Molecular beam photochemistry of organopolysilanes and organopolygermanes

Article abstract of DOI:10.1021/om990880u

The molecular beam photochemistry of various polysilanes and polygermanes was investigated. In most cases, the precursor compounds were photolyzed in the nozzle region of a supersonic jet by a 193 nm laser and the photoproducts analyzed downstream by 118 nm photoionization pulses followed by time-of-flight mass spectrometry. The polysilane and polygermane compounds included in this study were PhMeSi(SiMe₃)₂, PhSi(SiMe₃)₃, (Me₂-Si)₆, (Me₂Ge)₆, and 1,3-diphenyl-1,2,2,3-tetramethyl-1,2,3-trisilacycloheptane. The 193 nm photoproducts of PhSiMe₃, Me₃SiSiMe₃, and vinyltrimethylsilane were also examined for comparison purposes. Dimethylsilylene (Me₂Si:) was directly observed as the major one-photon photoproduct from the cyclic precursors (Me₂Si)₆ and 1,3-diphenyl-1,2,2,3-tetramethyl-1,2,3-trisilacycloheptane. Likewise, dimethylgermylene (Me₂Ge:) was directly observed in the photolysis of(Me₂Ge)₆. One-photon photolysis of the noncyclic polysilanes PhMeSi(SiMe₃)₂ and PhSi(SiMe₃)₃, however, gave radical products derived from the homolytic scission of a single Si-Si bond with little or no evidence of silylene being generated directly. A mechanism explaining the difference in photochemical outcome for cyclic vs noncyclic molecules is presented. Finally, molecular Si2C is found to be a ubiquitous product resulting from the multiphoton photochemistry of a number of organosilicon precursors.

Full text of DOI:10.1021/om990880u

Organometallics 2000, 19, 139-146
139
Molecu la r Bea m P h otoch em istr y of Or ga n op olysila n es
a n d Or ga n op olyger m a n es
Ian Borthwick, Lawrence C. Baldwin, Mark Sulkes,* and Mark J . Fink*
Department of Chemistry, Tulane University, New Orleans, Louisiana 70118
Received November 2, 1999
The molecular beam photochemistry of various polysilanes and polygermanes was
investigated. In most cases, the precursor compounds were photolyzed in the nozzle region
of a supersonic jet by a 193 nm laser and the photoproducts analyzed downstream by 118
nm photoionization pulses followed by time-of-flight mass spectrometry. The polysilane and
polygermane compounds included in this study were PhMeSi(SiMe3)2, PhSi(SiMe3)3, (Me2-
Si)6, (Me2Ge)6, and 1,3-diphenyl-1,2,2,3-tetramethyl-1,2,3-trisilacycloheptane. The 193 nm
photoproducts of PhSiMe3, Me3SiSiMe3, and vinyltrimethylsilane were also examined for
comparison purposes. Dimethylsilylene (Me2Si:) was directly observed as the major one-
photon photoproduct from the cyclic precursors (Me2Si)6 and 1,3-diphenyl-1,2,2,3-tetramethyl-
1,2,3-trisilacycloheptane. Likewise, dimethylgermylene (Me2Ge:) was directly observed in
the photolysis of (Me2Ge)6. One-photon photolysis of the noncyclic polysilanes PhMeSi(SiMe3)2
and PhSi(SiMe3)3, however, gave radical products derived from the homolytic scission of a
single Si-Si bond with little or no evidence of silylene being generated directly. A mechanism
explaining the difference in photochemical outcome for cyclic vs noncyclic molecules is
presented. Finally, molecular Si2C is found to be a ubiquitous product resulting from the
multiphoton photochemistry of a number of organosilicon precursors.
In tr od u ction
phase precursors to silylenes in a molecular beam:
PhMeSi(SiMe3)2 (1), PhSi(SiMe3)3 (2), and (Me2Si)6 (3).6
Organosilylenes, R2Si:, are divalent reactive interme-
diates which have been implicated in such diverse
processes as the direct synthesis reaction,1 the CVD of
thin films from organosilanes,2 and the photochemical
degradation of polysilanes.3 Organosilylenes have been
previously generated both in solution4 and in cold inert
matrixes5 by photolysis of either cyclopolysilanes (eq 1)
or 2-aryltrisilanes (eq 2).
(RR′Si)n f (RR′Si)n-1 + RR′Si:
(1)
(2)
ArRSi(SiMe3)2 f Me3SiSiMe3 + ArRSi:
Extension of these photochemical reactions to the gas
phase would allow the generation of a variety of orga-
nosilylene species for molecular beam study. Mass-
selected molecular beam experiments could then be
carried out to assess/identify reaction products of si-
lylenes with selected coreactants added to the beam.
Alternatively, molecular beam laser spectroscopy could
be carried out on cold, collisionless silylene species (e.g.
ionization potentials, vibronically resolved excitation
spectra). In a preliminary study, we examined the gas-
phase photolysis of the following known condensed-
The photochemical generation of Me2Si: from (Me2Si)6
was inferred indirectly through the mass spectrometric
detection of the ring-contracted (Me2Si)5 fragment. In
stark contrast, the linear trisilane 1 and the branched
tetrasilane 2 only gave silyl radical products resulting
from the homolytic cleavage of a single Si-Si bond. No
silylene products were detected even at the highest laser
fluences utilized.
We have now carried out additional experiments
which are more general in two respects. First, the
previous experiments employed the same 266 nm pulses
(Nd:YAG fourth harmonic, 4.66 eV per photon) down-
stream across the molecular beam both to photolyze the
(1) Bazant, V.; J oklik, J .; Rathousky, J . Angew. Chem., Int. Ed. Engl.
1968, 7, 112.
(2) Scott, B. A.; Pecenik, R. M.; Simonyi, E. E. Appl. Phys. Lett. 1981,
39, 73 and references therein.
(3) Miller, R. D.; Michl, J . Chem. Rev. 1989, 89, 1359.
(4) Gaspar, P. P.; Holten, D.; Konieczyny; Corey, J . Y. Acc. Chem.
Res. 1987, 20, 329.
(5) Michalczyk, M. J .; Fink, M. J .; DeYoung, D. J .; Welsh, K.; West,
R.; Michl, J . Silicon, Germanium, Tin Lead Compd. 1986, 9, 75.
(6) Huang, Y.; Sulkes, M.; Fink, M. J . J . Organomet. Chem. 1995,
499, 1.
10.1021/om990880u CCC: $19.00 © 2000 American Chemical Society
Publication on Web 12/10/1999

140 Organometallics, Vol. 19, No. 2, 2000
Borthwick et al.
was stirred for an additional 21 h and the THF removed by
rotary evaporation. The resulting solid was quenched with H2O
and extracted with cyclohexane. The tannish organic layer was
dried over MgSO4 and the solvent removed. Crystallization
from EtOH yielded 4.75 g (28%) of the white solid. Mp: 62.4-
62.8 °C. GCMS: m/z 452 (M+). 1H NMR (δ (ppm), d6-acetone):
7.35 (m, 10H), 0.44 (s, 6H), 0.25 (s, 6H). 13C NMR (δ (ppm),
d6-acetone): 137.1 (Ph), 134.9 (Ph), 128.8 (Ph), 127.3 (Ph), -4.1
(SiMe), -4.9 (SiMe2). 29Si NMR (δ (ppm), CDCl3): -24.3 (SiMe),
-60.7 (SiMe2).
1,4-Bis(br om om a gn esio)bu ta n e. A solution of 1,4-dibro-
mobutane (0.011 mol, 1.3 mL) in 8 mL of THF was added
dropwise to a suspension of 0.029 mol (0.70 g) of Mg turnings
in 50 mL of THF for 30 min at room temperature. The solution
was stirred for 48 h.
1,3-Dip h en yl-1,3-bis(tr ifla to)tetr a m eth yltr isila n e. (Ph2-
MeSi)2SiMe2 (0.011 mol, 5.0 g) and 65 mL of freshly distilled
CH2Cl2 were placed in a 250 mL three-necked flask equipped
with an addition funnel and reflux condensor. This solution
was cooled to 0 °C; then triflic acid (0.022 mol, 1.95 mL) was
added dropwise over 10 min with stirring. The solution was
stirred overnight at room temperature, the CH2Cl2 was
removed under vacuum, and 70 mL of dry pentane was added.
The ditriflate was used in the next step without further
purification.
precursors as well as to photoionize species present for
mass analysis. The new experiments were two laser
experiments. The photolysis pulses were now directed
just below the gas expansion nozzle, prior to gas
skimming to form a molecular beam. The analyzing
photoionization was carried out with 118 nm pulses
which were directed further downstream across the
molecular beam, adjacent to the draw-out electrode. The
118 nm pulses have a one-photon energy of 10.5 eV,
sufficient to photoionize almost any species that might
be present without significant fragmentation, leading
to their time-of-flight mass detection. Second, a wider
range of precursors has been studied. In addition to
compounds 1-3, we have also examined the additional
ring compounds (Me2Ge)6 (4) and 1,2,2,3 tetramethyl-
1,3-diphenyl-1,2,3 trisilacycloheptane (5) as well as
1,3-Dip h en yl-1,2,2,3-tetr a m eth yl-1,2,3-tr isila cycloh ep -
ta n e. The solution of 1,4-bis(bromomagnesio)butane was can-
nulated into the addition funnel attached to the round-bottom
flask containing the ditriflate. The di-Grignard was then added
dropwise over a 1.3 h period. The resulting solution was stirred
overnight at room temperature. The solvent was then removed
under vacuum, and the product was quenched with H2O and
extracted with diethyl ether. The product was then purified
over a silica gel column with hexane as the eluent, which
yielded a 0.79 g (20%) mixture of cis (45%) and trans (55%)
isomers as a clear oil. GCMS: m/z 354 (M+), both isomers.
some mono- and disilanes, which help to qualify product
analyses in the other cases. Using beam techniques, we
have been able to study the photolysis mechanism
independent of the ionization process and have demon-
strated unambiguous detection of dimethylsilylene and
dimethylgermylene.
Exp er im en ta l Section
1
UV: λmax 239. H NMR (δ (ppm), CDCl3): 1.8 (m, 8H), 1.3 (m,
Syn th esis. All manipulations were conducted under an
inert atmosphere of argon employing standard Schlenk tech-
niques. All glassware was oven-dried at 140 °C prior to use.
Tetrahydrofuran was distilled from sodium benzophenone
ketyl under nitrogen. Pentane was distilled from CaH2 under
nitrogen. Methylene chloride was washed with concentrated
H2SO4, H2O, and aqueous K2CO3(aq), dried over CaCl2, and
finally distilled over P2O5 under argon. 1H, 13C, and 29Si NMR
data were recorded on a GE Omega 400 NMR spectrometer.
GCMS data were obtained using an HP5890 Series 2 GC/
HP5989A mass spectrometer with an ULTRA 2 cross-linked
5% PhMe siloxane column. Silica gel (40 µm, 230-400 mesh)
was purchased from Scientific Adsorbents Inc. The polysilanes
1 and 2 were synthesized according to previously reported
procedures.7 Dodecamethylcylohexasilane (3) was purchased
from Petrarch, now United Chemical Technology, Inc. Dodeca-
methylhexagermane, vinyltrimethylsilane, phenyltrimethyl-
silane, and hexamethyldisilane were obtained from Aldrich
and used without further purification. The synthesis of 1,3-
diphenyl-1,2,2,3-tetramethyltrisilacycloheptane (5) was ac-
complished using a modified literature procedure.8
4H), 1.1 (m, 4H), 0.50 (s, 3H), 0.45 (s, 6H), 0.30 (s, 3H), 0.14
(s, 6H), -0.04 (s, 3H). 13C NMR (δ (ppm), CDCl3): 139.5 (Ph),
139.3 (Ph), 134.2 (Ph), 128.5 (Ph), 128.0 (Ph), 25.5 (CH2), 12.4
(CH2), -3.8 (SiMe), -5.3 (SiMe), -5.4 (SiMe). IR (neat, cm-1):
1717 (s), 1423 (s), 1362 (s), 1222 (s), 531 (s).
Molecu la r Bea m Stu d ies. All experiments were performed
in
a molecular beam photoionization time-of-flight mass
spectrometer system (R. M. Jordan and Co.). Precursor samples
entrained in He gas were introduced through a pulsed solenoid
gas valve (General Valve Series 9).9 To attain sufficient sample
vapor pressure when it was not available at room temperature,
the sample holder and, downstream from it, the pulsed nozzle
could be heated. The photolysis laser (Lambda Physik Lextra
200) was operated with ArF to produce 193 nm photons. In a
few experiments 222 nm photolysis pulses were also employed
(Questek Series 2000 excimer using KrCl). Unless noted
otherwise, the photolysis wavelength used was 193 nm. The
193 nm laser beam was loosely focused by a 30 cm lens just
below the pulsed valve, producing a spot across the gas jet of
approximately 2 mm2. The power measured immediately
before the lens was up to 20 mJ . The ensuing gas pulses, now
containing photolysis products, were passed through a 0.5 ×
4 mm slit skimmer to form a molecular beam. The downstream
distance between the pulsed valve and the photoionization
beam was approximately 8 cm. The delay between the pho-
tolysis and ionization lasers was varied using a Stanford
DG535 delay generator; it was typically 40 µs. The 118 nm
vuv photoionization pulses were produced by frequency-
tripling the 355 nm output from a Continuum YG680 Nd:YAG
laser (third harmonic) in a mixture of 9% Xe in Ar.10 The 355
nm pulses were focused into the tripling cell with a 15 cm lens,
(P h 2MeSi)2SiMe2. Ph2MeSiCl (0.076 mol, 16.0 mL) was
added dropwise with stirring for 15 min to a suspension of
cut lithium wire (0.43 mol, 3.0 g) and 20 mL of THF in a 250
mL three-neck round-bottom flask. After 18 h of stirring at
room temperature, an additional 60 mL of dry THF was added.
The resultant dark brown-green solution was cannulated into
an addition funnel and added dropwise with stirring for 1 h
to a flask containing Me2SiCl2 (0.041 mol, 5.0 mL). The solution
(7) (a) Duffaut, N.; Donogues, J .; Calas C. R. Acad. Sci. Paris, Ser.
C 1969, 967. (b) Puranik, D. B.; J ohnson, M. P.; Fink, M. J . Organo-
metallics 1989, 8, 770.
(8) Sakurai, H.; Kobayashi, Y.; Nakadaira, Y. J . Am. Chem. Soc.
1974, 96, 2657.
(9) Huang, Y.; Sulkes, M. Rev. Sci. Instrum. 1994, 65, 3868.
(10) Bramer, S. E.; J ohnston, M. V. Appl. Spectrosc. 1992, 46, 255.

Photochemistry of Organopolysilanes and -germanes
Organometallics, Vol. 19, No. 2, 2000 141
F igu r e 2. log(intensity) versus log(pulse energy) depen-
dence monitoring the 58 amu product peak of (Me2Si)6
following 193 nm photolysis. The 193 nm pulse energies
are the internal energy readings from the Lextra 200 laser.
As can be seen by the reference line of slope 1.00,
experimental fits are close to this slope. A slope of 1
indicates that the photolysis process was one photon.
from only a single photolysis mechanism, the slope of
the plots should equal the number of photolysis photons
involved. Typically for a larger peak, particularly mass
68 here, the plots are of better quality. In general, low
pulse energy points may be less reliable due to smaller
product peak sizes; plateau effects, due to saturation,
may take place at higher pulse energies.
For the mass peak of paramount interest, mass 58
(Me2Si), four separate series of 193 nm power depen-
dence measurements were all consistent with a one-
photon photolysis process. The 43 amu (MeSi) peak
appeared concomitantly with the 58 amu (Me2Si) peak
and was prominent at low laser fluences, suggesting
that this mass is a daughter ion of dimethylsilylene. The
290 amu (Me2 Si)5 peak decreased with increasing 193
nm photolysis energy, consistent with additional pho-
tochemistry further reducing its population.
The 68 amu peak is due to Si2C. Corroboration is
found by examination of peak ratios at masses 68, 69,
and 70, the latter peaks arising primarily from 29Si and
30Si. (At higher laser powers saturation effects can cause
the relative sizes of smaller mass isotopic peaks to be
unduly large.) A Si2C species would be expected to arise
from multiphoton processes; two series of power depen-
dence measurements are closely consistent with a two-
photon process. Mass peaks at 28 amu (Si), 40 amu
(SiC), and 56 amu (Si2) closely parallel the growth of
the 68 amu peak and are likely daughter ions of Si2C.
The 68 amu peak is noteworthy because of its size and
its appearance in the photolysis of a number of different
silanes.
F igu r e 1. Mass spectra series following 193 nm photolysis
of compound 3 and subsequent 118 nm photoionization.
The spectra shown correspond to increasing 193 nm pulse
energy (internal readings from Lextra 200 laser) from
bottom to top (lowest 10 mJ ), as indicated in the figure
legend.
and the cell was isolated from the molecular beam chamber
by a 63 mm focal length LiF lens. By offsetting the 355 nm
beam from the center of the quartz lens, it was possible to
have the two beams traverse slightly different paths through
the mass spectrometer, so that two-color photoionization did
not occur. The ions produced were mass-detected in a reflectron
time-of-flight mass spectrometer; the signal from the micro-
channel plate detector was digitized by a Tektronix 2440
digital oscilloscope.
Resu lts
P h otolysis of (Me2Si)6. Figure 1 shows mass spectra
as a function of photolysis beam power. A small product
peak, out of the figure range, also appears at 290 amu,
corresponding to (Me2Si)5, but the major photolysis
products all appear below 100 amu. Even with no
photolysis pulse at the nozzle (lowest trace) photoprod-
ucts still appear, principally at 73 amu, corresponding
to a Me3Si radical. These photoproducts arise from the
118 nm pulse. Growth of strong peaks can be seen at
28 amu (Si), 42 amu (SiCH2), 43 amu (MeSi), 58 amu
(Me2Si), and 68 amu (Si2C). Peak sizes for one species
compared to another do not necessarily reflect directly
relative species populations.
A 42 amu (SiCH2) peak appeared at higher laser
fluences and ultimately became larger than the 43 amu
(MeSi) peak. This peak also likely arises from a two-
photon process.
Separate series of power dependence measurements
were made for each of the foregoing mass products. Each
independent series was plotted in a log(peak area)
versus log(pulse energy) form. A representative plot for
the 58 amu mass product is shown in Figure 2. Gener-
ally for a given mass peak multiple independent power
dependence series were carried out. If a product arose
P h otolysis of (Me2Ge)6. Typical photolysis results
for the cyclic polygermane (Me2Ge)6 are shown in Figure
3. All significant photolysis-induced peaks are in the
mass range shown. The major structure arises from Ge
(predominant mass at 74 amu), MeGe (predominant
mass at 89 amu), Me2Ge (predominant mass at 104
amu), and Me3Ge (predominant mass at 119 amu). The

142 Organometallics, Vol. 19, No. 2, 2000
Borthwick et al.
F igu r e 3. Mass spectrum monitoring the following 193
nm photolysis of (Me2Ge)6 (2) and subsequent 118 nm
photoionization (upper scan). The middle scan is with 193
nm photolysis pulses only and the bottom scan with 118
nm pulses only.
F igu r e 5. Mass spectrum following 193 nm photolysis of
2-phenyl-2-(trimethylsilyl)hexamethyltrisilane (2) and sub-
sequent 118 nm photoionization (upper scan). The lower
scan is with 118 nm pulses only.
ated with definite species are at 68 amu (Si2C) and 105
amu (PhSi).
The product of primary interest is the silylene PhMeSi
(120 amu). Sufficiently high laser fluences do lead to
production of a small 120 amu product peak. Meaningful
power dependence studies on this peak were not pos-
sible, but several considerations indicate that the peak
arose from higher order processes. It does not appear
at lower laser fluences, when only weak product peaks
at 43 and 105 amu were evident. It is not certain that
the foregoing MeSi and PhSi fragments arose from one-
photon photolysis, but some other fragments that ap-
peared at stronger laser fluences such as Si2C (68 amu)
almost certainly arise from multiphoton processes. It
was also the case that for mono- and disilanes similar
laser fluences produced significant product peaks that
also probably arose from higher order photolysis pro-
cesses (below). In our earlier studies employing 266 nm
photolysis/photoionization6 no product at 120 amu was
detected, even for the strongest laser fluences.
F igu r e 4. Mass spectrum following 193 nm photolysis of
2-phenylheptamethyltrisilane (compound 1) and subse-
quent 118 nm photoionization (upper scan). The lower scan
is with 118 nm pulses only.
P h otolysis of 2-P h en yl-2-(tr im eth ylsilyl)h exa m -
eth yltr isila n e (2). Typical results in Figure 5 show
again a number of product peaks due to absorbance of
118 nm photons. The principal photochemical product
peaks are at 28, 43, 58, 68, and 73 amu. Aside from the
73 amu (Me3Si) peak, which may substantially arise
from one-photon 193 nm photolysis, the likely structures
of the other products suggest that they probably arose
from higher order processes. Much as with compound
1, a very small photoproduct peak sometimes appeared
at 178 amu, probably corresponding to Ph(Me3Si)Si. As
was the case with the silylene in compound 1, we do
not attribute its formation to one-photon 193 nm pho-
tolysis but would instead argue that it likely arose from
a higher order photolysis process. No such product was
observed in our earlier beam experiments using 266 nm
photolysis/photoionization pulses.11
peaks associated with Me3Ge are almost entirely due
to the photoionization and are not associated with the
193 nm photochemistry. There is more complexity for
each of the foregoing species in the germanium analogue
compared to silicon because germanium has five iso-
topes with significant natural abundance. Additional
complications, as seen in the MeGe peaks, can arise
from the loss of one or more hydrogens. Consistent with
the silicon analogue results, and of central interest to
us, is the production of a nontrivial amount of Me2Ge.
While photolysis of (Me2Si)6 resulted in significant
production of Si2C, there is no analogous observed
production of Ge2C (expected mass centered at 160 amu)
with (Me2Ge)6.
P h otolysis of 2-P h en ylh ep ta m eth yltr isila n e (1).
Representative broad-scale mass spectra (Figure 4)
show that the 118 pulses themselves produce many ionic
fragments. Significant product peak growth due to the
193 nm photolysis pulses is evident in the 40-60 amu
range. The product at 43 amu can likely be associated
with MeSi. Other photolysis peaks that can be associ-
(11) In ref 6, the parent mass was incorrectly identified as 338 amu
instead of 324 amu; similarly, the fragment corresponding to a loss of
one Me3Si unit should have been identified as mass 251 rather than
265.

Photochemistry of Organopolysilanes and -germanes
Organometallics, Vol. 19, No. 2, 2000 143
F igu r e 6. Mass spectrum following 193 nm photolysis of
1,2,2,3-tetramethyl-1,3-diphenyl-1,2,3-trisilacycloheptane
(5) and subsequent 118 nm photoionization (upper scan).
The lower scan is with 118 nm pulses only.
F igu r e 7. Mass spectrum following 193 nm photolysis of
vinyltrimethylsilane and subsequent 118 nm photoioniza-
tion (upper scan). The lower scan is with 118 nm pulses
only.
P h otolysis of 1,2,2,3-Tetr a m eth yl-1,3-d ip h en yl-
1,2,3-tr isila cycloh ep ta n e (5). Principal photolysis
peaks are shown in Figure 6. No significant photolysis
product peaks appeared at higher mass. In this instance
a significant 58 amu peak appeared, corresponding to
Me2Si. Similar results also were obtained using 222 nm
photolysis pulses. Three different independent power
dependence plots each showed the photolysis process to
be first order for the 58 amu peak.
P h otolysis of Vin yltr im eth ylsila n e, Hexa m eth -
yld isila n e, a n d P h en yltr im eth ylsila n e. Since the
linear polysilanes 1 and 2 produced trimethylsilyl
radical as a major photoproduct in the molecular beams,
independent trimethylsilyl radical generators were
sought in order to interpret their fragmentation pattern.
Vinyltrimethylsilane, hexamethyldisilane, and phenyl-
trimethylsilane were therefore examined as potential
F igu r e 8. Mass spectrum following 193 nm photolysis of
hexamethyldisilane and subsequent 118 nm photoioniza-
tion (upper scan). The lower scan is with 118 nm pulses
only.
sources of trimethylsilyl radical in the gas phase.12
A
trimethylsilyl (73 amu) peak was produced by all these
precursors upon 118 nm photoionization alone; however,
only vinyltrimethylsilane and hexamethyldisilane re-
sulted in substantial increases in this mass upon 193
nm photolysis irradiation.
The 193 nm photolysis of vinyltrimethylsilane (Figure
7) resulted in an increase in the 73 amu peak as well
as a small but clear 58 amu (Me2Si) product mass peak.
Smaller mass peaks were found at 58, 43, 42, and 28
amu.
to the appearance of new peaks at 77, 53, 52, 50, 46,
and 45 amu (Figure 9). A 68 amu peak was also evident
following photolysis of phenyltrimethylsilane. Assuming
that the product is again Si2C, it is necessary that an
initial photolysis product incorporating one silicon have
an encounter with a second species incorporating an-
other silicon. These considerations speak to a strong
propensity for formation of Si2C.
The 193 nm photolysis of hexamethyldisilane afforded
a more substantial peak at 58 amu along with an
increase of the 73 amu peaks, as shown in Figure 8. As
with the case of vinyltrimethylsilane, mass peaks at 58,
43, 42, and 28 amu were observed. In addition, a
prominent peak at 68 amu was observed. Scatter in
power dependence measurements did not allow for an
unambiguous determination of the photolysis order for
58 amu product formation, but 68 amu product forma-
tion was clearly second order.
Discu ssion
The present study extends and improves upon the
previous investigation in a number of significant ways.
First, two separate lasers were used, one for the
photolysis and one for the photoionization. The physical
separation of photolysis and photoionization events
allow for a more unambiguous determination of photo-
chemical power dependence in many cases. In addition,
the 118 nm ionization laser photons are sufficiently high
in energy (10.5 eV) to photoionize nearly all possible
photofragments.
Photolysis of phenyltrimethylsilane at 193 nm only
resulted in a slight increase of the 73 amu peak relative
Photoionization at 118 nm alone, however, occasion-
ally led to significant production of photoionic frag-
ments. This was clearly evident for the photoionization
(12) Hexamethyldisilane has been reported to be a source of trim-
ethylsilyl radical upon 193 nm photolysis in the gas phase. Shimo, N.;
Nakashima, N.; Yoshihara, K. Chem. Phys. Lett. 1986, 125, 303.

144 Organometallics, Vol. 19, No. 2, 2000
Borthwick et al.
Sch em e 1
dimethylsilylene to methylsilene (Me(H)SidCH2) is
believed to be nearly isothermal.13 Loss of methane by
either R-elimination or â-elimination would then give
rise to silylidene (:SidCH2) or silyne (HSitCH), respec-
tively, with a mass of 42 amu.14
Phenyltrimethylsilane, in contrast to vinyltrimethyl-
silane and hexamethyldisilane, was a poor generator of
trimethylsilyl radical. Only small amounts of 73 amu
ions were produced, as well as the expected fragment
ions at 58, 43, and 42 amu. Instead, new masses at 77,
68, 53, 52, 50, 46, and 45 amu were observed. Most of
these mass peaks appear to correspond to the fragmen-
tation of either benzene or phenyl radicals. Although a
small quantity of atomic Si at 28 amu was observed, no
other identifiable silicon-containing fragments were
found. This may be due to a much greater photoioniza-
tion cross section for the aromatic ring relative to the
organosilicon fragments. The photolysis of the noncyclic
polysilanes yielded masses consistent with the photo-
chemical cleavage of a single silicon-silicon bond. In
the case of PhSi(SiMe3)3, a large peak at 73 amu along
with masses at 58, 43, 42, and 28 amu clearly showed
the presence of trimethylsilyl radical. Additional peaks
at 50 and 52 amu arise from phenyl group degradation.
The photolysis of MePhSi(SiMe3)2 also gave mass frag-
ments at 73 and 43 amu, indicative of trimethylsilyl
radical formation. However, these mass peaks were
obscured by fragments from the photoionization of the
precursor trisilane and, consequently, the relative in-
tensities of these peaks were difficult to ascertain. Mass
fragments from the phenyl group were also conspicuous
at 52, 50, and 45 amu. For each trisilane, an extremely
weak peak due to the expected silylene was observed.
These peaks were only observed under the highest laser
F igu r e 9. Mass spectrum following 193 nm photolysis of
phenyltrimethylsilane and subsequent 118 nm photoion-
ization (upper scan). The lower scan is with 118 nm pulses
only.
of all precursors at 118 nm, whereupon fragment ions
occurred even in the absence of the photolysis laser.
Prior irradiation of the molecular beam by the photolysis
laser resulted in the additional appearance of molecular
ions of the photoproducts along with their respective
daughter ions.
To interpret the photochemical processes involving
higher molecular weight precursors, some simple mono-
and disilanes were examined. Vinyltrimethylsilane and
hexamethyltrisilane were found to be excellent photo-
chemical sources of trimethylsilyl radical (m/e ) 73
amu). In addition to the 73 amu peak, masses at 58,
43, and 28 amu were also found. These were ascribed
to daughter ions of trimethylsilyl radical corresponding
to the consecutive loss of methyl groups to give Me2Si,
MeSi, and Si atoms.
(13) At the best level of theory, Me2Si: is 2 kcal/mol more stable
than Me(H)SidCH2 with a barrier height of 41 kcal/mol. (a) Nagase,
S.; Kudo, T. J . Chem. Soc. Chem. Commun. 1984, 141. (b) Nagase, T.;
Kudo, T.; Ito, K. In Applied Quantum Chemistry; Smith, V. H. J r.,
Schaefer, H. F., III, Morokuma, K., Eds.; Reidel: Dordrecht, The
Netherlands, 1986; pp 249-267.
(14) On the SiCH2 energy surface, silyne is calculated to be 49.1
kcal/mol less stable than the global minimum silylidene. Isomerization
of silyne to silylidene is predicted to have little or no activation
barrier: Hoffman, M. R.; Yoshioka, Y.; Schaefer, H. F., III. J . Am.
Chem. Soc. 1983, 105, 1084.
A significant 42 amu peak was detected in the
photolysis of both vinyltrimethylsilane and hexameth-
yldisilane. One pathway to this product mass may arise
from the dimethylsilylene (Me2Si) daughter of the
trimethylsilyl radical (Scheme 1). Rearrangement of

Photochemistry of Organopolysilanes and -germanes
Organometallics, Vol. 19, No. 2, 2000 145
Sch em e 2
fluences used and certainly arose from multiphoton
processes.
Photolysis of the cyclic precursor (Me2Si)6, in stark
contrast to the linear and branched polysilanes, does
efficiently produce the silylene, Me2Si:. The appearance
of the Me2Si: peak (58 amu) is first order with respect
to photolysis laser intensity, indicating that it is the
result of a one-photon event. This is the first direct
observation of Me2Si in a molecular beam. In our
preliminary study,6 Me2Si was inferred indirectly from
the observation of the (Me2Si)5 coproduct. Additional
masses at 43 and 28 amu are likely fragments from the
photoionization of Me2Si, formed by consecutive losses
of methyl groups. Power dependence measurements on
the intensity of the (Me2Si)5 mass peak showed satura-
tion and eventual loss of intensity at very high laser
powers. This is probably due to additional photochem-
istry of (Me2Si)5, consistent with the known photoreac-
tivity of this compound in solution.15
The photolysis of the cyclic germanium compound
(Me2Ge)6 was completely analogous to that of (Me2Si)6.
Dimethylgermylene, Me2Ge:, was very cleanly produced
during the photolysis, giving the expected isotope cluster
centered at 104 amu. Isotope clusters centered at 89 and
74 amu correspond to fragmentation involving loss of
methyl groups.
Our results suggest a common mechanism for polysi-
lane photolysis in the gas phase in which the primary
photochemical event is the homolytic cleavage of a single
silicon-silicon bond. For noncyclic polysilanes, the
resulting silyl radicals recoil from each other and
become the ultimate detectable products. Homolytic Si-
Si cleavage in cyclic systems, on the other hand, yield
R,ω-diradicals which are capable of subsequent in-
tramolecular reactions. Silylene formation subsequently
becomes possible by an SH2 mechanism by which a Si-
Si bond is attacked by the distal silyl radical (Scheme
2). SH2 displacements of Si-Si bonds are well-prece-
dented reactions.17
The dramatic difference in photochemistry in going
from the cyclic compounds (Me2Si)6 and (Me2Ge)6 to the
radical producing linear and branched polysilanes may
rest in differences in electronic properties and/or mo-
lecular topology. For instance, the photoreactive excited
state of a cyclohexasilane may be so significantly
different from that of a linear trisilane that an intrinsi-
cally different photochemical outcome may result. On
the other hand, differences in photochemistry may be
simply attributed to whether the molecule is linear or
whether it is cyclic. In an attempt to distinguish these
possibilities, we examined the photochemistry of the
1,2,3-trisilacycloheptane 5 in the molecular beam.
In solution, 1,2,3-trisilacycloheptanes photochemically
extrude the central silicon atom as a silylene to give 1,2-
disilacyclohexanes.16 For the specific case of 5, this
photoextrusion occurs with retention of stereochemistry
for both silicon atoms in the 1,2-disilacyclohexane
product.8 For our purposes, these compounds are es-
sentially trisilanes with the terminal silicon atoms
tethered by an electronically insulating tetramethylene
bridge. These precursors therefore should be virtually
identical with the linear trisilanes in electronic struc-
ture but are constrained to a cyclic structure.
The trisilacycloheptane 5 is an efficient photochemical
generator of a silylene. This is apparent from the intense
peaks due to dimethylsilylene (58 amu) and its daugh-
ters at 43, 42, and 28 amu. Present with lesser intensity
were peaks at 50, 52, and 78 amu, indicative of phenyl
group fragmentation. The dramatic difference in pho-
tochemical outcome between this cyclic trisilane and the
two noncyclic ones strongly suggest that silylene (and
germylene) formation lies not in the nature of the
excited state but is dependent on the geometric con-
straints of a cyclic system.
The reaction rate for the second thermal step which
generates the silylene must be very high. In solution,
the trisilacycloheptane 5 is known to produce dimeth-
ylsilylene with the retention of stereochemistry at both
silicon atoms. Any intermediate diradical therefore must
extrude silylene at a rate greater than that of a silicon-
carbon bond rotation. The elusiveness of diradicals to
spectroscopic detection in the solution photolysis of
cyclic silylene precursors may be a testament to their
high reactivity and very short lifetimes.
The proposed mechanism may be cautiously extended
to the condensed phase, in which the silicon-containing
precursors 1-3 and 5 are all observed to generate
silylenes. Although linear and branched polysilanes give
only silyl radicals under molecular beam conditions,
they are highly efficient silylene precursors in solution
and in cold matrices.4,5 This apparent discrepancy is still
reconcilable by a common primary photochemical step.
In the condensed phases, the radical pair generated is
initially bound in either a solvent or matrix cage where
further reaction chemistry may take place. The SH2
displacement of the Si-Si bond of one radical by the
other may effectively generate the silylene and disilane
products in a mechanistic step similar to that proposed
for the cyclic precursors in the gas phase. Again in such
a case, the second bimolecular step for the production
of silylene must be very fast and efficient in order to be
competitive with radical separation from the cage.
Although it is tempting to accept such a SH2 mech-
anism as a general process for the production of si-
lylenes in the condensed phase, theoretical studies and
experimental results strongly suggest a concerted ex-
trusion under these conditions. Particularly convincing
is the photochemical extrusion of dimethylsilylene from
5 with the retention of stereochemistry16 and the
(15) In solution, (Me2Si)5 photochemically decomposes to (Me2Si)4
with the extrusion of Me2Si: Ishikawa, M.; Kumada, M. J . Organomet.
Chem. 1972, 42, 325.
(16) Sakurai, H.; Kobayashi, Y.; Nakadaira, Y. J . Am. Chem. Soc.
1971, 93, 5292.
(17) SH2 reactions are known to occur at Si-Si bonds: (a) Hosomi,
A.; Sakurai, H. J . Am. Chem. Soc. 1972, 94, 1384. (b) Nakadaira, Y.;
Sakurai, H. J . Organomet. Chem. 1973, 47, 61.

146 Organometallics, Vol. 19, No. 2, 2000
Borthwick et al.
photochemical ring contraction of a “racked” dodeca-
methylhexasilane with the concomitant extrusion of
dimethylsilylene.18 A recent detailed study by Kira on
the photochemistry of the linear trisilane Ph2Si(SiMe3)2
also strongly suggests that the extrusion of diphenyl-
silylene is concerted in solution.19
where bimolecular chemistry may occur. The general
appearance of this ion may be due to a remarkable
stability relative to other small silicon molecules, a fact
that possibly has astrophysical implications.
Con clu sion s
In most cases, it is believed that the concerted silylene
extrusion arises from excited singlet states of the
polysilane.20 While these reactive singlet states may
predominate in solution, other radical producing states
may be much more important for isolated molecules in
the gas phase. It has been proposed that both triplet21
and Rydberg states20a,22 may also be important in the
photochemistry of polysilanes. The observation of radi-
cal products in the gas-phase experiments may possibly
be the result of these states. Further experiments are
still needed, both in the gas phase and in solution, to
understand the relative importance of all of the various
reactive excited states of polysilanes.
Finally, in the photolysis of many organosilicon
precursors, a prominent mass at 68 amu is found. This
ubiquitous peak is ascribed to Si2C on the basis of
isotope cluster intensities. This bent triatomic molecule
has been characterized experimentally23 and theoreti-
cally.24 Whereas Si2C has been typically produced in the
past by laser vaporization of silicon carbide25 or vapor-
ization of solid silicon and carbon,26 we have for the first
time produced Si2C via photolysis of gas-phase precur-
sors. Where power dependences can be determined, the
appearance of Si2C is second order in photolysis laser
intensity. It is noteworthy that Si2C is present even
when the gas-phase precursor has one silicon atom,
possibly implying its formation under plasma conditions
Photochemical silylene precursors in the condensed
phase do not always photochemically generate silylenes
in the gas phase under molecular beam conditions. Of
the condensed-phase organosilylene precursors exam-
ined, only the cyclic polysilanes 3 and 5 give organosi-
lylenes in significant quantities. An analogous processes
is also seen for the cyclic polygermane 4 to give a
germylene. The linear and branched polysilanes 1 and
2, in contrast, give photochemical products derived from
the homolytic cleavage of a single Si-Si bond. The
difference in photochemical outcome between these two
sets of polysilanes appears to derive from molecular
topology (cyclic vs noncyclic) rather than inherent
electronic effects. This is most clearly demonstrated by
the strikingly different photochemical outcomes of the
cyclic trisilane precursor 5, which produces a silylene,
and the electronically similar linear trisilane 1, which
produces only silyl radicals.
A unifying mechanism which would account for the
observed photochemical behavior in the gas phase
involves the initial homolytic cleavage of a Si-Si bond
for all silicon precursors. For the noncyclic precursors,
the resulting silicon radicals cannot undergo further
bimolecular reactions in the gas phase. However, the
R,ω-diradicals produced from cyclic precursors can
subsequently react to give silylenes via an intramolecu-
lar scission. This mechanism may also be operable in
the condensed phase as well, but probably only as a
minor pathway. The observation of extensive homolytic
scission in the gas phase is likely due to excited states
not readily accessible in solution. Experiments to fur-
ther elucidate photochemical processes both in a mo-
lecular beam and in the condensed phase are currently
underway.
(18) Mazieres, S.; Raymond, M. K.; Raabe, G.; Prodi, A.; Michl, J .
J . Am. Chem. Soc. 1997, 119, 6682.
(19) Miyazawa, T.; Koshihara, S.; Liu, C.; Sakurai, H.; Kira, M. J .
Am. Chem. Soc. 1999, 121, 3651.
(20) (a) Halevi, E. A.; Winkelhofer, G.; Meisl, M.; J anoschek, R. J .
Organomet. Chem. 1985, 294, 151. (b) Balaji, V.; Michl, J . Polyhedron
1991, 10, 1265.
(21) Venturini, A.; Vreven, T.; Bernardi, F.; Olivucci, M.; Robb, M.
A. Organometallics 1995, 14, 4953.
(22) Bock, H.; Wittel, K.; Veith, M.; Wiberg, N. J . Am. Chem. Soc.
1976, 98, 109.
Ack n ow led gm en t. We gratefully acknowledge sup-
port from the Department of Chemistry, the DoE/Epscor
program, and the Center for Photoinduced Processes,
funded by the National Science Foundation and the
Louisiana Board of Regents.
(23) Presilla-Marquez, J . D.; Graham, N. R. M. J . Chem. Phys. 1991,
95, 5612.
(24) Bolton, E. E.; DeLieww, B. J .; Fowler, J . E.; Grev, R. S.;
Schaefer, H. J . III. J . Chem. Phys. 1992, 97, 5586 and references
therein.
(25) Weltner, W., J r.; McLeod, D., J r. J . Chem. Phys. 1964, 41, 235.
(26) Presilla-Marquez, J . D.; Graham, W. R. M. J . Chem. Phys. 1991,
95, 5612.
OM990880U