Vinyl dichlorosilane and vinyl dibromosilane (H2C=CH-SiHX2, X = Cl, Br): Conformational structure and vibrational properties determined by gas-phase electron diffraction, ab initio molecular orbital calculations, and variable-temperature Raman spectroscopy

Article abstract of DOI:10.1016/S0022-2860(00)00390-2

The molecular structures, conformations, vibrational spectra, and torsional potentials of vinyl dichlorosilane (VDC) H₂C=CH-SiHCl₂, and vinyl dibromosilane (VDB) H₂C=CH-SiHBr₂, have been studied using gas-phase electron diffraction (GED) data at 23-25°C and variable-temperature Raman spectroscopy, together with ab initio molecular orbital calculations. The GED data were handled by a dynamic theoretical model using a cosine Fourier potential function in describing the torsional coordinate. According to the GED refinements, these molecules exist in the gas phase at room temperature as a mixture of two minimum energy conformers, syn (torsional angle φ(CCSiH) = 0°) and gauche (torsional angle φ(CCSiH) ? 120°). Relevant structural parameters for syn-VDC are as follows: Bond lengths (r(g)): r(Si-C) = 1.847(5) A?, r(Si-Cl) = 2.042(2) A?, r(C=C) = 1.357(7) A?. Bond angles (angle(α)): angle CSiCl = 110.3(6)°, angle CCSi = 121.8°(calc.). Relevant structural parameters for syn-VDB are as follows: bond lengths (r(g)): r(Si-C)= 1.827(9) A?, r(Si-Br) = 2.206(2) A?, r(C=C) = 1.366(10) A?. Bond angles (angle α): angle CSiBr = 110.1(8)°, angle CCSi=121.7°(calc.). Uncertainties are given as 2σ (σ includes estimates of uncertainties in voltage/height measurements and correlation in the experimental data). From the variable-temperature Raman investigation in the liquid phase, the energy differences are: VDC, ΔE°(S-G) = + 0.11 ± 0.06 kcal mol^-1; VDB, ΔE°(S-G) = + 0.23 ± 0.07 kcal mol^-1. The Raman energies are average values obtained from two separate line doublets for each molecule, and they have been used in the GED least-squares refinements as valuable constraints. (C) 2000 Elsevier Science B.V.

Full text of DOI:10.1016/S0022-2860(00)00390-2

Journal of Molecular Structure 550–551 (2000) 257–279
www.elsevier.nl/locate/molstruc
Vinyl dichlorosilane and vinyl dibromosilane
ꢀH2CyCH–SiHX2; X ˆ Cl; Br† : conformational structure and
vibrational properties determined by gas-phase electron diffraction,
ab initio molecular orbital calculations, and variable-temperature
Raman spectroscopyଝ
a
a,
b
c
b
a
*
¨
T.H. Johansen , K. Hagen , K. Hassler , A.D. Richardson , U. Patzold , R. Stølevik
aDepartment of Chemistry, Norwegian University of Science and Technology, NTNU, N-7491 Trondheim, Norway
b
¨
¨
Institut fur Anorganische Chemie, Technische Universitat Graz, Stremayrgasse 16, A-8010 Graz, Austria
cDepartment of Chemistry, Oregon State University, Corvallis, OR 97331-4003, USA
Received 19 August 1999; received in revised form 30 September 1999; accepted 8 October 1999
Abstract
The molecular structures, conformations, vibrational spectra, and torsional potentials of vinyl dichlorosilane (VDC)
H2CyCH–SiHCl2; and vinyl dibromosilane (VDB) H2CyCH–SiHBr2; have been studied using gas-phase electron diffraction
(GED) data at 23–25ЊC and variable-temperature Raman spectroscopy, together with ab initio molecular orbital calculations.
The GED data were handled by a dynamic theoretical model using a cosine Fourier potential function in describing the torsional
coordinate. According to the GED refinements, these molecules exist in the gas phase at room temperature as a mixture of two
minimum energy conformers, syn (torsional angle fꢀCCSiH† ˆ 0Њ† and gauche (torsional angle fꢀCCSiH† Ϸ 120Њ†: Relevant
ꢀ
ꢀ
structural parameters for syn-VDC are as follows: Bond lengths (rg): rꢀSi–C† ˆ 1:847ꢀ5† A; rꢀSi–Cl† ˆ 2:042ꢀ2† A; rꢀCyC† ˆ
ꢀ
1:357ꢀ7† A: Bond angles (Єa ): ЄCSiCl ˆ 110:3ꢀ6†Њ; ЄCCSi ˆ 121:8Њ (calc.). Relevant structural parameters for syn-VDB are
ꢀ
ꢀ
ꢀ
as follows: bond lengths (rg): rꢀSi–C† ˆ 1:827ꢀ9† A; rꢀSi–Br† ˆ 2:206ꢀ2† A; rꢀCyC† ˆ 1:366ꢀ10† A: Bond angles ꢀЄa†:
ЄCSiBr ˆ 110:1ꢀ8†Њ; ЄCCSi ˆ 121:7Њ (calc.). Uncertainties are given as 2s (s includes estimates of uncertainties in
voltage/height measurements and correlation in the experimental data). From the variable-temperature Raman investigation
in the liquid phase, the energy differences are: VDC, DEЊSϪG ˆ ϩ0:11 ^ 0:06 kcal molϪ1; VDB, DEЊSϪG ˆ ϩ0:23 ^
0:07 kcal molϪ1: The Raman energies are average values obtained from two separate line doublets for each molecule, and
they have been used in the GED least-squares refinements as valuable constraints. ᭧ 2000 Elsevier Science B.V. All rights
reserved.
Keywords: Vinylsilane; Conformation; Gas electron diffraction; Raman spectroscopy; Torsional potential
1. Introduction
ଝ
Dedicated to Professor James R. Durig on the occasion of his
65th birthday.
We have recently studied the structures and confor-
mational properties of methyl vinyl difluorosilane
(VFS), methyl vinyl dichlorosilane (VCS),
* Corresponding author. Tel.: ϩ 47-73-596223; fax: ϩ 47-73-
596255.
E-mail address: kolbjorn.hagen@chembio.ntnu.no (K. Hagen).
0022-2860/00/$ - see front matter ᭧ 2000 Elsevier Science B.V. All rights reserved.
PII: S0022-2860(00)00390-2

258
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Fig. 1. Models showing the atom numbering for the syn and gauche vinyl dihalosilane conformers (X ˆ Cl or Br).
H2CyCH–SiX2–CH3 ꢀX ˆ F; Cl† [1], and perchloro-
vinylsilane (PCV), Cl2CyCCl–SiCl3 [2], by gas-
phase electron diffraction (GED), ab initio molecular
orbital calculations, and variable temperature Raman
spectroscopy. It has been established that the mole-
cules VFS and VCS exist as a two-conformer mixture
in the gas phase, one conformer with the methyl group
eclipsing the planar vinyl group (syn), and the other
with the halogen atom eclipsing the vinyl group
(gauche). Alternatively the experimental results for
PCV [2] show the anti conformer (CyC–Si–Cl
torsional angle f ˆ 180Њ† to be more stable in the
gas phase than the syn form by 1:0 ^
0:6 kcal molϪ1: In the syn form the silicon–chlorine
bond eclipses the vinyl group.
The GED analysis of PCV was made using a
dynamic model [2], because the syn form exhibits a
very shallow torsional potential minimum, and a large
torsional motion of the –SiCl3 group. Ab initio calcu-
lations clearly indicated the shallow minimum, and
the experimental potential curve resulting from the
GED analysis also strongly suggested the same [2].
However, the similar molecule vinylchlorosilane
ꢀH2CyCH–SiH2Cl†; which has been thoroughly
studied both theoretically and experimentally [3],
has been found to have a torsional potential with
two “deep” minima corresponding to the syn and
gauche conformers. The syn-to-gauche energy barrier
is 2.09 kcal molϪ1 and the gauche-to-syn barrier is
1.74 kcal molϪ1. The torsional motion of vinylchlor-
osilane was described by a cosine Fourier potential
function (V(f)) calculated from observed torsional
transitions for the two conformers [3]. The syn
conformer (syn-chloro) was found to be lower in
energy by 0:354 ^ 0:054 kcal molϪ1: In comparison,
the syn-to-anti barrier in PCV was experimentally
determined from GED data to be ca. 0.5 kcal molϪ1
[2]. This value is probably somewhat large. The calcu-
lated ab initio value was much lower at only
0.08 kcal molϪ1. The two vinylsilanes, VDC and
VDB (Fig. 1), are a natural extension of our earlier
work on these types of organosilicon compounds
[1,2]. It is of interest to determine whether the syn-
preference in relation to the vinyl group also holds
when the halogen is a large bromine atom. An
accepted model for explaining the syn-preference
phenomenon was recently reviewed by Page et al.
[4]. In short, this model regards the double bond as
consisting of a pair of bent single bonds, giving the
syn form a completely staggered possibility. This
explains why only the syn conformer was observed
in vinyl trimethyl silane ꢀCH2yCH–SiꢀCH3†3† [4].
Some of the difficulties in applying the GED
method to these types of organosilicon molecules
may be indicated from the study of vinyl dimethyl
chlorosilane, H2CyCH–SiClꢀCH3†2; studied using
unconstrained GED data in 1982 [5]. In this study, it
was concluded that a 100% syn (syn-chloro) model
could easily be used to fit the experimental data.
However, the gauche (syn-methyl) form could not
be entirely excluded as up to 30% could be included
without making the agreement noticeably worse.
Refinement of the mole fraction parameter in a
model containing both conformational forms resulted
in a gauche mole fraction of 0.17 with a standard
deviation of 0.12. Inclusion of more than 40% of the
gauche conformer made the fit between experimental
and theoretical RD-curves worse in the conformation-
ally sensitive regions. This would suggest that in the
gas phase, the syn conformer of vinyl dimethyl

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
259
differences between silicon and carbon (alkane and
alkene) compounds, as well a general interest in
expanding knowledge in the organosilicon structural
chemistry. This article presents a room temperature
structural investigation of VDC and VDB, using
dynamic GED models of the Fourier cosine-potential
type, constrained with information from liquid phase
variable-temperature Raman studies, and from ab
initio MO and density functional theory (DFT) calcu-
lations at the Hartree–Fock (HF), Møller–Plesset
(MP2) and Becke3/Lee–Yang–Parr (DFT/B3LYP)
levels of theory.
2. Experimental section
2.1. Synthesis
Scheme 1.
2.1.1. General remarks
chlorosilane is more stable than the gauche form.
However, a recent vibrational investigation [6]
based on a variable temperature study of the infrared
spectrum with the sample dissolved in liquid xenon,
found the gauche form to be lower in energy by
0:117 ^ 0:026 kcal molϪ1: These experimental condi-
tions had been thought to give results close to those in
the gas phase, but evidently do not produce agreement
with the GED results.
All operations with trifluoromethanesulfonic acid
or halogen compounds were carried out under a
nitrogen atmosphere. Trifluoromethanesulfonyloxy,
chloro and bromo derivatives of silicon are highly
sensitive to moisture. In contrast, they are stable for
several hours when exposed to dry air. Solvents were
distilled from potassium, lithium halides were dried
by heating for several hours in vacuo. Trifluorometha-
nesulfonic acid was used as received from Merck–
Schuchardt (Art. 821166). 29Si-NMR spectra were
measured with a BRUKER MSL 300 MHz spectro-
meter, chemical shifts are referenced against tetra-
methylsilane (TMS).
It is of great importance to use additional theore-
tical or experimental data when carrying out modern
GED investigations. It would seem particularly
important, in the case just described, to use additional
information to reasonably constrain the difference in
the C(vinyl)–Si and the Si–C(methyl) bond distances
in each conformer. This difference is very difficult, if
not impossible, to obtain precisely from GED
measurements. The difference in bond length for
vinyl dimethyl chlorosilane, obtained from the GED
2.1.2. Preparation of the compounds
Both H2CyCH–SiHCl2 and H2CyCH–SiHBr2
were
prepared
from
vinyldiphenylsilane
H2CyCH–SiHPh2 which reacts quantitatively with
trifluoromethanesulfonic acid CF3SO3H ((TfOH)
forming bis(trifluoromethanesulfonyloxy)vinylsilane
H2CyCH–SiHꢀOTf†2: Subsequent treatment with
LiCl or LiBr affords VDC and VDB which can be
purified by distillation. The purity of the samples
was checked with 29Si-NMR spectroscopy as well as
infrared and Raman spectroscopy. In the following
reaction (Scheme 1 (steps necessary for the prepara-
tion of VDC and VDB)), all the necessary steps are
summarized.
˚
data in Ref. [5], is 0.038 A, while ab initio calcula-
tions performed on several methyl vinyl silanes give a
˚
difference close to 0.010 A [1,4,6]. For VCS we found
earlier [1] that ab initio calculations at the MP2/6-
311G(d) level of theory gave a bond length difference
˚
˚
of 0.006 A, compared to the value of 0.023 A used by
Naumov et al. [7].
Our continuing studies on the vinyl silane mole-
cules [1,2] primarily stem from the interest in obser-
ving the potential structural and thermodynamic

260
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
2.1.2.1. Preparation of chlorodiphenylvinylsilane
H2CyC H –S i C l P h2. H2CyC H –S i C l3 (50.0 g,
309.6 mmol) is dissolved in 200 ml of diethylether
(Et2O) and a solution of 620 mmol of PhMgBr
(prepared from 97.3 g of C6H5Br and 14.6 g of Mg)
is added dropwise over a period of approximately 8 h.
Dry toluene (2000 ml) is added to the reaction mixture
and all diethylether is then removed by distillation.
The salts are separated by filtration and the filtrate is
fractionated in vacuo. H2CyCH–SiClPh2 (48.9 g,
65%) is obtained as an oily, colourless liquid,
b:p:0:05 ˆ 125–135ЊC: The chlorodiphenylvinylsilane
prepared by this procedure always contains a small
amount of bromodiphenylvinylsilane formed by Cl/
Br exchange.
2.1.2.4. Preparation of vinyl dichlorosilane,
H2CyCH–SiHCl2; and vinyl dibromosilane, H2Cy
CH–SiHBr2. H2CyCH–SiHꢀOTf†2; prepared accord-
ing to the procedure described above from 6.7 g
(31.9 mmol) of H2CyCH–SiHPh2; is dissolved in
approximately 40 ml of dry di-n-butylether (n-
Bu)2O. The solution is cooled to Ϫ40ЊC and approxi-
mately 9 g (210 mmol) of dry LiCl is added. The
suspension is stirred vigorously at that temperature
for 8 h and then allowed to slowly warm up to room
temperature. H2CyCH–SiHCl2 is then distilled off the
reaction mixture at a temperature of 60–70ЊC and
purified by fractionation. It is a colourless liquid,
b:p:760 ˆ 62ЊC: The yield is 1.5–2.0 g (37–49%).
For the preparation of H2CyCH–SiHBr2; bis(tri-
fluoromethanesulfonyloxy)vinylsilane is dissolved in
approximately 40 ml of dry Et2O. At Ϫ40ЊC, 6 g
(69 mmol) of LiBr is added. The mixture is again
stirred vigorously for several hours. Then all volatile
compounds are separated by evaporation in vacuo.
H2CyCH–SiHBr2 ꢀ3:0–4:0 g ˆ 43–58%† is separated
from the solvent and purified by fractionation under
reduced pressure, b:p:100 ˆ 40ЊC:
29Si{1H}-NMR: dˆ Ϫ20:2 ppm; singlet; ꢀH2Cy
CH–SiBrPh2 : Ϫ21:3 ppm†
2.1.2.2.
Preparation
of
diphenylvinylsilane,
H2CyCH–SiHPh2. A solution of 150 mmol of
LiAlH4 in 200 ml of Et2O is added dropwise to a
solution of 50.0 g (204.5 mmol) of H2Cy
CH–SiClPh2 in approximately 250 ml of Et2O. The
reaction mixture is cooled with an ice bath (0ЊC)
during the process and then allowed to come to
room temperature. It is then refluxed for several
hours to complete the reaction. Subsequently, excess
LiAlH4 is destroyed by slowly adding 1000 ml of 1 N
H2SO4 to the ice-cooled reaction mixture through a
dropping funnel. The organic layer is then separated,
dried over Na2SO4 and is finally fractionated in vacuo.
H2CyCH–SiHPh2 (43.8 g, 85%) is isolated as a
viscous, colourless liquid, b:p:0:05 ˆ 82ЊC:
29Si-NMR:
◦ H2CyCH–SiHCl2 : d ˆ Ϫ3:2 ppm;
1JꢀSiH† ˆ 287:5 Hz;
◦ H2CyCH–SiHBr2 : d ˆ Ϫ17:2 ppm;
1JꢀSiH† ˆ 287:6 Hz:
2.2. Infrared and Raman spectra
29Si-NMR: d ˆ Ϫ20:8 ppm; 1JꢀSiH† ˆ 197:8 Hz:
Infrared spectra of films between CsBr plates in the
range 4000–250 cmϪ1 were obtained on a Perkin–
Elmer Model 883 infrared spectrophotometer.
Raman spectra from 4000 to 50 cmϪ1 were recorded
with a Jobin–Yvon T61000 triple monochromator
spectrometer equipped with an argon laser operating
on the 514.5 nm line. Samples of VDC and VDB were
distilled into 1 mm capillary tubes which were then
sealed under an N2 atmosphere. The capillary was
mounted on a copper block, placed in a vacuum
chamber that was cooled with liquid nitrogen, and
heated resistively. The temperature of the sample
was monitored by a thermocouple. The reported
2.1.2.3. Preparation of bis(trifluoromethanesulfonyl-
oxy)vinylsilane, H2CyCH Ϫ SiHꢀOTf†2. H2CyCH–
SiHPh2 (6.7 g, 31.9 mmol) is dissolved in approxi-
mately 40 ml of toluene. The solution is cooled to
Ϫ40ЊC and 4.88 g ꢀ32:5 mmol ˆ 2:85 ml† CF3SO3H
is added dropwise. The reaction mixture is stirred for
several hours and allowed to warm up to room
temperature. Then the solvent, the benzene formed
during the reaction, and the excess of CF3SO3H is
removed in vacuo. The oily residue consists of pure
H2CyCH–SiHꢀOTf† whose purity can be checked
2
with 29Si-NMR spectroscopy.
lines are believed to be accurate to within ^3 cmϪ1
.

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
261
Fig. 2. Intensity curves ꢀsImꢀs†† for vinyl dichlorosilane (VDC). The experimental curves are averages of 3 traces for each photograph at the two
camera distances ꢀ3 ϩ 3†: The theoretical curve was calculated from the structural parameters given in Table 6. The difference curves are
experimental minus theoretical.
2.3. Gas electron diffraction
tip temperature of 296–298 K, and sample bath
temperatures of 223 K (VDC) and 258 K (VDB).
Nozzle-to-plate distances were nominally 750 and
300 mm for long (LC) and middle (MC) camera experi-
ments. A voltage-to-distance calibration was made
The samples of VDC and VDB were used without
further purification. Electron diffraction patterns were
recorded on Kodak projector slide plates using the
apparatus at Oregon State University. Experimental
conditions included an accelerating voltage of 60 kV,
ꢀ
using CO2 as the reference (raꢀCyO† ˆ 1:1646 A;
…
ꢀ
raꢀO O† ˆ 2:3244 A†: Three diffraction photographs
˚
from both the long and middle camera distances were
an average electron wavelength of 0.04905 A, a nozzle-

262
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Fig. 3. Intensity curves ꢀsImꢀs†† for vinyl dibromosilane (VDB). The experimental curves are averages of 3 traces for each photograph at the two
camera distances ꢀ2 ϩ 4†: The theoretical curve was calculated from the structural parameters given in Table 6. The difference curves are
experimental minus theoretical.
Ϫ1
˚
used in the analysis of VDC, and two photographs from
the long and four from the middle were used in the VDB
analysis. Each photograph was traced three times and
ments, at intervals of Ds ˆ 0:25 A . The data were
reduced in the usual way [8–10].
A calculated background [11] was subtracted from
the data for each plate to yield experimental intensity
curves in the form sImꢀs†: The averaged experimental
intensity curves for VDC are shown in Fig. 2, and the
corresponding curves for VDB in Fig. 3. Figs. 4 and 5
display the final experimental radial distribution (RD)
the results averaged to give a total of 6 data sets for each
ꢀ Ϫ1
compound. Data ranges are 2:00 Յ s=A
Յ
ꢀ Ϫ1
15:50 and 8:00 Յ s=A Յ 39:00 for the long and
ꢀ Ϫ1
middle camera VDC experiments and 2:00 Յ s=A
Յ
ꢀ Ϫ1
15:00 and 8:00 Յ s=A Յ 37:00 for the VDB experi-

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
263
Fig. 4. Radial distribution curves for vinyl dichlorosilane (VDC). The experimental curve was calculated from a composite of the six average
Ϫ1
2
ꢀ Ϫ1
Յ
experimental curves shown in Fig. 2 multiplied by the factor ZClZClꢀAClACl
†
expꢀϪ0:002s †: Theoretical data for the region 0 Յ s=A
1:75 were added to the composite curve prior to the transformation. The difference curve is experimental minus theoretical. The vertical lines
indicate the interatomic distances, with lengths proportional to the distance weights.
curves for VDC and VDB, respectively. The RD-
curves were calculated in the conventional way from
the modified molecular intensity curve I0ꢀs† ˆ
3. Structure analysis
3.1. Infrared and Raman spectra
Ϫ1
sImꢀs†ZXZXꢀAXAX† expꢀϪ0:002s2†; where X ˆ Cl;
Br; A ˆ s2F and F is the absolute value of the
complex electron scattering amplitudes. A theoretical
inner part was added to the experimental intensity
data to represent the unobserved region, 0:0 Յ
3.1.1. Vibrational assignment
Tables 1 and 2 summarize the observed Raman
wavenumbers for liquid and solid VDC and VDB as
well as the infrared wavenumbers of the liquids.
Because of the moderate resolution of the infrared
spectrometer available (Ͼ3 cmϪ1), gas phase spectra
have not been recorded. Even after repeated
ꢀ Ϫ1
s=A Յ 1:75; before the RD curve was calculated.
The scattering amplitudes and phases were taken
from tables in Ref. [12].

264
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Fig. 5. Radial distribution curves for vinyl dibromosilane (VDB). The experimental curve was calculated from a composite of the six average
Ϫ1
2
ꢀ Ϫ1
Յ
experimental curves shown in Fig. 3 multiplied by the factor ZBrZBrꢀABrABr
†
expꢀϪ0:002s †: Theoretical data for the region 0 Յ s=A
1:75 were added to the composite curve prior to the transformation. The difference curve is experimental minus theoretical. The vertical lines
indicate the interatomic distances, with lengths proportional to the distance weights.
annealing, the glassy solids still contained both
conformers. Obviously, a large number of funda-
mental frequencies for the syn and gauche conformers
coincide. The assignments that are included in the
Tables 1 and 2 are based on the results of the ab initio
calculations described later on, and they are given for
the syn conformer. Bands due to the gauche confor-
mers are indicated by a prime.
and Raman active. For the gauche conformers with C1
symmetry, all 21 fundamentals are infrared and
Raman active too. As no infrared band contours
have been measured, the present assignment in the
region of 3200–1000 cmϪ1 is based not only upon
the ab initio calculations, but also on previous vibra-
tional studies of vinylchlorosilane [3,13,14], vinylsi-
lane [15], and substituted vinylsilanes such as methyl
vinyl difluorosilane [16]. The numbering of the funda-
mentals is presented in Table 3. The assignments of
the three CH stretching modes and the SiH stretch are
The syn conformers of VDC and VDB possess Cs
symmetry, and the 21 fundamentals are divided into
14A0 and 7A00 normal modes, all of which are infrared

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
265
Table 1
Observed infrared (IR) and Raman (Ra) wavenumbers (cmϪ1) for VDC (s ˆ strong; b ˆ broad; sh ˆ shoulder; m ˆ medium; w ˆ weak;
v ˆ very)
IR(l)
Ra(l)
Ra(s)
ni
Assignment
3184 w
3071 m
3021 sh
2992 vs
3184 w
3073 m
3022 m
2993 vs
3071 ms
3023 m
2980 vs
2970 sh
n1
n2
CH2 antisym. stretch
CH stretch
2957 mw
2938 vvw
2911 vw
2876 vvw
2215 vs
1595 vs
1407 ms
1402 ms
1379w
1267 vs
1008 w
1003 w
992 vw
977 vw
2955 m
2938 w
2912 w
2879 w
2220 vs
1597 vs
1407 ms
1400 ms
2936mw
n3
CH2 sym. stretch
2871 m
2219 vs
1596 s
1403 s
n4
n5
n6
n 0
SiH stretch
CyC stretch
CH2 deformation
6
1380 m,sh
1268 m
1266 vs
1008 w
n7
CH in-plane bend
CH2 twist
CC torsion
CH2 wag
n15
n16
n8
1004 m
990 sh
973 m
840 sh
816 vs
801 vs
990 w
976 w
n 0
8
815 m
802 m
785 vs
727 m
693 m
575 mw
816 m
799 m
783 sh
727 m
692 m
573 m
527 s
n 0
9
n9
CSiH deformation
ClSiH deformation
SiC stretch
n17
n10
n 0
728 ms
694 m
575 s
10
n19
n 0
SiCl2 antisym. stretch
530 ms
18
520 vs
518 s
n11, n18
SiCl2 sym. stretch, CH out-of-
plane bend
475 mw
440 w
370 vw
476 vs
439 mw
371 mw
322 s
226 mw
220 sh
176 ms
157 sh
87 w
474 s
n 0
11
372 m
322 s
228 m
n 0
n12
12
CCSi bend
n 0
14
n14
n13, n20
SiCl2 wag
SiCl2 deformation, SiCl2 twist
176 s
158 sh
n 0
n21
20
SiC torsion
in agreement with the studies cited above [3,13–16]
and the results of the ab initio calculations. In the
present work, the Raman lines at 1402 cmϪ1 (VDC)
out for VDC and VDB, as 2n10 must lie above n6
for both compounds.
Due to the presence of six fundamentals (n8, n 0 ,
8
and 1397 cmϪ1 (VDB) are assigned as n 0 of the
n15, n 015, n16, and n 016) the region between 1020 and
950 cmϪ1 is somewhat complicated. The proposed
assignments for VDC and VDB are partially based
upon the assignments given by Durig et al. [3] for
vinylchlorosilane as they rely upon an infrared band
contour analysis, but also on the predictions made by
the ab initio calculations. In particular, the succession
6
gauche conformers, because their intensity ratios
ꢀn6=n06† are temperature dependent.
A line at
1394 cmϪ1 has also been observed for vinylchlorosi-
lane [3], but has been assigned as the overtone of the
SiC stretch in Fermi resonance with the CH2 deforma-
tion. An analogous assignment clearly can be ruled

266
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Table 2
Observed infrared and Raman wavenumbers (cmϪ1) for VDB (s ˆ strong; b ˆ broad; sh ˆ shoulder; m ˆ medium; w ˆ weak; v ˆ very)
IR(l)
Ra(l)
Ra(s)
ni
Assignment
3177 w
3068 m
3018 m
2978 m
2960 sh
3068 mw
3016 mw
2988 vs
2956 mw
2925 mw
3066 m
3014 w
2985 vs
2956 w
2921 w
2869 w
2206 vs
n1
n2
n3
n4
n5
CH2 antisym. stretch
CH stretch
CH2 sym. stretch
SiH stretch
2205 vs
1944 w,b
1592 s
2204 vs
1592 vs
1592 vs
1572 w
CyC stretch
1494 w
1404 vs
1395 sh
1405 m
1397 m,sh
1404 s
n6
n 0
CH2 deformation
1396 ms
1380 vw
1283 w
1261 s
1208 w
1029 w
1001 m
6
1283 vw
1262 ms
1208 vw
1029 w
1002 m
990 vvw
970 vvw
797 mw
785 mw
771 mw
722 w
684 mw
562 mw
536 w
519 mw
468 w
1263 w
n7
CH in-plane bend
n16
n15
n8
CC torsion
CH2 twist
CH2 wag
1004 ms
989 s
972 s
972 vw
797 mw
785 m
771 mw
723 w
683 mw
560 mw
537 w
519 mw
470 mw
n 0
n 0
8
795 vvs
9
n9
CSiH deformation
BrSiH deformation
SiC stretch
770 vvs
724 s
690 sh
n17
n10
n 0
10
535 vs
516 vs
469 vs
432 vs
n18
n 0
CH out-of-plane bend
18
n 0
19
n19
n11
SiBr2 antisym. stretch
SiBr2 sym. stretch
423 m
393 sh
370 s
360 sh
274 s
194 m
160 w,sh
130 sh
118 s
423 m
396 sh
370 vs
360 sh
275 vs
197 m
372 s
360 sh
n 0
n 0
11
12
n12
n 0
CCSi bend
14
n14, n20
SiBr2 wag, SiBr2 twist
SiBr2 deformation
140 sh
119 vs
n 0
n13
20
of the fundamentals has been chosen according to the
latter. The proposed assignments for these fundamen-
tals are by no means unambiguous, as the calculated
VDC are replaced by the heavier bromine atoms in
VDB. The assignments of the SiC stretching modes
n10 is straightforward for both compounds, with n 0
10
wavenumbers do not differ by more than 25 cmϪ1
.
possessing the smaller wavenumber for both VDC and
VDB. This is opposed to the succession reported for
the SiC stretching modes of the conformers of vinyl-
chlorosilane [3].
The SiH bending modes n9 and n17 appear between
820 and 770 cmϪ1; they are easily assigned with the
help of the ab initio calculations. Both modes shift to
slightly lower wavenumbers as the chlorine atoms of
The assignments for the SiCl stretching modes n11

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
267
Table 3
Symmetry coordinates (not normalized) used in the normal coordinate analysis for syn- and gauche-VDC and -VDB
Species
Cs (syn)
A0
Approximate description
Symmetry coordinate
C1 (gauche)
A
CH2 antisymmetric stretch
CH stretch
CH2 symmetric stretch
SiH stretch
S1 ˆ t1 Ϫ t2
S2 ˆ t3
S3 ˆ t1 Ϫ t2
S4 ˆ u1
C ˆ C stretch
S5 ˆ R
CH2 deformation
CH in-plane bend
CH2 wag
S6 ˆ 2d Ϫ g1 Ϫ g2
S7 ˆ g4 Ϫ g3
S8 ˆ g1 Ϫ g2
CSiH deformation
SiC stretch
SiX2 symmetric stretch
CCSi bend
S9 ˆ 4b Ϫ w2 Ϫ w3 Ϫ c2 Ϫ c3
S10 ˆ s
S11 ˆ u2 ϩ u3
S12 ˆ 2e Ϫ g3 Ϫ g4
S13 ˆ 5a Ϫ b Ϫ w2 Ϫ w3 Ϫ
c2 Ϫ c3
SiX2 deformation
SiX2 wag
S14 ˆ w2 ϩ w3 Ϫ c2 Ϫ c3
A00
CH2 twist
CC torsion
ClSiH deformation
CH out-of-plane bend
SiX2 antisymmetric stretch
SiX2 twist
S15 ˆ p1
S16 ˆ u
S17 ˆ w2 Ϫ w3 ϩ c2 Ϫ c3
S18 ˆ p2
S19 ˆ u2 Ϫ u3
S20 ˆ w2 Ϫ w3 Ϫ c2 ϩ c3
S21 ˆ t
SiC torsion
and n19 and the CH out-of-plane bend n18 of VDC are
based on the ab initio calculation, which predicts iden-
tical wavenumbers for n11 and n18 of the syn
conformer (498–499 cmϪ1, scaled). For the gauche
conformer, n18 is predicted to shift to higher wave-
numbers by approximately 24 cmϪ1. The infrared
band at 530 cmϪ1 therefore is assigned as n 018. For
shift to lower wavenumbers for VDB, which is easily
understood taking into account the larger masses of
the Br-atoms. Strong vibrational couplings between
the low frequency deformational modes occur for
both syn-VDC and syn-VDB, as given by the potential
energy distributions (PEDs) in Tables 4 and 5. For the
gauche conformers, further kinetic couplings between
these two groups of modes occur as they are now
allowed by symmetry.
VDB, both n18 and n 0 (536 and 519 cmϪ1, Raman)
18
have been observed, as they are clearly separated from
the SiBr stretching modes. For gauche vinylchlorosi-
lane, the out-of-plane CH-bend has been observed at
470 cmϪ1, and it is strongly coupled with the SiCl
stretch [3], very similar to the situation with VDC
which also exhibit considerable coupling between
the CH out-of-plane bend and the antisymmetric
SiCl2-stretch (cf. Section 3.2 on NCA and Table 4).
3.1.2. Rotational isomerism
A close inspection of the Raman spectra of VDC
and VDB reveals that intensity variations with
temperature occur due to restricted rotation around
the SiC bonds. For VDC, the line pair observed at
322/371 (n12/n 012) and 727/693 cmϪ1 (n10/n 010) turned
out to be most suitable for a determination of the
enthalpy difference DH ˆ HS Ϫ HG: Two line pairs
The assignments of the CCSi bends n12 and n 0 for
12
VDC (322 cmϪ1 for the syn conformer, 371 cmϪ1 for
the gauche) are based upon the ab initio results; they
are in agreement with the assignments proposed for
vinylchlorosilane (289 and 374 cmϪ1 for the syn and
gauche conformer, respectively [3]). n12 and n 012 both
at 274/360 (n12/n 012) and 1405/1397 (n6/n 0 ) (cmϪ1
)
6
have been chosen for VDB. Fig. 6 presents the van’t
Hoff plots for all four-line pairs, that have been

268
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Table 4
Observed and calculated wavenumbers (cmϪ1) and potential energy distributions ( Ͼ 10%) for syn- and gauche-VDC
Vib. no Description
syn
gauche
Theorya
Exp.
PED
Theorya
Exp.
PED
A0
n1
n2
n3
n4
n5
n6
n7
n8
n9
CH2 antisym. stretch
CH stretch
CH2 sym. stretch
SiH stretch
CyC stretch
CH2 deformation
CH in-plane bend
CH2 wag
3044
2985
2972
2218
1597
1403
1261
1004
809
3071 98(1)
3056
2984
2961
2215
1598
1397
1257
994
3071 99(1)
2992 91(2)
2938 90(3)
2215 100(4)
1595 72(5), 32(6)
1402 69(6), 19(5)
1267 55(7), 21(8)
977 69(8), 33(7)
815 51(9), 17(14), 22(17),
12(20)
2992 80(2), 18(3)
2938 82(3), 18(2)
2215 100(4)
1595 71(5), 34(6)
1407 67(6), 20(5)
1267 53(7), 22(8)
992 66(8), 32(7)
802 76(9), 26(14)
CSiH deformation
813
n10
n11
n12
SiC stretch
SiCl2 sym. stretch
CCSi bend
727
503
317
727 80(10)
692
471
365
693 74(10)
476 74(11), 12(18)
371 58(12), 12(14)
520 81(11), 11(12)
322 37(12), 11(10), 15(11),
17(14)
n13
n14
SiCl2 deformation
SiCl2 wag
172
186
176 74(13), 19(9), 18(14)
220 34(14), 49(12), 21(13),
12(9)
165
218
176 79(13), 19(9), 12(14)
226 37(14), 12(9), 19(13),
A00 n15
n16
n17
CH2 twist
CC torsion
ClSiH deformation
CH out-of-plane bend
1021
1015
810
1008 89(15)
1008
1003
794
1008 101(15)
1003 53(16), 13(15), 20(18)
785 72(17), 35(20)
520 33(18), 20(16), 49(19)
1003 59(16), 25(18)
785 51(17), 28(20)
527 38(18), 14(12), 16(14),
14(19), 24(16)
n18
502
528
n19
n20
SiCl2 antisym. stretch
SiCl2 twist
571
169
575 54(19), 17(16), 26(18),
176 66(20), 30(17), 15(19),
13(21)
578
143
575 82(19)
157 42(20), 22(12), 28(17)
n21
SiC torsion
79
87 92(21)
79
87 104(21)
a
Wavenumbers calculated from the scaled (refined) HF/6-31G(d) force fields as described in Section 3.2.
obtained by application of the equation
accuracy of the DH-values reported. Our final esti-
ꢀ
ꢁ
ꢀ
ꢁ
mates are therefore ϩ0:11 ^ 0:06 and ϩ0:23 ^
0:07 kcal molϪ1 for VDC and VDB, respectively,
with gauche lower in energy. These values have
been used to determine energy constraints in the
GED least-squares refinements, to reduce the uncer-
tainty and correlation in obtained parameter values.
IS
DH
RT
DS
aS
ln
ˆ Ϫ
ϩ
ϩ ln
ꢀ1†
IG
R
aG
assuming that DH, DS and the ratio of the scattering
coefficients aS=aG is constant with temperature. DH-
values are 0:49 ^ 0:03 and 0:42 ^ 0:05 kJ molϪ1
for VDC, 0:93 ^ 0:07 and 0:97 ^ 0:07 kJ molϪ1
for VDB, giving averaged values of ϩ0:45 ^
0:05 and ϩ0:95 ^ 0:07 kJ molϪ1 (ϩ0:11 ^ 0:01
and ϩ0:23 ^ 0:02 kcal molϪ1), respectively. Apart
from the random errors given above, systematic
errors caused by impure bands (i.e. bands that do
not originate from a single conformer) cannot be
3.2. Ab initio mo calculations and normal coordinate
analysis (NCA)
For the purpose of NCA, ab initio molecular orbital
(MO) calculations (RHF-SCF) of equilibrium geome-
tries were carried out for minimum energy confor-
mers—syn- and gauche-VDC and -VDB—as well
as five additional pseudo-conformers, using the 6-
31G(d) (VDC) and 6-311G(d) (VDB) basis sets in
excluded.
A
systematic error of at least
^0:2 kJ molϪ1 ꢀ^0:05 kcal molϪ1†; maybe even
larger, has to be added, reducing significantly the

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
269
Table 5
Observed and calculated wavenumbers (cmϪ1) and potential energy distributions ( Ͼ 10%) for syn- and gauche-VDB
Vib. no Description
syn
gauche
Theorya Exp.
PED
Theorya Exp.
PED
A0
n1
CH2 antisym. stretch
CH stretch
3037
2982
2963
2205
1595
1396
1259
1003
785
722
416
265
114
3068
2988
2925
2204
1592
1405
1262
990
785
722
423
274
118
160
1002
1029
771
536
432
160
–
98(1)
88(2)
91(3)
100(4)
3050
2977
2955
2204
1596
1389
1253
991
796
683
371
344
112
185
1003
1029
776
516
472
125
74
3068
2988
2925
2204
1592
1397
1262
970
797
684
370
360
118
194
1002
1029
771
519
468
130
–
99(1)
n2
88(2), 11(3)
88(3), 11(2)
100(4)
71(5), 33(6)
68(6), 19(5)
55(7), 21(8)
70(8), 32(7)
61(9), 18(14), 14(17)
79(10)
83(11), 11(13)
53(12), 29(19)
91(13)
n3
CH2 sym. stretch
SiH stretch
C ˆ C stretch
CH2 deformation
CH in-plane bend
CH2 wag
CSiH deformation
SiC stretch
SiBr2 sym. stretch
CCSi bend
n4
n5
70(5), 34(6)
66(6), 20(5)
53(7), 21(8)
67(8), 32(7)
79(9), 23(14)
83(10)
53(11), 27(12), 12(14)
34(12), 42(11)
92(13)
57(14), 22(9), 39(12)
63(15), 23(16)
37(16), 39(15), 23(18)
71(17), 35(20)
53(18), 34(16)
94(19)
n6
n7
n8
n9
n10
n11
n12
n13
n14
SiBr2 deformation
SiBr2 wag
CH2 twist
165
48(14), 18(9)
100(15)
A00 n15
n16
n17
1005
1009
782
539
415
CC torsion
60(16), 24(18)
61(17), 14(9), 34(20)
55(18), 13(14), 35(16)
72(19), 20(12)
41(20), 14(12), 25(17)
100(21), 11(18)
BrSiH deformation
CH out-of-plane bend
SiBr2 antisym. stretch
SiBr2 twist
n18
n19
n20
n21
150
71
65(20), 26(17), 11(18), 20(21)
84(21)
SiC torsion
a
Wavenumbers calculated from the scaled (refined) HF/6-311G(d) force fields as described in Section 3.2.
the program Gaussian 94 [17]. Harmonic wavenum-
bers and force fields were calculated for the syn and
gauche geometries at the HF level of theory. The ab
initio Cartesian force fields were converted into
symmetry-defined force fields using 21 non-redundant
symmetry coordinates in the normal coordinate
program asym40 [18], and were further used in calcu-
lating shrinkage corrections for the respective confor-
mers of VDC and VDB utilized in the GED least-
squares refinements. The symmetry coordinates are
linear combinations of 32 redundant internal coordi-
nates defined in Fig. 7, and they are summarized in
Table 3 together with a description of the normal
modes. Identical symmetry coordinates were used
for all conformers to facilitate the comparison of
force constants. The ab initio symmetry force fields
calculated at the HF levels were scaled according to
the types of internal coordinates using the scheme
Fijꢀscaled† ˆ Fijꢀab initio† × ꢀXiXj†1=2; where Xi and
Xj are the scale constants for the diagonal force
constants. The scale constants were obtained by a
refinement procedure, where the calculated normal
mode wavenumbers from the ab initio force fields
were fitted to the experimental (Raman) wavenumbers
within an average ^10 cmϪ1. For VDC, the scale
constants varied from 0.76 to 0.91, and for VDB the
scale constants varied from 0.78 to 0.92
(1s ˆ 0:01–0:06 for both molecules). The observed
and calculated wavenumbers, as well as the PEDs for
both conformers of VDC and VDB, are summarized
in Tables 4 and 5.
Results from the ab initio MO calculations were
used to establish trial values for bond distances and
bond angles in the GED least-squares refinements. In
addition, the structural information was used to estab-
lish constraints, by incorporating geometrical differ-
ences between all pseudoconformers comprised by
the dynamic models. Calculations of VDC were
performed at the HF level using the basis sets 6-
31G(d) and 6-311G(d), obtaining both a torsional
potential curve and geometry constraints utilized in
the GED refinements. For VDB, torsional potential
curves and geometry constraints were obtained
using the methods HF/3-21Gء and HF/6-311G(d). A

270
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Fig. 6. Van’t Hoff plots for the line pairs: (a) 322/371 (n12/n 012) and (b) 727/693 (n10/n 010) cmϪ1 of VDC (top), and (c) 274/360 (n12/n 012) and (d)
1405/1397 (n6/n 06) cmϪ1 of VDB (bottom).
Fig. 7. Internal coordinates of vinyl dichlorosilane (VDC) and vinyl dibromo-silane (VDB) used in the construction of symmetry coordinates
(cf. Table 3).

Table 6
Structural parameter results from GED/Raman refinements and ab initio MO calculations for the syn conformer of vinyl dichlorosilane ꢀX ˆ Cl† and vinyl dibromosilane (X ˆ Br)
˚
˚
(Distances (rg) are in angstrøms (A), angles (Єa ) in degrees. S and G refer to syn and gauche conformers, respectively. Parenthesized values are 2s. Quantities in square brackets
were held at ab initio values)
VDC
VDB
GED (rg/Єa )
Ab initioa (r/Є; HF)
Ab initiob (r/Є; MP2)
GED (rg/Єa )
Ab initioa (r/Є; HF)
Ab initiob (r/Є; MP2)
r(CyC)
r(C–Si)
r(Si–X)
r(Si–H)
r(͗C–H͘)
ЄCCSi
Є͗CCH͘
ЄXSiX
ЄCSiX
1.357(7)
1.847(5)
2.042(2)
[1.499]
[1.099]
[121.8]
[121.1]
108.2(6)
110.3(6)
[111.7]
Ϫ0.15 ^ 0.08d
[0.0]
1.324
1.845
2.057
1.460
1.078
121.8
121.1
107.5
110.6
111.7
0.55e
Ϫ0.10e
1.70e
47
1.346
1.840
2.048
1.470
1.088
120.8
121.0
108.2
110.3
111.1
–
1.366(10)
1.827(9)
2.206(2)
[1.501]
[1.100]
[121.7]
1.324
1.848
2.230
1.461
1.078
121.7
121.2
108.6
110.4
111.4
0.45
0.10
1.65
52
Ϫ0.46
–
1.347
1.842
2.222
1.472
1.088
120.6
121.0
109.2
110.0
110.7
–
[121.2]
109.0(2)
110.1(8)
[111.4]
ЄCSiH4
V1 (kcal molϪ1
V2 (kcal molϪ1
V3 (kcal molϪ1
aS ( × 100)f
)
)
)
Ϫ0.31 ^ 0.09d
[0.0]
c
c
c
–
–
–
–
1.2 ^ 0.8
32
[ϩ0.11]
0.136
1.4 ^ 1.0
27
[ϩ0.23]
0.133
38
Ϫ0.11
–
40
Ϫ0.18
–
DEЊSϪG (kcal molϪ1
)
Ϫ0.33
–
Rg
Dependent parameters:
ЄCCH7
ЄCCH8
ЄCCH9
ЄSiCH7
[118.9]
[122.5]
[121.9]
[119.3]
108.1(7)
[115.6]
[120.0]
118.9
122.5
121.9
119.3
108.1
115.6
116.4
119.0
121.9
121.9
120.1
108.4
116.2
114.8
[119.1]
[122.5]
[121.9]
[119.2]
108.1(9)
[115.6]
[120.0]
119.1
122.5
121.9
119.2
108.0
115.6
117.3
119.3
122.0
121.8
120.1
108.4
116.2
115.9
ЄXSiH
ЄHC1H
f(gauche)h
a
HF/6-311G(d) calculations, except where noted.
MP2/6-311G(d) calculations.
The refined potential constants from the dynamic GED model described in the text, and estimated theoretical values from simple curve fitting.
Value and estimated error limit based on the liquid Raman energy difference.
Based on the HF/6-31G(d) potential curve calculations.
Percent mole fraction of the syn conformer at 296 K (VDC) and 298 K (VDB). The GED values are estimated from the dynamic model (sum of mole fractions of
b
c
d
e
f
pseudoconformers 0, 30 and 60).
P
P
The ‘quality of fit’ factor R ˆ ‰ wiDi2= wiꢀsiIm ꢀs†† Š and Di ˆ siIm ꢀs† Ϫ siImcalcꢀs†.
g
exp
2
1=2
exp
h
The gauche torsional angle for f ˆ fꢀCCSiH†.

272
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
general basis set was utilized for VDB in an HF/6-
31G(d)/GEN calculation with a bromine contracted
basis set ([20s14p9d]/[13s8p6d]) [19], giving results
similar to the earlier HF calculations. Additionally,
MP2(fc) and DFT/B3LYP calculations with the 6-
311G(d) basis set were performed for both molecules
as a final check of the consistency in the obtained
geometries and conformational energy differences.
The procedure of using ab initio results to provide
model constraints in the GED analysis is now well
established [20–22]. Zero-point energies (ZPE) were
also estimated with the HF ab initio calculations.
These were scaled by the suggested factor of 0.893
[23] to compensate for the neglect of electron correla-
tion at this level of theory. This resulted in ZPE differ-
ence values of 0.0342 and 0.0230 kcal molϪ1 for VDC
and VDB, respectively, in favour of the syn
conformer.
anti (180Њ), comprising a total of seven pseudo-
conformers, a refined value for V3 was obtained.
3.3.2. Refinements and vibrational corrections
Refinement of the molecular structures based on the
GED data were made utilizing the least-squares
method [25]. In this procedure, a theoretical sImꢀs†
curve was simultaneously adjusted to six average
experimental intensity curves, one from each plate.
A unit weight matrix was used. The structures were
converted from the geometrically consistent ra to the
ra-type required by the formula for the scattered in-
tensities ꢀraꢀT† ˆ rgꢀT† Ϫ `2ꢀT†=r ˆ raꢀT† Ϫ `2=r ϩ
KT ϩ drꢀT†† [26,27], using values of the centrifugal
distortion constants (dr(T)), perpendicular amplitude
corrections (KT) and root-mean-square amplitudes of
vibration (`(T)) calculated for all pseudoconformers
at about room temperature (296–298 K). This was
performed using the asym40 NCA program as
described in Section 3.2, with the additional effort in
removing the contribution from the large-amplitude
torsional symmetry coordinate to the harmonic NCA
calculations.
3.3. Gas electron diffraction
3.3.1. Details of the dynamic model
The GED data for VDC and VDB were analysed by
fitting the cosine Fourier series
X
3.3.3. The molecular parameters
1
Vꢀf† ˆ
Vi‰1 Ϫ cos iꢀf†Š
ꢀ2†
The geometry of the two molecules was described
by a set of independent parameters which were to be
fitted to the experimental data. In our refinements
these parameters were chosen as (X ˆ Cl or Br):
r(CyC), r(C–Si), r(͗C–H͘) (average value), r(Si–
X), r(Si–H), Є(CCSi), Є(͗CCH͘) (average value),
P(XSiX) (projection angle; the angle between the
two Si–X bonds projected on a plane perpendicular
to the C–Si bond), Є(CSiX) and Є(CSiH), along with
the potential constants V1, V2 and V3. Ab initio
constraints were incorporated for both molecules as
described earlier. The r(Si–H) parameter value could
not be refined in either molecule, and was held
constant in the least squares refinements. The ra
values used for this parameter were calculated from
the assumed re ab initio values (HF/6-311G(d)
2
i
with i ˆ 1–3; to the experimental data, where f is the
value of the torsional angle fꢀCyCSiH†: The cosine
potentials [24] were obtained by using pseudo-confor-
mers, each one representative of a finite displacement
of the large-amplitude torsional coordinate, and each
given a weight determined by Boltzmann statistics in
which the cosine potential V(f) appeared as the
energy factor. The potential function V(f) has the
property Vꢀ0Њ† ˆ 0; Vꢀ120Њ† ˆ ꢀ3=4†ꢀV1 ϩ V2† and
Vꢀ180Њ† ˆ V1 ϩ V3 (in units of kcal molϪ1). Only
one potential parameter could be refined indepen-
dently, and the following procedures were under-
taken: The potential constant V2 was set to zero,
while constraints were introduced through the V1 para-
meter by using the variable-temperature Raman
energy results, which found gauche lower in energy
in both molecules. Therefore, V2 ˆ 0 implied that
Vꢀ120Њ† ˆ ꢀ3=4†V1; giving values of V1 ˆ
Ϫ0:15 ^ 0:08 kcal molϪ1 for VDC and V1 ˆ
Ϫ0:31 ^ 0:09 kcal molϪ1 for VDB. By using
pseudo-conformers at 30Њ intervals from syn (0Њ) to
method) by using the anharmonicity constant a3 ˆ
ꢀ Ϫ1
2:0 A
in the usual formula ra ˆ re ϩ 3a3`2=2:
The rg values then became rg ˆ re ϩ 3a3`2=2 ϩ K;
where K is the perpendicular amplitude correction.
The vibrational properties of the molecules were
specified by 36 amplitude parameters for each
pseudoconformer, corresponding to the number of

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
273
Fig. 8. Torsional potential curves for VDC as obtained from ab initio Hartree–Fock energy values using the 6-31G(d) basis set (top curve), and
as obtained from the GED data using a Raman-constrained dynamic model (bottom curve), showing the potential energy V(f) (kcal molϪ1) as a
function of the torsional angle fꢀCyC–Si–H†:
interatomic distances, making a total of 252 weighted
distances included in the dynamic model. Many, but
not all, of the amplitudes could successfully be
refined. Some of the amplitudes were refined together
as groups. The amplitudes which could not be refined
were held constant at the calculated values obtained
from asym40. In the final refinement on both VDC
and VDB, five geometrical parameters, four ampli-
tude parameters and the potential constant V3 were
refined simultaneously. The results of the refinements
are given in Table 6 along with the corresponding
geometrical values from the ab initio MO calculations
(HF/6-311G(d) and MP2/6-311G(d)). The full range
of ab initio and DFT calculations and the correlation
matrices have been deposited as supplementary data.
The correlation matrix for the refined parameters of
VDC showed the V3 parameter to be quite uncorre-
lated with the other parameters chosen for refinement.
The correlation matrix for VDB showed similar char-
acteristics. In Figs. 8 and 9, the energy values obtained
from the constrained dynamic GED models and the ab
initio MO calculations have been plotted as functions
of the torsional angle fꢀCyC–Si–H† for VDC and
VDB, respectively. Selected interatomic distances
for VDC and VDB, including refined and calculated
vibrational amplitudes (`), are given in Tables 7 and
8, respectively.
4. Discussion
The collection of vinylsilanes studied experimen-
tally by GED or spectroscopic methods is starting to
grow [1–7,28], and a list of those studied by GED has
been assembled in Table 9. Ab initio calculations of
conformer stability can often be inconclusive, as the
sign of the DEЊ value may often be heavily dependent
on the level of theory applied. The sign change should
imply that the energy difference between the confor-
mers is fairly small, rarely exceeding 0.5 kcal molϪ1

274
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Fig. 9. Torsional potential curves for VDB as obtained from ab initio Hartree–Fock energy values using the 6-311G(d) basis set (top curve), and
as obtained from the GED data using a Raman-constrained dynamic model (bottom curve), showing the potential energy V(f) (kcal molϪ1) as a
function of the torsional angle fꢀCyC–Si–H†:
[1–7]. In the present study, we have therefore chosen
to use the experimental Raman energy difference
values as constraints in the GED investigations,
hopefully improving the reliability of the results
obtained from the least-squares refinements. Using
only GED data and refining several potential para-
meters simultaneously gave results with uncertainties
too large for the values to be useful. The values
obtained for the refined V3 parameter were: V3 ˆ
1:0 kcal molϪ1 for VDB, along with the constrained
parameter values V1 ˆ Ϫ0:15 ^ 0:08 kcal molϪ1
(VDC) and V1 ˆ Ϫ0:31 ^ 0:09 kcal molϪ1 (VDB).
The error limits on the V1 parameters were simply
based on the estimated error limits in the Raman
energy differences. Unfortunately, the combined use
of liquid- and gas-phase data introduces an additional
source of uncertainty, realistically making the error
limits even larger than quouted here. In fact, the rela-
tively large 2s standard error for the refined V3 poten-
tial parameter in both molecules showed that the
potential energy values could not be determined
with any high degree of certainty from the experi-
mental data. The distance distribution does not vary
much during the conformational change in VDC and
VDB, and this is probably the main reason for the
difficulties in obtaining reliable energy information.
Presumably, a more reliable method of obtaining
conformer energy differences is performing variable
temperature studies of the samples in question.
Although liquid state studies using Raman spectro-
scopy may not be considered directly compatible
with gas-phase data, it should prove compatible
enough to give directional clues in most cases. The
Raman data gave lower energy for the gauche
1:2 ^ 0:8 kcal molϪ1
for
VDC;
V3 ˆ 1:4 ^
conformer for both VDC (0:11 ^ 0:06 kcal molϪ1
)
and VDB (0:23 ^ 0:07 kcal molϪ1†; with relatively
smaller uncertainties, by using plots of intensity ratios
and the relations given in Eq. (1). The Raman results
are average values obtained from two separate line
doublets for each molecule (322/371 and 727/
693 cmϪ1 for VDC; 274/360 and 1405/1397 cmϪ1

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
275
Table 7
˚
˚
Selected interatomic distances for VDC. (Distances (ra , rg) and amplitudes (`ij) are in angstrøms (A). Parenthesized values are 2s. Quantities in
square brackets were held at values derived from ab initio calculations (HF/6-311G(d)). ‘Quality of fit’ factor for final refinement was
R ˆ 0:136)
GED
Ab initio values
ra
rg
`
calc
`
exp
MP2/6-311G(d) distances
syn:
r(CyC)
r(C–Si)
r(Si–H)
r(Si–Cl5,6
r(C1···Si)
1.348(7)
1.843(5)
[1.483]
1.357
0.043
0.050
0.088
0.049
0.079
0.100
0.128
0.100
0.166
0.102
0.095
0.196
0.115
0.140
0.179
0.176
0.213
0.165
0.054(7)
0.050(5)
1.346
1.840
1.470
2.048
2.781
2.103
2.738
3.193
2.939
3.803
3.318
3.067
4.215
3.722
3.514
4.453
4.150
5.106
1.847
[1.499]
2.042
2.805
2.104
2.771
3.190
2.982
3.825
3.303
3.121
4.222
3.742
3.498
4.479
4.204
5.108
)
2.038(2)
2.799(7)
2.095(7)
2.760(5)
3.187(11)
2.971(8)
3.808(7)
3.300(12)
3.111(6)
4.220(11)
3.728(5)
3.492(17)
4.473(9)
4.185(7)
5.099(13)
0.057(3)
r(C1···H7)
r(C2···H4)
r(C2···Cl5)
r(Si···H8)
r(Si···H9)
r(Cl5···Cl6)
r(C1···H4)
r(C1···Cl5)
r(H4···H7)
r(Cl5···H7)
r(Cl5···H8)
r(H4···H9)
r(Cl5···H9)
0.101(13)a
gauche:
r(C1···Si)
r(C2···Cl5)
r(C2···Cl6)
r(Cl5···Cl6)
r(C1···H4)
r(C1···Cl5)
r(C1···Cl6)
r(Cl5···H7)
r(Cl5···H8)
r(Cl6···H8)
r(H4···H9)
r(Cl5···H9)
r(Cl6···H9)
2.851(7)
2.858
3.194
3.173
3.313
3.876
3.446
4.235
4.186
2.962
4.539
4.786
4.519
5.092
0.071
0.090
0.100
0.098
0.139
0.151
0.113
0.112
0.232
0.166
0.176
0.175
0.166
2.830
3.190
3.173
3.345
3.810
3.405
4.268
4.198
2.880
4.581
4.737
4.475
5.126
3.191(11)
3.170(11)
3.310(12)
3.868(8)
3.440(17)
4.232(16)
4.181(10)
2.950(19)
4.531(16)
4.772(8)
0.092(13)a
0.102(13)a
0.153(14)a
4.506(17)
5.084(17)
a
These amplitudes were refined as a group.
for VDB; cf. Fig. 6). It could be argued that it is not
unlikely that the energy difference between the
conformers is different in the liquid phase, compared
to the gas phase, as implied by the ab initio MO
calculations. We tried testing this by using the ab
initio HF/6-311G(d) energy differences for both
VDC and VDB (with syn lower in energy by 0.33
and 0.46 kcal molϪ1, respectively), as V1 constraints
in the least-squares refinements similar to the proce-
dure described previously for the Raman energies.
The resulting ‘quality-of-fit’ R-factors had almost
identical values to the ones obtained using the
Raman energy constraints. This indicated that the
GED data alone just cannot determine a reliable
value for the conformational energy difference in
either molecule. Adding to this, the MP2/6-311G(d)
ZPE-corrected conformational energy differences
were 0.11 and 0.18 kcal molϪ1 for VDC and VDB,
respectively, with syn still the lower energy form.
These values, however, were much closer to the

276
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
Table 8
˚
˚
Selected interatomic distances for VDB. (Distances (ra , rg) and amplitudes (`ij) are in angstrøms (A). Parenthesized values are 2s. Quantities in
square brackets were held at values derived from ab initio calculations (HF/6-311G(d)). ‘Quality of fit’ factor for final refinement was
R ˆ 0:133)
GED
Ab initio values
ra
rg
`
calc
`
exp
MP2/6-311G(d) distances
syn:
r(CyC)
r(C–Si)
r(Si–H)
r(Si–Br5,6
r(C1···Si)
1.356(10)
1.822(9)
[1.485]
1.366
0.043
0.050
0.089
0.052
0.079
0.100
0.129
0.105
0.166
0.102
0.103
0.197
0.121
0.140
0.185
0.179
0.213
0.175
0.053(9)
0.061(7)
1.347
1.842
1.472
2.222
2.780
2.106
2.735
3.337
2.935
3.802
3.622
3.055
4.349
3.720
3.634
4.587
4.138
5.230
1.827
[1.501]
2.206
2.793
2.114
2.749
3.307
2.973
3.813
3.588
3.096
4.340
3.724
3.592
4.600
4.181
5.214
)
2.201(2)
2.784(11)
2.103(9)
2.738(8)
3.304(14)
2.961(11)
3.792(11)
3.585(6)
3.083(9)
4.336(13)
3.708(8)
3.585(23)
4.594(10)
4.158(10)
5.202(15)
0.052(3)
r(C1···H7)
r(C2···H4)
r(C2···Br5)
r(Si···H8)
r(Si···H9)
r(Br5···Br6)
r(C1···H4)
r(C1···Br5)
r(H4···H7)
r(Br5···H7)
r(Br5···H8)
r(H4···H9)
r(Br5···H9)
0.109(12)a
gauche:
r(C1···Si)
r(C2···Br5)
r(C2···Br6)
r(Br5···Br6)
r(C1···H4)
r(C1···Br5)
r(C1···Br6)
r(Br5···H7)
r(Br5···H8)
r(Br6···H8)
r(H4···H9)
r(Br5···H9)
r(Br6···H9)
2.825(11)
3.309(14)
3.298(14)
3.594(7)
3.847(11)
3.486(22)
4.357(20)
4.314(12)
2.930(27)
4.646(20)
4.757(11)
4.542(24)
5.203(22)
2.834
3.313
3.301
3.597
3.858
3.492
4.361
4.320
2.942
4.655
4.774
4.556
5.213
0.071
0.092
0.104
0.105
0.139
0.153
0.119
0.114
0.233
0.168
0.176
0.178
0.174
2.839
3.349
3.311
3.646
3.817
3.532
4.406
4.361
2.969
4.729
4.738
4.598
5.246
0.096(12)a
0.108(12)a
0.156(13)a
a
These amplitudes were refined as a group.
experimental Raman energy differences than the HF
calculated ones, lowering the energy differences by
0.2–0.3 kcal molϪ1 in favour of the gauche conformer
compared to the HF results. The corresponding energy
difference values from the density-functional based
B3LYP/6-311G(d) calculations were 0.29 and
0.37 kcal molϪ1 for VDC and VDB, respectively,
with syn also lower in energy.
same range as those found experimentally for vinyl-
chlorosilane by using torsional transitions [3]. The
barrier values were found within 1.2–2.1 kcal molϪ1
for all three molecules, including also the GED
values. The geometrical parameter values obtained
from the GED/Raman refinements are shown in
Table 6, along with the values from the HF- and
MP2/6-311G(d) calculations for both VDC and
VDB. The experimental geometrical values agree
fairly well with the theoretical ones, especially
Ab initio MO calculations of the torsional potential
barriers for VDC and VDB resulted in values in the

Table 9
Comparison of results for several vinyl- and methyl vinylsilanes obtained by GED. Bond distances (rg) in angstrøm (A), valence and torsional angles in degrees (Њ)
˚
˚
Molecule
Conformer
r(CyC)
r(C2–Si)
r(Si–X)
ЄCyCSi
127.8(12)
125.5 (13) (s)
124.3 (13) (g)
123.3(17)
125.7(17)
123.3(8)
ЄCSiC
f (Torsional angle)
Reference
CH2yCHSiClꢀCH3†
syn
Both
1.347(5)
1.341(7)
1.838(3)
1.840(3)
2.078(2)
2.052(2)
110.0a
115.6(13)
17(6)
[5]
[7]
2
CH2yCHSiCl2CH3 ꢀVCS†
CH2yCHSiCl2CH3 ꢀVCS†
CH2yCHSiF2CH3 ꢀVFS†
0
103.8b
syn
gauche
syn
gauche
syn
syn
anti
syn
gauche
syn
gauche
1.341(6)
1.342(6)
1.344(5)
1.347(5)
1.359(10)
1.35a
1.349(12)
1.357(7)
1.356(7)
1.366(10)
1.364(10)
1.843(4)
1.844(4)
1.846(3)
1.845(3)
1.867(3)
1.81(2)
1.863(13)
1.847(5)
1.851(5)
1.827(9)
1.831(9)
2.051(2)
2.049(2)c
1.592(2)
1.591(2)c
111.6(17)
110.7(17)
113.4(11)
113.2(11)
112.6(8)
0
[1]
[1]
121.2b
0
122.3(8)
124.6(18)
121(3)
117.4b
0
CH2yCHSiꢀCH3†
[4]
[28]
[2]
3
CH2yCHSiCl3
2.060(5)
2.020(3)
2.042(2)
2.040(2)c
2.206(2)
2.206(2)c
Cl2CyCClSiCl3 ꢀPCV†
124.0(12)
[180]d
0
H2CyCHSiHCl2
121.8a
Present work
125.7a (constr.)
121.7a
[120.0]d
0
H2CyCHSiHBr2
Present work
124.7a (constr.)
[120.0]d
a
Assumed, theoretical or constrained value.
b
c
d
The reported uncertainties for these torsional angles are: For VCS from Ref. [7]: ^5.4Њ (3s). For VCS [1]: ^15.7Њ (2s). For VFS [1]: ^14.3Њ (2s).
These are the Si–Xin plane bond distances (X ˆ Cl, Br) for the gauche conformer.
Studied by using a dynamic model.

278
T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
˚
regarding the valence angle parameters, but this is to
some extent due to the applied constraints as the
central CyCSi bond angle was set constant to the ab
initio value in both cases. This bond angle could not
be determined accurately from the GED data alone.
As is typical [1,2] with HF calculations, the carbon–
carbon double bonds were calculated too short (ca.
disilane, Br2HSi–SiHBr2 (2.206(5) A) [29], and
1,1,1,3,3,3-hexabromotrisilane, Br3Si–SiH2–SiBr3
(2.205(4) A) [30].
˚
It should be noted that the reliability of the scaled
HF force fields is probably quite good; the agreement
between the refined and calculated vibrational ampli-
tudes was found acceptable (cf. Tables 7 and 8). Ab
initio calculations further revealed that corresponding
parameter values in the syn and gauche conformers
did not change by any substantial amount. The most
important structural change was found in the central
CyCSi bond angle, which opens up 3–4Њ in the
gauche conformer for both molecules, according to
the ab initio MO and DFT calculations. This is prob-
ably due to larger steric interactions between the vinyl
group and the halogen atoms, compared to the inter-
actions between the vinyl group and the hydrogen
atom on Si.
In conclusion, the present study, combining GED,
spectroscopy and theoretical calculations, has shown
that both conformers (syn and gauche) of VDC and
VDB exhibit eclipsing of the CyC double bond. The
conformer where the hydrogen atom lies in the plane
of the vinyl group (syn) was found lower in energy by
all the theoretical calculations (by 0.11–
0.43 kcal molϪ1 for VDC; 0.04–0.46 kcal molϪ1 for
VDB). However, in the liquid phase slight energy
preferences were found in favour of the gauche
forms. These experimental results were used as
constraints in our final GED refinements. Gas compo-
sitions consisting of approximately 68% (VDC) and
73% (VDB) of the gauche conformer were obtained
as a result from the cosine potential models used,
while the corresponding MP2 compositions were
62% (VDC) and 60% (VDB).
˚
1.324 A), even compared to the ra values. The
MP2-calculated CyC bond distances became 1.346
˚
and 1.347 A in syn-VDC and -VDB, respectively,
with the 6-311G(d) basis set. The corresponding
B3LYP values were somewhat in between the HF
and the MP2 results, with r(CyC) syn values of
˚
1.335 A for both VDC and VDB (cf. Tables 7 and
8). The observed CyC bond distances in VDC and
VDB were seemingly quite large, giving rg values
˚
above 1.35 and 1.36 A, while for instance in
anti-PCV the rgꢀCyC† parameter value was found to
˚
be 1.349(12) A [2]. However, the CyC bond distance
was also found quite large in the vinyl trimethyl silane
ꢀ
[4], with rgꢀCyC† ˆ 1:359ꢀ10† A; which was insignif-
icantly different from the values obtained in the
present work. The value of the rg(C–Si) parameter
˚
refined to 1.847(5) A for VDC, while a value of
˚
only 1.827(9) A was obtained for the VDB molecule,
which was unexpectedly different. The ab initio calcu-
lations gave almost similar values for this parameter.
˚
We found 1.845 and 1.848 A for syn-VDC and -VDB
from the HF/6-311G(d) results, respectively (cf. Table
6); and the MP2 results also gave similar values
˚
(1.840 and 1.842 A for syn-VDC and -VDB, respec-
tively). The DFT values were also in that range (1.848
˚
and 1.851 A for syn-VDC and -VDB, respectively).
We have no immediate explanation for this observed
difference in the C–Si bond distance between VDC
and VDB. This difference could most likely be caused
by a systematic error inherent in the GED data for the
bromine molecule, as ab initio calculations are usually
able to calculate such differences quite accurately.
However, none of the other parameters seem to be
affected significantly by this eventual data error.
In Table 9, results for several vinylsilanes studied
by GED are compared. The value for the rg(Si–Cl)
Acknowledgements
We are very grateful to Professor Kenneth W.
Hedberg for use of the Oregon State University
GED-apparatus, and to Dr Lise Hedberg for help
with the normal coordinate calculations used in the
GED part of this work. This work has received
support from National Science Foundation under
grant CHE95-23581, and from The Research Council
of Norway (Program for Supercomputing) and
Oregon State University through grants of
˚
parameter in VDC (2.042(2) A) was observed to be
slightly smaller than the value found in methyl vinyl
˚
dichlorosilane (2.051(2) A) [1], while the Si–Br bond
ꢀ
distance in VDB was similar (rg ˆ 2:206ꢀ2† A) to the
ones found in the molecules 1,1,2,2-tetrabromo-

T.H. Johansen et al. / Journal of Molecular Structure 550–551 (2000) 257–279
279
Petersson, J.A. Montgomery, K. Raghavachari, M.A. Al-
Laham, V.G. Zakrzewski, J.V. Ortiz, J.B. Foresman, J.
Cioslowski, B.B. Stefanov, A. Nanayakkara, M. Challacombe,
C.Y. Peng, P.Y. Ayala, W. Chen, M.W. Wong, J.L. Andres,
E.S. Replogle, R. Gomperts, R.L. Martin, D.J. Fox, J.S.
Binkley, D.J. Defrees, J. Baker, J.P. Stewart, M. Head-
Gordon, C. Gonzalez, J.A. Pople, Gaussian 94, Revision
C.2 and D.4, Gaussian, Inc., Pittsburgh, PA, 1995.
¨
computing time. K. Hassler and U. Patzold thank
¨
the ‘Osterreichische Nationalbank’ for financial
support.
References
[1] T.H. Johansen, K. Hagen, R. Stølevik, Hassler, J. Phys. Chem.
A 101 (1997) 3580.
[2] T.H. Johansen, K. Hagen, K. Hassler, A. Richardson, U.
[18] L. Hedberg, I.M. Mills, J. Mol. Spectrosc. 160 (1993) 117
(asym20). The asym40 version presented in: L. Hedberg,
Abstracts, 15th Austin Symposium on Molecular Structure,
Austin, TX, 1994.).
[19] S. Huzinaga, M. Klobukowski, J. Mol. Struct. (Theochem)
167 (1988) 1.
¨
Patzold, R. Stølevik, J. Phys. Chem. A 101 (1997) 9641.
[3] J.R. Durig, J.F. Sullivan, G.A. Guirgis, M.A. Qtaitat, J. Phys.
Chem. 95 (1991) 1563 (and references therein).
[4] E.M. Page, D.A. Rice, R. Walsh, K. Hagen, J. Mol. Struct. 403
(1997) 199.
[20] D. Van Hemelrijk, L. Van den Enden, H.J. Geise, H.L. Sellers,
¨
L. Schafer, J. Am. Chem. Soc. 102 (1980) 2189.
[5] Q. Shen, J. Mol. Struct. 95 (1982) 215.
¨
[21] L. Schafer, J.D. Ewbank, K. Siam, N. Chiu, H.L. Sellers, in: I.
[6] G.A. Guirgis, Z. Shens, M.A. Qtaitat, J.R. Durig, J. Mol.
Struct. 403 (1997) 57.
[7] V.A. Naumov, M.B. Zuev, D.W.H. Rankin, H.E. Robertson, J.
Mol. Struct. 318 (1994) 151.
[8] K. Hagen, K. Hedberg, J. Am. Chem. Soc. 95 (1973) 1003.
[9] B. Andersen, H.M. Seip, T.G. Strand, R. Stølevik, Acta Chem,
Scand. 23 (1969) 3224.
[10] G. Gundersen, K. Hedberg, J. Chem. Phys. 51 (1969) 2500.
[11] L. Hedberg, Abstracts of papers, Fifth Austin Symposium on
Gas-Phase Molecular Structure, Austin, TX, 1974, p. 37.
[12] A.W. Ross, M. Fink, R. Hilderbrandt, International Tables of
Crystallography, vol. 4, Kluwer Academic, Dordrecht, 1992
(p. 245).
[13] V.I. Goreva, L.V. Khristenko, Yu.A. Pentin, Vopr. Stereo-
khim. 2 (1972) 57.
[14] L.V. Khristenko, Yu.A. Pentin, Vestn. Mosk. Univ. Ser. 2
Khim. 31 (1976) 304.
[15] S.G. Frankiss, Spectrochim. Acta 22 (1966) 295.
[16] J.R. Durig, G.A. Guirgis, M.A. Qtaitat, J. Raman Spectrosc.
26 (1995) 413.
[17] M.J. Frisch, G.W. Trucks, H.B. Schlegel, P.M.W. Gill, B.G.
Johnson, M.A. Robb, J.R. Cheeseman, T. Keith, G.A.
Hargittai, M. Hargittai (Eds.), Stereochemical Applications of
Gas-Phase Electron Diffraction, Part A, VCH Publishers, New
York, 1988, p. 301.
[22] V.J. Klimkowski, J.D. Ewbank, C. Van Alsenoy, J.N. Scar-
¨
dale, L. Schafer, J. Am. Chem. Soc. 104 (1982) 1476.
[23] J.A. Pople, M. Head-Gordon, D.J. Fox, K. Raghavachari, L.A.
Curtiss, J. Chem. Phys. 90 (1989) 5622.
[24] K. Hedberg, in: J. Laane, M. Dakkouri, B. van der Veken, H.
Oberhammer (Eds.), Structures and Conformations of Non-
Rigid Molecules, NATO ASI series C, vol. 410, Kluwer
Academic, Dordrecht, 1993, pp. 423–445.
[25] K. Hedberg, M. Iwasaki, Acta Crystallogr. 17 (1964) 529.
[26] L.S. Bartell, J. Chem. Phys. 38 (1963) 1827.
[27] K. Kuchitsu, Y. Morino, Bull. Chem. Soc. Jpn 38 (1965) 841.
[28] L.V. Vilkov, V.S. Mastryukov, P.A. Akishin, Russ. J. Struct.
Chem. 5 (1964) 168.
[29] H. Thomassen, K. Hagen, R. Stølevik, K. Hassler, J. Mol.
Struct. 147 (1986) 331.
[30] T.H. Johansen, K. Hagen, R. Stølevik, M. Ernst, K. Hassler, J.
Mol. Struct. 372 (1995) 161.