The Reaction of Hydrogen Atoms with Silyl Radicals; the Decomposition Pathways of Chemically Activated Silanes

Abstract

The reactions of hydrogen atoms with silane and the methylated silanes - with the exception of tetramethylsilane -have been investigated in a fast flow reactor. Under our experimental conditions hydrogen abstraction from the Si-H bond is followed by combination of hydrogen atoms with the corresponding silyl radicals. The molecules formed in this way are activated by about 375 kJ/mol of vibrational energy. Two decomposition channels have been unequivocally identified, namely the elimination of molecular hydrogen and of methane, both with concomittant formation of the respective silylenes. In a subsequent step, silylene inserts into the substrate under formation of disilanes. With increasing degree of methylation. stabilization of the activated molecule competes with decomposition and dominates the kinetics in the case of trimethylsilane. With methyl -and dimethyl-silane, methyl radicals are observed as an additional reaction product. On the basis of RRKM calculations it is unlikely that they originate from a direct decomposition of the activated molecules.
Absolute values for the room temperature rate constants of the abstraction reactions are given; for H+CH

Full text

The Reaction of Hydrogen Atoms with Silyl Radicals;
the Decomposition Pathways of Chemically Activated Silanes
K. Wörsdorfer. B. Reimann. and P. Potzinger
Max-Planck-Institut für Strahlenchemie. Mülheim (Ruhr)
Z. Naturforsch. 38 a, 896-908 (1983); received April 30, 1983
Dedicated to Prof. G. O. Schenck on the occasion of his 70th birthday
The reactions of hydrogen atoms with silane and the methylated silanes
—
with the exception
of tetramethylsilane - have been investigated in a fast flow reactor. Under our experimental
conditions hydrogen abstraction from the Si-H bond is followed by combination of hydrogen
atoms with the corresponding silyl radicals. The molecules formed in this way are activated by
about 375 kJ/mol of vibrational energy. Two decomposition channels have been unequivocally
identified, namely the elimination of molecular hydrogen and of methane, both with concomit-
tant formation of the respective silylenes. In a subsequent step, silylene inserts into the substrate
under formation of disilanes. With increasing degree of methylation. stabilization of the
activated molecule competes with decomposition and dominates the kinetics in the case of
trimethylsilane. With methyl- and dimethyl-silane, methyl radicals are observed as an additional
reaction product. On the basis of RRKM calculations it is unlikely that they originate from a
direct decomposition of the activated molecules.
Absolute values for the room temperature rate constants of the abstraction reactions are given;
for H + CH3SiH3, Arrhenius parameters have been determined.
Introduction
The reactions of hydrogen atoms with silane and
the methylsilanes have been the subject of a number
of studies [1], There exists agreement that the
primary step comprises a simple abstraction reac-
tion (1)
H + R3SiHH2 + R3Si. (1)
Investigating this type of reaction by the flow tech-
nique. the primary step is inevitably followed by a
very fast reaction (2)
H + R3Si -> R3SiH(v) (2)
not only under the condition of atoms in excess but
also when the substrate is introduced in ten to
hundred fold excess, due to the small rate constant
for the dimerisation of the silylradicals compared to
(2) [2]. The fate of the silane molecule generated in
(2) and possessing approximately 375 kJ/mol excess
energy has been investigated by Michael and co-
workers [3]. In the case of trimethylsilane complete
stabilization of the molecule activated by (2')
D + R3Si -> R3SiD(v) (2')
Reprint requests to Dr. P. Potzinger and Dr. B. Reimann.
Max-Planck-Institut für Strahlenchemie. Stiftstraße 34-36.
D-4330 Mülheim (Ruhr).
was found under flow conditions. This is in agree-
ment with findings in our group [4], In the case of
silane. mono- and dimethyl-silane all the activated
molecules were thought to decompose by simple
bond breaking processes, a silicon-hydrogen split in
the case of SiH4 and a silicon-carbon split other-
wise. Further reactions of the radicals with D atoms
lead ultimately to CD4 and SiD4.
Based on our knowledge at present it is improb-
able that silane and the methylsilanes activated by
(2) decompose as suggested by [3]. Energetically
much lower fragmentation pathways - silylene for-
mation with concomitant loss of H2 or CH4 — are
available and are not only realized in thermally
activated systems [5, 6, 7] but also in chemically
activated methylsilanes, produced by methylene
insertion into the Si—H bond [8].
It has been pointed out by us [9, 10] and others [3]
that flow experiments involving silanes are severely
hampered by wall effects and this also holds true
with pyrolysis experiments [5] leading in the latter
case at least to a contamination of the reaction
mechanism by radical reactions. Accordingly, the
scatter of the reported rate constants is large espe-
cially for the systems H/SiH4 and H/CH3SiH3 [1],
We have therefore decided to repeat the experi-
ments of Michael et al. [3] taking special care to
minimize wall effects. Rate constants as well as de-

composition pathways of vibrationally excited
silane, mono-, di- and trimethylsilane as measured
in a flow reactor are reported. The experimental
results are compared with RRKM calculations. To
test the quality of our results obtained in the flow
experiment we also measured the Arrhenius param-
eters for reaction (3)
H + CH3SiH3-+ H2+CH3SiH2 (3)
in a virtually wall free experiment by time resolved
Lyman a resonance absorption resp. resonance
fluorescence.
Experimental
SiH4 99.999%) was purchased from Air Li-
quide, CH3SiH3, (CH3)2SiH2 and (CH3)3StH from
PCR. He (99.996%), H2 (99.999%), and D2 (isotopic
purity 99.6%) were passed through molecular sieve
cooled by liquid nitrogen to remove any trace of
moisture. Ar (99.997%) was dried over P?05. Pro-
pane (Phillips research grade) was used as received.
Methyldisilane was prepared by reaction of disi-
lane with methyllithium [13]. 1,2-Dimethyldisilane
was prepared by reduction of 1,2-dimethyltetra-
chlorodisilane with LiAlH4. 1,1,2-Trimethyldisilane
and 1,1,2,2-tetramethyldisilane were not available as
pure substances for calibration purposes, only their
mass spectra are known from earlier work [14]. The
intensity of the prominent fragment ions of the four
methylated disilanes are given in Table 1. The mass
spectra of methyldisilane and 1,2-dimethyldisilane
are not in good agreement with published spectra
by Longeway and Lampe [15].
Our flow reactor is basically of the same design as
that of Wagner and coworkers [11]. Differences
occur in a) the mass spectrometer; a quadrupole
mass spectrometer has been used, and b) the signal
processing; the molecular beam was chopped in the
first vacuum chamber allowing phase sensitive de-
tection to be employed. A detailed description of
the apparatus is given in [12]. A special problem
was posed'in the flow experiments by the high wall
activity of the atom-silane systems. Measurements
of rate constants using a plain quartz reactor were
virtually impossible. Poisoning of the surface with
diluted HF. halocarbon wax, or methylated chloro-
silanes brought only a slight improvement. An ap-
preciable decrease of the influence of the walls has
been achieved by inserting teflon rings of 6 mm
height with a conical opening (smallest diameter
22.7 mm, largest diameter 25 mm) into the reactor.
The teflon rings were inserted with the sharp edge
of the smaller opening facing the streaming gas at
regular distances of 3.5 cm. The working hypothesis
made was that at high flow velocities the rings
would lead to a contraction of the flow leading to a
"walless" reactor. The effective cross section of the
reactor was reduced in this way by a factor of 1.5.
No extensive investigations to substantiate this pro-
position have been carried out. What has been
shown, however, is that in the case of the
D + (CH3)3SiH reaction, where wall effects are less
severe, consistent results are obtainable both with
and without the teflon rings. For the reaction
D + transbutene the rate constant k (D + f-C4H8) =
(7 ± 1) • 10~13 cm3 s-1 (experimental details: flow
velocity 25.3 ms"1, P = 425 Pa, [D]0=4.3• 1014cm-3,
[/-C4H8]0 = 4
•
1013cm~3) was obtained with teflon
rings in satisfactory agreement with the literature
[16].
The time resolved Lyman a absorption technique
has been described previously [17]. It is very similar
to a conventional flash photolysis-resonance absorp-
tion system, the main difference being the produc-
tion of H atoms by means of pulsed mercury radia-
tion. This is achieved by putting a fast mechanical
shutter in front of a DC powered low pressure
mercury lamp. The gas mixture in the reaction cell
Table 1. Intensities of the four most prominent fragment ions of four methylated disilanes.
Substance m/e
43 44 45 58 59 73 76 90 104 118
CH3Si->H5
l,2-(CH3)2Si2H4
l,l,l-(CH3)3Si2H3
l,l,2,2-(CH3)4Si2H2
65
53
40
46
100
72 49
54
100
100
86 100
22 69 22 44

consists of - 65 kPa He, - 1 kPa H2, - 0.1 Pa Hg,
and various, but well known amounts of substrate in
the order of a few Pa. The mercury light creates
thermal H atoms which react with the substrate
under pseudo first order conditions. Systematic
variation of the substrate partial pressure yields
absolute values of the bimolecular rate constant for
the reaction of H atoms with the substrate.
The advantage of this experimental setup lies in
the way the H atoms are produced. Up to the
present time, most investigations employing time
resolved resonance techniques for the study of H
atom reactions involved H20 photolysis as an atom
source. There the H atoms are generated with a
varying amount of undesirable translational energy,
depending on the wavelength of the photolysing
radiation. In addition, the creation of one OH
radical for each H atom invariably complicates the
kinetics. Neither of these disadvantages is present
with the chemical quenching of Hg(3P,) by H2. The
translational excess energy of the H atoms is very
low and no radical species are created besides H.
A few other points make this method convenient.
Firstly, the light pulses are obtained with high re-
producibility 5% pulse to pulse variation).
Secondly, absolute intensity calibrations are easily
performed by propane actinometry [18], and finally,
systematic variations of the light intensity are sim-
ply achieved by either varying an iris or by attenu-
ating the radiation by means of neutral density
filters.
A disadvantage is the limited shutter closing time
which sets an upper limit of ~ 250 s_l to the range
of pseudo first order rate constants.
A few improvements have been made over the
apparatus of [17]: In order to avoid product build-
up in the reaction cell, a slow flow system has been
incorporated. The premixed gases are supplied from
a 20
1
glass vessel and the pressure in the cuvette is
kept constant using an automatic pressure controller
(MKS 250A). The closure time of the shutter in
front of the Hg lamp was reduced to il ms by
inserting a circular disc (0 ~ 1 cm) to cut off the
central light rays. In the resonance fluorescence ex-
periments a combination of mechanical shutter and
electrical lamp shut off has been used. The light
pulses are generated by opening the shutter with a
moderately fast rise time of 2-3 ms. After a pre-
selected time interval, typically of 10-20 ms dura-
tion. the lamp is electrically shut off with a fall time
of less than 50 us. Immediately after the lamp shut
down the shutter is closed too and the mercury arc
is turned on again. The pulse repetition rate used in
these experiments is in the range of
1
to 0.1 s"1.
Photon pulse counting was employed exclusively
for signal detection. The data accumulated in the
Nicolet signal averager were subsequently trans-
ferred to a mainframe computer for further proces-
sing.
All errors quoted are single standard deviation
and are precision only.
Results
The system H/SiH4 and D/SiH4
The rate constants for the abstraction of hydrogen
from silane by hydrogen atoms (4)
H + SiH4-> H2+SiH3 (4)
as well as by deuterium atoms (4') have been deter-
mined in a flow experiment. For k (4) a value of
(4.4 ± 0.7) • 10~I3cm3s_l at room temperature was
obtained under the following experimental condi-
tions: the argon volume flow F varied in 6 experi-
ments from (1.4- 1.5) • 104 cm3 s~\ the pressure P
from 220-240 Pa, the initial atom concentration
H o ranged from (2-3.6) • 1014 cm-3 and the atom-
substrate ratio H 0/ SiH4 0 from 160 to 270. k (4')
equals (3.9 ± 0.7) • 10"l3cm3s~' and is an average
value of 7 determinations. The following experi-
mental conditions were used: 0.85 -lO4^/7^
1.4-104. 225 ^ P ^ 640. 2.6- 1014^ D 0 = 4.3-1014,
5 < D 0/ SiH4 0< 230. In all cases the disappear-
ance of SiH4 was measured at m/e 31. The H and D
atom concentrations have been determined mass
spectrometrically.
The products observed in the system D/SiH4 can
be inferred from Fig. 1 and Figure 2. At reactant
ratios SiH4 0/ D
o
= 2 only Si2H6 and Si2H5D are
observed in the ratio of ~ 2 and no deuterated
silanes are formed. For smaller SiH4 0/ Do ratios
higher deuterated disilanes are formed and SiD4 is
observed in increasing amounts. Interestingly, a
quantitative evaluation of the mass spectra of the
silane part (Fig. 2) does not show the appearance of
any partially deuterated silanes. Finally, it should
be mentioned that m/e 56 at medium reactant ratios
cannot be explained solely as a fragment peak of the
differently deuterated disilanes.

The system H/CH3SiH3 and D/CH3SiH3
For reaction (3)
H + CH3SiH3-+ H2+CH3SiH2 (3)
and its isotopic analogue (3'), hydrogen abstraction
by D atoms, the rate constants obtained by the flow
technique are k (3) = (3.9 ± 0.7) • 10~13 cm3 s-1 and
k (3') = (3.5 ± 0.4) • 1CT14. The following experimen-
tal conditions prevailed for reaction (3): 8 experi-
ments measured either on m/e 44 or m/e 45,
1.38- 104 ^ F^ 1.53
•
104, 226 ^P^ 246, 2.4 1014
^ H o^ 3.75- 1014, 160 ^ H 0/ CH3SiH3 0 = 250,
room temperature; for reaction (3') 5 experiments
measured on m/e 45, 1.35 • 104 ^ F 1.38 • 104,
P = 234, 3.4-1014 ^ |D!0^ 3.8 • 1014, 160^|D|0/
CH3SiH3 o = 230, room temperature.
The rate constant for reaction (3) has also been
determined in a real time experiment at pressures
ll 1
Si2H6
ll.
1
1 1
1
[SiH4]0/[D]0 = 2
l 1
.
1
1 1 ill
1
[SiHJ0/[D]0 = 0.2
.1,1
1
.1, 1
i
[SiHJo/[Dlo=0.05
1
1.
1
1
.Ii ,L
I
Si2D6
50 60 70 m/e
Fig. 1. Comparison of the mass spectra of Si2H6 and Si2D6
with product spectra of the silane system for different
SiH4n/ D
o
ratios.
100
öi
c
a>
c
0.03 0.05 0.1 0.2 0.6 1.0 2.0 [SiHj0/[D]0
1.8 6.6 oo SiH4/SiD4
(calculated)
Fig. 2. Profiles of masses m/e 28
—
m/e 34 for different
SiH4 0/ D o ratios.
around 1 bar where wall effects are negligible. Most
experiments have been done by a pulsed mercury
resonance photolysis — Lyman i resonance absorp-
tion technique. For determination of the room tem-
perature rate constant the following experimental
conditions applied: He = 1.5 • 1019 cm-3, H2
j
=
1.7 • 10l7cm~3, H0 varied from 5 • 1010- 1 • 1012
cm-3.
Because of the anticipated large difference in the
rate constants for (3) and (5)
H + CH3SiH2 CH3SiH3(v) (5)
a study of the intensity dependence of the apparent
rate constant was regarded advisable. Such a depen-
dence was indeed observed but in the resonance ab-
sorption experiments the limiting low intensity rate
constant has not been reached. We have therefore
extrapolated the pseudo-first-order rate constant by
assuming a linear dependence on the square root of
the mercury light intensity. This procedure was sug-

gested by a computer simulation of a reaction
system containing the three steps (3), (5) and (6)
2CH3SiH2 CH3H2SiSiH2CH3 (6)
to give a lower bound to the true rate constant. All
pseudo first order rate constants graphed in Fig. 3
by open circles were obtained in this way. In the
case of the resonance fluorescence experiments,
where a better signal to noise ratio has been
achieved, we believe that the true rate constant for
(3) has been reached (Figure 4). The values obtain-
ed by this method are filled in Fig. 3 by black
circles. From Fig. 3 one obtains for the bimolecular
rate constant for reaction (3) k (3) = (3.8 ± 0.2)
• 10~13cm3s_1.
Exploratory experiments on the temperature de-
pendence of k (3) have shown that the activation
energy obtained was the same within experimental
error regardless of whether the rate constants were
extrapolated to zero mercury light intensity or the
rate constants at a specific mercury light intensity
were used. The data shown in Fig. 5 were all
obtained with a mercury light intensity of 3.4- 1013
cm_3s_l. The activation energy obtained for reac-
tion (3) was £4(3) = 10.1 ± 0.8 kJ moP1. Using the
extrapolated room temperature rate constant an A
factor of A (3) = (2.3 ± 0.9) • 10_l
1
cm3 s-1 is ob-
tained.
ICH3SiH3j -10~U [cm"3]
Fig. 3. Dependence of the extrapolated first order rate
constant for the reaction H + CH3SiH3 on substrate con-
centration.
28
26
jfl 24
22
20
18
16
U
Fig. 4 Dependence of the pseudo first order rate constant
on mercury light intensity.
Product analyses for H + CH3SiH3 as well as
D + CH3SiH3 have been carried out with substrate in
excess. The results are shown in Figure 6. Peaks
with positive intensities are product peaks while the
negative intensities give the minimum amounts of
the reactant peaks that have disappeared. A closer
inspection of the spectra immediately suggests that
l,2-dimethyldisilane, methyldisilane and methane
are products of reactions initiated by (3). Methyl-
disilane has a parent peak at m/e 76 which is not
occupied by a fragment ion of dimethyldisilane
according to our reference spectrum (Table 1) as
well as that of Longeway and Lampe [15]. In the
system D/CH3SiH3 the formation of dimethyl-
disilane-d] and methyldisilane-d, is observed as
well and also CD4 besides CH4. The observation of
CD4 indicates the appearance of methyl radicals
which undergo repetitive atom addition followed by
unimolecular decomposition of vibrationally excited
methane [3]. Subsequently it was indeed possible to
show the presence of methyl radicals at high flow
velocities by the use of low voltage spectra. The
following relative ionization cross sections have
been determined o(m/e 16 <- CH4): o{m/el6+-
CH3Si2H5): a(m/e 90(CH3)2Si2H4)= 1.0:0.2:0.15
which allow us to deduce from Fig. 6 the following
relative product yields (CH3)2Si2H4 : CH3Si2H5 :
CH4 = 1 : 0.2 : 0.16. One further result can be ex-
tracted from Fig. 6b, the ratios (CH3)2H3DSi2 /
(CH3)2SiH4 - 0.4 and CH3Si2H4D / CH3Si2H5
^ 2.
The spectrum in Fig. 6 b might suggest that
methylsilane-d| is also a prominent product mani-

fested in the appearance of a peak at m/e 46. But as
can be seen from Fig. 6 a, the relative "negative"
intensity of m/e 45 is much too small as compared
to the mass spectrum of pure methylsilane, suggest-
ing that m/e 45 is a prominent product peak. This
is supported by the reference spectra of the methyl-
ated disilanes (see Table 1). In the D/CH3SiH3
system this product peak will partly show up at
m/e 46.
The system H/(CH3)2SiH2 and D/(CH3)2SiH2
The evaluation of the data in these two systems is
much more difficult due to a number of factors.
First of all lack of reference substances allows only a
qualitative interpretation of the product spectrum,
secondly partially stabilization of the chemically
activated dimethylsilane, generated in analogy to
reaction (5) and interference by product peaks
falsify the rate constant determinations.
In Fig. 7 the product spectrum of the system
D/(CH3)2SiH2 is shown. The peak group around
m/e 120 is due to tetramethyldisilane, partially iso-
topically substituted. More specifically we will as-
sume, for reasons given later, that the product is
1,1,2,2-tetramethyldisilane. According to the known
reference spectrum of this substance there appear to
be no fragment ions at m/e 104 and m/e 105, split-
ting off a methyl group yields m/e 103 of approxi-
mately the same intensity as the parent peak. This
means that m/e 104 and m/e 105 in the product
spectrum must be predominantly due to trimethyl-
disilane. Here again we find isotopically substituted
and unsubstituted products of approximately the
5 c
C D
01
c-e
ss
Fig. 6a. Mass spectrum of the system H + CH3SiH3 ob-
tained in the flow system. Peaks with positive intensities
are product peaks while the negative intensities give the
minimum amounts of the reactant peaks that have dis-
appeared.
Fig.
6 b.
Mass spectrum of the system D + CH3SiH3.
i • i—i— » i , i i . . i
2.0 2.5 3.0 3.5
,000/T[K-1!
same concentration. If the mass spectrometric sen-
sitivity for the parent peaks of tri- and tetramethyl-
disilane is not too different, then the two substances
are formed in approximately equal amounts. As in
the case of methylsilane we find CH4, CH3D, CD4
and also methyl radicals.
The loss spectrum of dimethylsilane shows much
too small intensities for m/e 58 and m/e 59. This can
be qualitatively explained by the very intensive
fragment peaks at m/e 58 and m/e 59 of tri- and
tetramethyldisilane which will show up partially at
m/e 60 in the case of reaction with D atoms. For
m/e 46 this explanation is insufficient because tri-
as well as tetramethyldisilane show only very small
fragment peaks at m/e 45. We therefore have to

üi c
C 3
C L.
— o
Fig. 7. Mass spectrum of the system D + (CH3)2SiH2.
assume that m/e 46 originates predominantly from
(CH3)2SiH2D formed by reaction (7)
(CH3)2SiH + D-> (CH3)2SiHD . (7)
If we neglect higher molecular weight product
contributions to m/e 46 and m/e 45 altogether then
we arrive at a stabilization fraction of deuterated
dimethylsilane formed via (7) of ~ 40%.
The rate constants for the abstraction reaction by
H atoms (8)
H + (CH3)2SiH2 H2 + (CH3)2SiH (8)
as well as by D atoms (8') have been determined in
a flow experiment with atoms in excess. The dis-
appearance of dimethylsilane as a function of reac-
tion time was monitored at m/e 59. (Other experi-
mental conditions for a) H/(CH3)2SiH2: 5 deter-
minations. F= 1.4 • 104. £ = 234. 1.9-1014 ^ H0
^ 2.6 • 1014, 80=1 H 0/(CH3)2SiH2 o^ 160, room
temperature; b) D/(CH3)2SiH2: 7 determinations,
F= 1.35- 104 P = 240. 3.4- 1014 ^ D 0 = 4.1 • 1014,
130 ^ D0/ (CH3)2SiH2 0 =i 360. room tempera-
ture).
The rate constants obtained are k (8) = (2.8 ± 0.2)
• lO-'-WV, k (8') = (2.2 ± 0.2) • 10~13 cm3 s-1.
Two error sources influence these values. Dimethyl-
silane formed by (7) leads to too small a decrease of
m/e 59 so that the rate constant k (8) has to be in-
creased by the stabilization fraction while k (8') has
to be increased by only half of this value (neglecting
isotope effects). The second error source concerns
the heavy contributions of tri- and tetramethyl-
disilane to m/e 59 but here one has to take into
account that for every disilane molecule formed a
second dimethylsilane molecule is used up (see
below) so that this error will largely cancel itself
out.
The system D/(CH3)3SiH
The rate constant of the reaction
D + (CH3)3SiH -> HD + (CH3)3Si (9')
has been determined in the usual way to k (9') =
(1.4 ±0.2)- 10-13 cm3 s"1.
Experimental conditions: 7 determi-
nations, substrate concentration measured at m/e 59.
1.07- lO4^/7^ 1.5- 102, 192 351, 3.7 • 1014
< D o = 6.6
•
1014, 15 D 0/(CH3)3SiH o^ 180,
room temperature.
The product spectrum is shown in Figure 8. With
the mass spectrum of trimethylsilane-d in mind [4]
it is immediately clear that (CH3)3SiD is the main
product. There are higher deuterated species formed
as well as can be seen from m/e 61, m/e 62 and
m/e 74. m/e 75. The appearance of a negative

20 50 60 70 -I
80 m/e
Fig. 8. Mass spectrum of the system D + (CH3)3SiH.
m/e 73 indicates that other reaction channels than
(10)
D + (CH3)3Si -* (CH3)3SiD (10)
contribute to the disappearance of the trimethylsilyl
radicals formed in (9'), but no higher molecular
weight products have been found. The product loss
seems to be dependent on the atom substrate ratio.
At large ratios the loss is negligible while at smaller
values it increases to almost 50%. Small amounts of
CD4 have also been found.
Discussion
It is generally well accepted [19] that the Si-H
bond dissociation energy in silane and the methyl-
ated silanes is hardly dependent on the degree of
methylation. Accordingly one expects no great
variation in the rate constants for hydrogen abstrac-
tion in going from silane to trimethylsilane. This
expectation is borne out by our experimental re-
sults. Our rate constant for H/SiH4 and D/SiH4 is in
very good agreement with earlier measurements
from our laboratory using a flow- [9] as well as a
pulse radiolysis- Ly i resonance absorption-experi-
ment [10], and with the recommended value by
Arthur and Bell [1] in their critical review on hydro-
gen abstraction reactions from silanes. This finding
is of importance for the main subject of the present
work because it shows that wall effects to which
flow experiments with silanes are especially prone
have been successfully suppressed. Michael and co-
workers [3] in their paper on the very same subject
obtained a much smaller rate constant with atoms in
excess (£app = 0.5- 10-l3cm3s-1) and a much higher
rate constant (&app = 52 • 10~13 cm3 s_l) with sub-
strate in excess.
The reported rate constants for hydrogen abstrac-
tion from methylsilane by H atoms cover almost a
power of ten ranging from (1.5 ± 0.4) • 10~13cm3s_1
[20] over (6.1 ± 1.0) • lO^cmV [21] to (11.5 ± 2)
• 10~13cm3s~' [3]. Here again the value obtained in
a flow experiment is the highest one. Our two room
temperature values obtained by different techniques
are in agreement with each other and also with the
value of Lampe et al. [21] within the combined error
limits. There exists no other measurement of the Ar-
rhenius parameters to compare our values with.
Activation energy as well as A factor are very
similar to the values obtained for the systems
D/SiH4 [9] and D/(CH3)3SID [17]. The small differ-
ences in the activation energies merely reflect the
minor variations in the SiH bond dissociation
energies, while the expected increase of A factors
with the number of Si—H bonds per molecule seems
to be buried in experimental uncertainties.
Our rate constants for the systems H/(CH3)2SiH2
and D/(CH3)2SiH2 are only lower bounds due to
difficulties inherent in these systems (see experi-
mental part). Relative rate determinations by Lam-
pe et al. [21] and Strausz et al. [22] yielded consider-
ably larger values.
The reason for reporting another set of rate
constants is to show that our efforts to diminish wall
effects were successful and that the product mea-
surements are not falsified by heterogeneous pro-
cesses.
Michael et al. found SiD4 as the only product of
the decomposition of chemically activated SiH3D(v)
generated by the reaction sequence (4') and (11)
D + SiH3 -»• SiH3D(v). (11)
They explained this by a primary Si—H bond split
of SiH3D(v) followed by successive D atom addi-
tion and Si-H bond rupture. For the thermal de-

composition of SiH4. Ring et al. [23] proposed the
same primary step while Purnell and his group [24]
favoured a silylene mechanism
SiH44siH2 + H2 (12)
which has been finally agreed on [25].
Our results at large SiH4 0/ D 0 ratios are in
qualitative accordance with the latter mechanism:
SiHjD(v) SiH2 + HD (13)
-> SiHD + H2 (14)
SiH2+SiH4 - Si2H6 (15)
SiHD + SiH4 -*• Si2H5D . (16)
Neglecting isotope effects, a ratio of one for the two
disilane products should be observed, while Si2H6
in twofold excess has been found experimentally.
An explanation could be reaction (17)
SiH3 + D SiH2+HD (17)
competing with (11). At smaller SiH4 0/ D 0 ratios,
SiD4 as well as higher deuterated disilanes are
observed (see Figs. 1 and 2). The reaction sequence
leading to these products is possibly initiated by a
reaction of the D atoms with SiH2.
There are additional reaction channels open for
methylated silanes excited by about 375 kJ/mol [19]:
1. Deactivation
(CH3)„SiH4_„(v) + M -> (CH3)„SiH4_„ + M , (18)
where in our case M is predominantly argon and
(18) is certainly a multistep process.
2. Elimination of molecular hydrogen with concom-
itant formation of silylene
(CH3)„SiH4_„(v) - H2 + (CH3)„SiH2_„. (19)
3. Elimination of methane with concomitant forma-
tion of silylene
(CH3)„SiH4_„(v) CH4 + (CH3)„_iSiH3_„. (20)
4. Elimination of molecular hydrogen with concom-
itant formation of silaethylene p
I ^
(CH3)„SiH4_„(v) - H2 + (CH3)„_,H3_„Si = CH2.
5. Breaking of a silicon-carbon bond
(CH3)„SiH4_„(v) - CH3 + (CH3)„_|SiH4_„. (22)
Reactions (19) and (20) have been observed in
the thermal decomposition of methylsilane [5, 6, 7],
For hydrogen elimination (19) an activation energy
of about 270 kJ/mole_l has been reported, while for
reaction (20) a somewhat higher activation energy
as well as a lower A factor has been estimated [7],
Both reaction channels are clearly observed in our
methylsilane system; the silylenes formed in reac-
tions (19) and (20) will yield the observed products
1,2-dimethyldisilane and methyldisilane via an in-
sertion reaction in methylsilane, e.g.
CH3SiH + CH3SiH3 CH3SiH2SiH2CH3. (23)
The observation of monodeuterated products in the
system D/CH3SiH3 is in accordance with this ex-
planation. By analogy with the methylsilane system
we associate the mass spectrum in Fig. 7 with
1.1.2.2-tetramethyldisilane and 1,1,2-trimethyldisila-
ne. In the case of trimethylsilane these two decom-
position pathways are not observable anymore due
to reaction (18).
Concerning the other reaction channels our results
do not allow similar definitive conclusions as for
reactions (19) and (20).
We expect increasing importance of reaction (18)
with increasing methylation and indeed there is no
SiH3D observed in the silane system while
(CH3)3SiD is clearly the main product in the
D/(CH3)3SiH system. A quantitative assertion to
the occurrence of (18) in the dimethylsilane system
is difficult to make and in the methylsilane system
even a qualitative statement cannot be made.
Reaction (21) has not been discussed in the in-
vestigations of the thermal decompositions of me-
thylsilane despite the fact that the exothermicity of
reaction (21) is at least as large as those for reac-
tions (19) and (20) [19]. Nothing is known about the
activation energy of reaction (21). If it is taking
place we would expect very fast addition of deu-
terium atoms to silaethylene [26] leading to a
chemically activated (CH3)„_,H3_„SiDCH2D(v)
which incorporates two D atoms. These two D
atoms should then, at least partially, show up in the
reaction products. Our results are ambiguous. In the
case of methylsilane we do not observe higher
deuterated disilane products while in the dimethyl-
silane case deuterated tetramethyldisilane shows up.
Deuteration at the methyl group is also observed
with trimethylsilane. But other reaction paths lead-

ing to these products can be envisaged and the
whole problem must be considered unsettled.
Methyl radicals are observed in the methyl- and
dimethyl-silane system and one is inclined to at-
tribute this product to reaction (22). But reaction
(22) is close to thermoneutral and despite its ex-
pected large A factor it is questionable whether it
can compete with the other decomposition channels.
In order to gain more quantitative insight we
carried out RRKM calculations.
The input data are given in Tables 2, 3, and 4.
The vibrational frequencies for the molecules have
been taken from Ball et al. [27] while the necessary
energetic parameters were taken from Walsh [19]
with the exception of the activation energies for
reactions (19) and (20). In the case of methyl- and
dimethylsilane the activation energy for (19) has
been measured by Neudorfl and Strausz [5] and for
(20) a value has been suggested by Davidson and
Ring [7] for the methylsilane system, all other values
are estimated. The transition state models for the
three processes have been chosen to achieve agree-
ment with A factors reported in the literature for
the same or similar reactions [5, 7, 28]. For reaction
(22) a Gorin type transition state has been used
with free tumbling motion of the two fragments.
However, in the case of trimethylsilane this leads to
too large an A factor and therefore the tumbling
motion has been restricted [29]. Direct count of
states has been employed for the sum of states of
the activated complex of reaction (22); in all other
cases the Whitten-Rabinovitch approximation has
been used.
The results of these calculations (Table 2, 3, and
4) substantiate our surmise made above, that reac-
tion (22) cannot compete with the energetically
favoured channels (19) and (20). In the case of me-
thylsilane reaction, (22) amounts to only 0.3% of the
total decomposition and this value may be even
further lowered by competing deactivating colli-
sions. Taking up a suggestion by Hase et al. [30] to
approximate the unknown Lennard-Jones param-
eters of the methylsilanes we can calculate a colli-
sion frequency of the activated molecules with the
carrier gas Ar of co = 2- 107s-1. In the case of di-
methylsilane the occurrence of reaction (22) can be
excluded with certainty. The calculations further
suggest that methylsilane cannot be stabilized under
our experimental conditions, that trimethylsilane is
completely and dimethylsilane is partially stabi-
lized. In this connection one has to keep in mind that
Ar is not a strong collider for the very exothermic
Table 2.
Model CH3SiH3 (CH3... SiH3)+ (CH3HSi... H2)* (H2Si ... CH4)*
2975(3) 3023(3) 2975(3) 2975(3)
2170(3) 2200(3) 2170(2) 2170(2)
Frequencies 1412(2) 1420(2) 1266 1412(2)
[crrr1] 1266 950(4) 1200(2) 1266
900(2) rotf (5) 800(2) 900
870(2) 701 800
750 600(2) 700(2)
701 500 600(2)
539(2) 360(2) 400(2)
roth rotf 300
/• (Si-C) [A] 1.87 5.8
Moments of inertia
[10-40 g cm2]
7V,/V 75.4 583
I. 15.1 15.7
7red (CH3) 3.4 3.4
/(CH,)a 2.95
/(SiH3)a 6.0
Symmetry number a b 9
18
c 3 1
4![kJ mol"1] 360 260 269
E*r [kJ moF1] 393 376 376
lg/Hs"1] 17.0 d 15.0 e 14.6 f
k(E*) [s-1] 2.2
•
107 6.6
•
109 1.0
•
109
a tumbling motion; b o= crext
•
erin(; c factor 2 comes from the tumbling motion of planar CH3; d T= 950 K; e T= 660 K;
f r=900 K.

Table
3.
Model (CH3)2SiH2 (CH3... Si(CH3)H2)* ((CH3)2Si...H2)* (CH3HSi... CH4)*
2950(6) 3023(3) 2950(6) 2950(6)
2145(2) 2950(3)
2145 2145
1440(4) 2200(2) 1260(2) 1440(4)
Frequencies 1260(2) 1440(2) 1200(4) 1260(2)
[cm"1] 930(2) 1420(2) 800(4)
930
[cm"1] 870(4)
1260
700(2) 870(2)
760 950
600(2)
800
700(2) 930(2)
500
700(3)
470 760 300
600(2)
223 700 200 350
roth(2) 870(2)
rotf
(6)
rotf
(2) 300
rotf
r
(Si—C)
[A] 1.87 5.8
Moments
of
inertia
[10"40
g cm2]
A 169.5 754
/, 51.2 71
I. 138.7 711
7red
(CH3) 4.9 4.9
/(CH,)
2.95
/(SiH2CH3)
30 a
Svmmetry number
a 18 18 18 3/2 b
£{j[kJ mol"1]
356 280 294
E* [kJ
mol"1]
395 378 378
\aA [s"1] 17.4 c 14.6 d 14.5 e
k(E*)[
s-1] 5.3
•
105 2
•
106 1.4
•
106
a
mean value;
b
factor 1
/2 due to
optical activity
of the
activated complex;e
T = 950 K; d T = 660 K;e T = 900 K.
Table
4.
Model (CH3)3SiH (CH3... SiH(CH3)2)* ((CH3)2Si...CH4)*
2950(9) 3023(3) 2950(9)
2125
2950(6) 1440(6)
Frequencies 1440(6) 2200 1260(3)
[cm"1] 1260(3) 1440(4)
905
905(2) 1420(2)
840
840(4) 1260(2)
800
710(2)
950 710
625
840(4) 700(3)
616(2)
710 625
252(2) 616(2) 616(2)
216 600 600
roth(3)
220
rotf
(3)
roth(4)
300
200(3)
rotf
(2)
/•
(Si-C)
[A] 1.87 5.8
Moments
of
inertia
[10-40
g cm2]
/v 153 802
/, 153 177
1. 260 888
4d(CH3)
5.1 5.1
/(CH,)
2.95
/(Si(CH3)2H)
80 a
Svmmetrv number
81 54 9
[kJ
mol-1]
363 293
E* [kJ
mol"']
397 387
lg A [s"1] 17.45 b 14.6 c
k (E*) [s"1] 6.6
•
102 1.7- 104

K. Wörsdorfer et al. • The Reaction of Hydrogen Atoms with Silyl Radicals
Table 5.
Model CH3SiH2 (CH3... SiH2)* (CH3)2SiH (CH3...Si(CH3)H)*
2975(3) 3023(3) 2950(6) 3023(3)
2170(2) 2200(2) 2145 2950(3)
1412(2) 1420(2) 1440(4) 2200
Frequencies 1266 950 1260(2) 1440(2)
[cnr1] 870(2) 750 930 1420(2)
750 rotf (5) 870(4) 1260
701 700(2) 950
600(2) 470 930
rotf 223
rotf (2)
870(2)
rotf (6)
/• (Si-C) 1.87 5.8 1.87 5.8
Moments of inertia
[10-40 g cm2]
A 74 520 166 747
Iy 74 520 136 710
I. 12 12.5 45 61
Eed (CH,) 3.4 3.4 4.9 4.9
/(CHO 2.9 2.95
/(SiH2), 7(SiHCH3) 5.7 30
Symmetry number 3 6 9 18/2 a
[kJ mol-1]15 253 247
E* [kJ rnol-^] 257 268
k(E) 4- 107 2.2
•
107
a factor 1/2 due to optical activity of the activated complex; b assuming zero activation energy for reaction (—26).
reactions (19) and (20) and co has to be multiplied
by a collisional deactivation efficiency factor.
We are still left with the question: where do the
CH3 radicals come from? There is of course the pos-
sibility of formation by wall reactions. A second
possibility might be a reaction sequence suggested
by O'Neal [31] and formulated here for the methyl-
silane system:
CH3SiH + H CH3SiH2(v), (24)
CH3SiH2(v) -*CH3 + SiH2. (25)
A number of assumptions must be made for this
mechanism:
1. Reaction (24) must be fast enough to compete
with the rapid silylene insertion reaction; under
our condition of high atom concentration this
might well hold true.
2. The difference in the activation energies of (24)
and the back reaction of (25) must be lower than
the exothermicity of (25).
3. The exoergic reactions (26), (27)
CH3SiH2(v) CH3Si + H2 (26)
SiH + CH4 (27)
must not compete with (25). This calls for activation
energies for the reversed reactions (—26) and (-27)
of about 80 kJ/mole. The activation energies for
(-19) and (-20) are about that order of magnitude
and higher values are expected for (—26) and
(-27), so condition 3 seems to be fullfilled. RRKM
calculations given in Table 5 suggest that 4 kJ/mol
and 20 kJ/mol of excess energy are necessary for the
CH3SiH2 radical and the (CH3)2SiH radical, resp.,
in order to decompose via (25).
Our experiments do not allow an extraction of
rate constants for the different decomposition path-
ways. Therefore, our RRK.M calculations cannot be
tested quantitatively. However, the agreement be-
tween theory and experiment with respect to the
stabilization of the dimethylsilane molecule indi-
cates that our calculations are quantitatively not too
far off. Simons and coworkers [30, 8] investigated
the decomposition of methylsilanes activated by
CH2 insertion into the Si-H bond e.g.
CH2 + CH3SiH3 -> (CH3)2SiH2(v) (28)
and gave rate constants for the different product
channels. Our calculated rate constants k( 19) =
6.7-10V, k (20) = 4.0 • 108 s-1 and k (22) =
6.5- 109s-1 are in reasonably good agreement with
the experimental results if the free available energy
of the dimethyl silane molecule is simply computed
from the exothermicity of reaction (28). Taking into

account an excess energy of methylene as high as
suggested by Simons [32] leads to values too large
for the various rate constants.
Conclusions
Silane and the methylated silanes produced by
reacting hydrogen atoms with the corresponding
silyl radicals decompose mainly by the ejection of
hydrogen or methane with the concomitant forma-
tion of silylenes. The observed formation of methyl
radicals is in all probability not due to a decomposi-
tion of the activated molecule, but rather to
secondary reactions. RRK.M calculations show that
the results obtained by thermal and chemical
activation are consistent.
[1] N. L. Arthur and T. N. Bell. Rev. Chem. Intermed. 2,
37 (1978).
[2] B. Reimann and P. Potzinger, to be published.
[3] J. A. Cowfer, K. P. Lvnch. and J. V. Michael, J. Phys.
Chem. 79,1139 (1975).
[4] M. A. Contineanu. D. Mihelcic, R. N. Schindler, and
P. Potzinger, Ber. Bunsenges. phys. Chem. 75, 426
(1971).
[5] P. S. Neudorn and O. P. Strausz. J. Phys. Chem. 82,
241 (1978).
[6] C. G. Newman. M. A. Ring, and H. E. O'Neal, un-
published results.
[7] I. M. T. Davidson and M. A. Ring, J. Chem. Soc.
Faraday Trans. I, 76, 1520 (1980).
[8] T. H. Richardson and J. W. Simons, Int. J. Chem.
Kinet. 10,1055 (1978).
[9] D. Mihelcic. P. Potzinger. and R. N. Schindler, Ber.
Bunsenges. phys. Chem. 78,82 (1974).
[10] D. Mihelcic.'V. Schubert, R. N. Schindler, and
P. Potzinger. J. phys. Chem. 81, 1543 (1977).
[11] K H. Homann. W. C. Solomon, J. Warnatz, H. Gg.
Wagner, and C. Zetsch, Ber. Bunsenges. Phys. Chem.
74,585 (1970).
[12] K. Wörsdorfer. Ph.D. Thesis. University of Essen,
1979.
[13] E. A. Groschwitz. W. M. Ingle, and M. A. Ring, J. Or-
ganomet. Chem. 9,421 (1967).
[14] P. Potzinger, A. Ritter, and J. Krause, Z. Naturforsch.
30 a, 347 (1975).
[15] P. A. Longeway and F. W. Lampe, J. Photochem. 14,
311 (1980).
[16] J. A. Kerr and M. J. Parsonage, Evaluated Kinetic
Data on Gas Phase Addition reactions, Butterworths,
London 1972.
[17] R. Ellul, P. Potzinger, B. Reimann. and P. Camilleri,
Ber. Bunsenges. Phys. Chem. 85,407 (1981).
[18] R. A. Back. Can. J. Chem. 37,1834 (1959).
[19] R. Walsh. Acc. Chem. Res. 14, 246 (1981).
[20] J.-G. Hong, Ph.D. Thesis, University of Detroit, 1972.
[21] E. R. Austin and F. W. Lampe, J. Phys. Chem. 81,
1134 (1977).
[22] K Obi, H. S. Sandhu, H. E. Gunning, and O. P.
Strausz. J. Phys. Chem. 76,3911 (1972).
[23] M. A. Ring. M. J. Puentes. and H. E. O'Neal. J. Amer.
Chem. Soc. 92,4845 (1970).
[24] J. H. Purnell and R. Walsh. Proc. Roy. Soc. London
A. 293, 543 (1966); P. John and J. H. Purnell. J. Or-
ganometal. Chem. 29, 233 (1971).
[25] C. G. Newman. H. E. O'Neal, M. A. Ring. F. Leska,
and N. Shipley. Int. J. Chem. Kinetics 11,1167 (1979).
[26] E. Bastian. P. Potzinger. A. Ritter. H.-P. Schuchmann,
C. von Sonntag, and G. Weddle, Ber. Bunsenges.
Phys. Chem. 84,56 (1980).
[27] D. F. Ball. P. L. Goggin. D. C. McKean, and L. A.
Woodward. Spectrochim. Acta 16, 1358 (1960).
[28] I. M. T. Davidson and A. V. Howard, J. Chem. Soc.
Faraday Trans. I, 71,69 (1975).
[29] S. W. Benson. Thermochemical Kinetics, 2nd ed.,
John Wiley, New York 1976.
[30] W. L. Hase, C. J. Mazac, and J. W. Simons, J. Amer.
Chem. Soc. 95,3454(1973).
[31] H. E. O'Neal, private communication.
[32] J. W. Simons and G. W. Taylor. J. Phvs. Chem. 73,
1274 (1969).