of
silylidene, :Si5CH2 , following a HV discharge of
Me4Si in pure argon carrier gas, employing a ring
Ž
HV electrode configuration. They did not carry out
.
mass measurements, however. In contrast, our HV
toluene, a significant amount of tropylium cation,
discharge experiments on Me4Si in He resulted in
production of atomic silicon but no detected silyli-
dene. A future endeavour for us will be to see how
systematic changes in HV discharge conditions can
alter some of the product distributions we have
already obtained.
In addition to identifying novel species that can
subsequently be studied individually employing beam
REMPI spectroscopies, there is the practical possibil-
ity of upscaling HV discharge preparative schemes.
It would appear that a species such as Si2 N might be
of use as CVD species in silicon nitride thin film
depositions.
q
w x
C7 D7 , is observed 5 ; the tropylium cation is aro-
matic, accounting for its stability. It appears that the
C8 Dq9 rC8 Hq9 ion peaks in the HV discharge experi-
ments are most likely due to methyltropylium. What
is remarkable in view of the N2 addition experiment
is an analogous experiment with co-added O2 , also
shown in Fig. 5. Here, for both deuterated and
undeuterated toluene, a dramatic product peak ap-
peared at 56 amu, leading to the conclusion that the
detected cation must have the formula C2Oq2 . Note
that here the formation of the neutral precursor to
methyltropylium cation was completely uncompeti-
tive compared to formation of the neutral species
leading to the detected C2Oq2 peak. Significant ef-
forts to observe the neutral C2O2 species, ethylene-
dione, have been made in recent years, so far without
References
w x
1
Y. Huang, M. Sulkes, M.J. Fink, J. Organomet. Chem. 499
w
x
success 6–11 . Because of the highly non-equi-
librium conditions involved at the HV discharge,
followed in a matter of microseconds by drastic
internal cooling due to collisions followed by a free
molecular flow regime, the possibility that neutral
C2O2 was produced in this experiment may not be
nil.
Ž
.
1995 1.
w x
2
I. Borthwick, L. Baldwin, M. Sulkes, M.J. Fink,
Ž
.
Organometallics 19 2000 139.
w x
Ž .
3
w x
4
D. Brugh, M. Morse, Chem. Phys. Lett. 267 1997 370.
Ž
.
S.E. Van Bramer, M.V. Johnston, Appl. Spectrosc. 46 1992
255.
w x
5
F.W. McLafferty, Interpretation of Mass Spectra, 2nd edn.,
Benjamin, New York, 1973, p. 109.
w x
6
D. Schroder, C. Heinemann, H. Schwartz, J. Harvey, S. Dua,
Ž
.
S. Blanksby, J. Bowie, Eur. J. Chem. 4 1998 2550.
w x
7
D. Sulze, T. Weiske, H. Schwarz, Int. J. Mass Spectrom. Ion
4. Conclusions
Ž
.
Processes 125 1993 75.
w x
8
L. Pandolfo, G. Paiaro, S. Catinella, P. Traldi, J. Mass
Smaller Group-14 atoms apparently can form sta-
bility favored species in high product yields follow-
ing HV discharge. It may be the case that different
favored products or product branching ratios may be
attainable as a function of discharge conditions. This
can be seen by comparing our HV discharge results
on tetramethylsilane, Me4Si, with those of Clouthier
Ž
.
Spectrom. 31 1996 209.
w x
9
D. Dawson, H. Chen, J.L. Holmes, Eur. J. Mass Spectrom.
Ž
.
373 1996 .
w
w
w
x
10 H. Chen, J.L. Holmes, Int. J. Mass Spectrom. Ion Proc. 133
Ž
.
1994 111.
x
11 P.G. Raine, H.F. Schaefer III, R.C. Haddon, J. Am. Chem.
Ž
.
Soc. 105 1983 194.
x
12 W. Harper, K. Waddell, D. Clouthier, J. Chem. Phys. 107
w
x
Ž
.
and co-workers 12 . They obtained LIF spectra of
1997 8829.