Conformational polymorphism of solid tetramesityldisilene Mes2Si=SiMes2 (Raman, UV-vis, IR and fluorescence study)

Article abstract of DOI:10.1016/S1386-1425(03)00004-0

Conformational polymorphism of solid tetramesityldisilene (1) has been studied by the methods of optical spectroscopy. The three known modifications of 1: orange unsolvated 1a and two 1:1 solvates with toluene (1b) and THF (1c) have been found to transform under specific conditions to a new, most thermodynamically stable polymorph, yellow unsolvated powder 1d. The latter has been characterized by the Raman, IR, UV-vis and fluorescence data. All forms of 1 exhibit Raman spectra differing in details, which reflect their different crystal and molecular structures. Unsolvated 1a and 1d differ significantly in electronic absorption and fluorescence emission. The yellow form 1d can be converted to the orange form 1a upon illumination with laser light in the region 514-457 nm. Similarity of the Raman and UV-vis spectra of 1d to those of the solutions of 1 provides some evidence for a quasi-trans conformation of 1d.

Full text of DOI:10.1016/S1386-1425(03)00004-0

Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
www.elsevier.com/locate/saa
Conformational polymorphism of solid tetramesityldisilene
Mes2SiÄSiMes2 (Raman, UVꢀ
vis, IR and fluorescence study)ꢀ
/
/
L.A. Leites a,*, S.S. Bukalov a, J.E. Mangette b, T.A. Schmedake b, R. West b
a A.N. Nesmeyanov Institute of Organoelement Compounds, Scientific and Technical Center on Raman Spectroscopy,
Russian Academy of Sciences, 28 Vavilova str., 119991 Moscow, Russia
b Organosilicon Research Center, University of Wisconsin, Madison, WI 53706 USA
Received 12 November 2002; accepted 22 November 2002
Abstract
Conformational polymorphism of solid tetramesityldisilene (1) has been studied by the methods of optical
spectroscopy. The three known modifications of 1: orange unsolvated 1a and two 1:1 solvates with toluene (1b) and
THF (1c) have been found to transform under specific conditions to a new, most thermodynamically stable polymorph,
yellow unsolvated powder 1d. The latter has been characterized by the Raman, IR, UVꢀvis and fluorescence data. All
/
forms of 1 exhibit Raman spectra differing in details, which reflect their different crystal and molecular structures.
Unsolvated 1a and 1d differ significantly in electronic absorption and fluorescence emission. The yellow form 1d can be
converted to the orange form 1a upon illumination with laser light in the region 514ꢀ
and UVꢀvis spectra of 1d to those of the solutions of 1 provides some evidence for a quasi-trans conformation of 1d.
# 2003 Elsevier Science B.V. All rights reserved.
/457 nm. Similarity of the Raman
/
Keywords: Disilene; Conformational polymorphism; Raman; UVꢀvis spectra; Fluorescence
/
1. Introduction
have been reported for this compound: an un-
solvated modification 1a, orange at room tem-
perature and thermochromic, obtained from
solution in hexane [6]; a yellow, crystalline 1:1
Tetramesityldisilene (1) was the first SiÄ
doubly bonded compound to be synthesized [2]
and is undoubtedly the best studied of disilenes [3ꢀ
5]. Nevertheless, details of its structural and
conformational behaviour are not completely
understood. Three different crystalline forms
/
Si
/
toluene solvate, (Mes4Si2×
similar solvate with one molecule of THF,
THF), 1c [8]. Recently, formation of
/C7H8), 1b [7]; and a
(Mes4Si2×
/
the fourth, unsolvated yellow form 1d was de-
scribed but briefly [1,9].
The X-ray data [6ꢀ
and the molecular structures of 1aꢀ
several respects (Table 1; Fig. 1). The C2SiÄ
/
8] show that both the crystal
c differ in
SiC2
/
ꢀ Some of these data were briefly published in Mendeleev
Commun. (1998) 43.
/
moiety is slightly distorted from planarity in all
three forms, but differently, with the twisting and
* Corresponding author. Fax: ꢁ
E-mail address: buklei@ineos.ac.ru (L.A. Leites).
/
7-95-135-5085.
1386-1425/03/$ - see front matter # 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S1386-1425(03)00004-0

1976
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
Table 1
X-ray data for conformational polymorphs of 1
Compound
Space group
Site sym
Z
r(SiÄ/Si), pm
Twist angle, (8)a
Pyram. angle, (8)b
Ring orient, (8)c
1a Orange
P21/c
I41/a
Ccca
C1-1
C2-2
4
8
4
214.3
216.0
214.6
3
12
13
12, 14
42, 52, 67, 70
35, 35, 78, 78
62.6
1b 1×
/
C7H8
THF
18
0
1c 1×
/
C2-222
a
b
c
The dihedral angle between the CÃ
The angle between the CÃSiÃC plane and the SiÃ
The dihedral angle between the aromatic ring plane and the CÃ
/
SiÃ
/
C and C?Ã
/
Si?Ã
/
C? planes.
/
/
/
Si vector.
/
SiÃSi plane.
/
pyramidalization both largest for 1b. The aromatic
rings are twisted out of this plane in all structures,
giving a helical arrangement with all torsion angles
different (42, 52, 67 and 708) for 1a, and all angles
equal at 628 for 1c. The aromatic rings are also
helically disposed in 1b, but with the two of the
cis-mesityl groups being more nearly coplanar
(358) and the other two more nearly orthogonal
(788). There is no specific interaction between 1
and the solvent molecules in the solvates 1bꢀc,
/
Fig. 1. Structures of 1aꢀd and conditions of their interconversions.
/

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1977
both being typical ‘packing crystals’ [10] where the
role of the solvent molecule is simply to fill space
in the loose crystal lattice. The three structures
demonstrate that 1 enjoys considerable conforma-
3) to try to propose a structure for the newly
found form 1d which does not form single
crystals suitable for X-ray diffraction study;
4) to investigate the temperature dependence of
the electronic and vibrational spectra of 1a in
connection with the problem of its thermo-
chromism.
tional flexibility. Distortions around the SiÄ
double bond are consistent with theoretical (ab
initio) calculations for H2SiÄSiH2 and Me2SiÄ
/
Si
/
/
SiMe2 [11a], which predict a very flat potential
energy surface for these model disilenes, with a
difference of energy between the planar and
2. Results
distorted skeletons of only 1ꢀ2 kcal/mol. The
/
results [11] suggest that the structures which will
actually be found for disilenes with bulky sub-
stituents may well be strongly affected by sub-
stituent steric hindrance, crystal packing forces
and other environmental effects.
2.1. Interconversions of 1aꢀd
/
These interconversions as well as different
structures of solid 1 are presented in Fig. 1. In
the course of experimenting with the orange
crystals 1a (in inert atmosphere or in high
vacuum), we found that, when heated or sublimed
or ground in a mortar, they easily transform to a
yellow powder 1d whose UV band at ꢀ
confirms it is also a disilene [2ꢀ5,13]. The same
substance 1d was obtained when solvates 1b or 1c
were exposed to high vacuum and heated to
remove the solvent. The identity of 1d, obtained
Different conformations due to different mesityl
ring orientations are also to be expected. Conjuga-
tion between the aromatic rings and the SiÄ
/Si
double bond would be maximized in a fully
coplanar conformation, but this is impossible
sterically, so that twisting of the rings must take
place. Fischer [12] has shown that in the related
trans-diarylethylenes and similar compounds, al-
most isoenergetic rotamers are present.
/425 nm
/
from unsolvated 1a and from the solvates 1bꢀ
was proved by the complete coincidence of their
Raman and UVꢀvis spectra. Numerous efforts to
/
c,
The orange crystals of 1a were found to exhibit
reversible thermochromism, turning to pale yellow
on cooling and to red on heating [6]. X-ray
diffraction studies of 1a carried out at 173 and
295 K showed that both crystal and molecular
structures are experimentally equivalent at these
two temperatures. This led to the suggestion that
the thermochromism might result from a tempera-
ture redistribution of molecules among low-lying
vibrational levels of the ground electronic state.
No corresponding thermochromism was observed
for the yellow crystals of 1b or for solutions of 1.
The aims of this work are as follows:
/
prepare a good crystal of unsolvated 1d for single-
crystal X-ray analysis were unsuccessful. Crystal-
lization from solutions lead to formation of either
1a or the corresponding solvates, whereas sub-
limation of 1d resulted in a fine powder on a glass
surface. It seems that 1d molecules are incapable of
forming perfect crystals due to their particular
shape incovenient for close packing (Aufbau
principle of Kitaigorodsky [10]).
Unsolvated forms 1a and 1d were found to
interconvert upon laser irradiation (in the course
of obtaining Raman or fluorescence spectra), if the
laser light density overcomes a certain threshold.
Moreover, the yellow solvates 1b and 1c can also,
sometimes, be converted to orange 1a. For in-
stance, long exposure to the 514.5 nm laser beam
produced a bright orange spot in the yellow
sample of 1c sealed in a capillary. The spot
persisted when the capillary was removed from
light. When exposed to light of the same wave-
1) to obtain and to compare in detail vibrational
(Raman, IR) and electronic (UVꢀ
tion, fluorescence emission) spectra of the
reported modifications (1aꢀd) in order to
/vis absorp-
/
find out how the small changes in their
molecular and crystal structure affect their
spectra;
2) to study interconversions of modifications
(1aꢀd);
/

1978
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
length but of higher intensity, the crystals of 1b,
sealed in vacuo in a quartz cell, on one occasion
suddenly emitted a yellow ‘cloud’, a solid part of
which settled on the upper wall of the cell, well
separated from the initial crystals. According to
the Raman spectrum, this solid film appeared to
be a mixture of 1a and 1d.
2.2. Vibrational spectra of solid 1aꢀd
/
When examining vibrational spectrum of a
disilene, a chemist is most interested in identifica-
tion of the stretching vibration of the SiÄ
/
Si bond,
the so-called n(SiÄSi). This problem was consid-
/
ered in detail in our previous papers [14] where the
Raman and IR spectra of three disilenes were
reported along with the results of normal coordi-
nate analysis (NCA) for several disilene models.
Herewith we report the Raman spectra of four
solid forms of 1 and of its solution in hexane. The
results are presented in Figs. 2ꢀ
/
4, 10ꢀ12, 15 and
/
16. IR spectra of unsolvated forms 1a and 1d are
also given (Fig. 5).
Molecule 1 consists of 82 atoms and, formally
speaking, its vibrational spectrum must have 240
fundamentals. However, vast experience in vibra-
tional spectroscopy of organometallics suggests
that internal vibrations of mesityl groups do not
interact kinematically through the heavier silicon
atoms, so that the frequencies of their similar
vibrations should coincide, leading to a great
simplification of the spectrum. Vibrational spec-
trum of mesitylene was investigated in detail in
Ref. [15], and vibrational modes of a mesityl group
attached to silicon atom were analyzed in terms of
Cs local symmetry in Ref. [14]. Most of the latter
vibrations are situated in the region above 750
Fig. 2. Raman spectra of solid 1aꢀ
at ambient temperature in the region 450ꢀ
1b; (c) 1c; (d) 1d.
/
d with 514.5 nm excitation
/
750 cmꢂ1. (a) 1a; (b)
in the SiÄ
/
Si bond. The SiÄ
/
Si and SiÃ
/C internal
stretching coordinates are heavily mixed. Their in-
phase combination results in a symmetric Raman
active normal mode with frequency in the range
460ꢀ
550 cmꢂ1 (n1) while their out-of-phase com-
/
bination gives a symmetric Raman active normal
mode near 700 cmꢂ1 (n2). Variation of the force
constant FSiÄSi within reasonable limits does not
alter this general conclusion. Particular contribu-
cmꢂ1. Vibrations of the framework C2SiÄ
/SiC2
should obey their own symmetry and their fre-
quencies are anticipated in the region below 750
cmꢂ1. However, some of the mesityl group modes
are also situated in this region. For correct assign-
ment of the bands observed, the results of normal
coordinate calculations [14] appeared particularly
useful.
tions from the SiÄ
/
Si and SiÃ/C stretching coordi-
nates to the eigenvectors of n1 and n2, the normal
modes of principal interest, depend on particular
molecular structure and substituents but both
contributions are always significant. Therefore,
The main result of the NCA obtained for the
the assignment of the n1 mode as the n(SiÄ
/
Si)
stretch and of the n2 mode as the totally symmetric
C2SiÄ
/
SiC2 moiety is that there is no normal mode
in the spectrum of a disilene which is well-localized

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1979
Fig. 3. Evolution of the Raman spectrum of 1c sample in the
process of its degradation in air (514.5 nm excitation; room
temperature). (1) initial spectrum; (2ꢀ8) the process of degra-
dation during 5 days; (9) Raman spectrum of Mes2Si(O)2Mes2.
/
ns(SiÃ
/
C) stretch is in essence arbitrary. The kine-
Fig. 4. Raman spectra of solid 1aꢀ
region; 514.5 nm excitation, ambient temperature. (a) 1a; (b) 1b;
(c) 1c; (d) 1d.
/
d in the low-frequency
matic effect of the out-of-plane twisting of the SiC2
fragment on the NCA results is small, with the
biggest frequency shift in the spectrum being only
10 cmꢂ1
.
For mesityl substituted disilenes, the results of
NCA [14] predict that the Raman active n1 mode,
mode, with similar frequencies but with different
intensities, because the intensity depends on the
mode eigenvector as well. In addition, NCA has
also shown that two of the symmetric Raman
active vibrations, localized in the mesityl group
(the ring breathing and deformation modes, n3 and
n4, respectively), should fall in the same frequency
still being an in-phase combination of the SiÄ
and SiÃC stretching coordinates, also becomes
/Si
/
mixed with symmetric angle deformations in
mesityl group. The degree of the latter mixing
(reflected by the eigenvector elements) depends
strongly on the mutual orientation of the disilene
and mesityl moieties. The more coplanar these
fragments, the more coupled are their vibrations.
At the same time, it is very important that the
frequency of the n1 mode is not affected by this
mixing. Therefore mesityl-containing disilenes
with different ring orientations are expected to
exhibit Raman bands, corresponding to the n1
range as n1, 500ꢀ
/
550 cmꢂ1
.
The Raman spectra of 1aꢀ
/d (excited by the
514.5 nm line) in the diagnostic region 450ꢀ
/
750
cmꢂ1 are presented in Fig. 2 and can be easily
rationalized on the basis of the NCA results. All of
the spectra exhibit a triplet in the region 500ꢀ550
/
cmꢂ1, which is evidently an overlap of three lines,
corresponding to n1, n3 and n4. Contrary to the
Raman spectra of E-tBuMesSiÄ
/SiMes(tBu) [14]

1980
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
Fig. 5. IR spectra of 1a and 1d as mulls in degassed apiezon (Section 4). The bands of apiezon are marked by asterisks. Identity of the
compounds was checked by the UVꢀvis band positions of the very same samples.
/
and of a Z-diaminodisilyldisilene [13], where the n1
mode manifests itself as a most intense line at
about 530 cmꢂ1, the Raman intensity of the n1 line
for all polymorphs of 1 is medium, comparable
with those of n3 and n4. Which of these could be
assigned as n1 will be discussed below. The assign-
ment of the most intense line at about 680 cmꢂ1 to
n2 is straightforward. Significant contribution
optimum registration conditions) revealed reliable
differences between their band parameters. For
instance, the frequencies of the n2 mode are 681,
676.5, 675, and 674 cmꢂ1 for 1a, 1b, 1c and 1d,
respectively. The half-width of these lines also
varies within 9ꢀ
/
15 cmꢂ1. The frequencies of the
550
components of the triplet in the region 500ꢀ
/
cmꢂ1 are close in value for all polymorphs (Fig.
2), whereas their intensities vary distinctly, each
modification exhibiting its own characteristic pat-
tern in this region. Also notable are different
intensity ratios of the n2 line and the neighboring
from the SiÄ
/
Si stretch to this so-called ns(SiÃ
/
C)
mode is confirmed by gradual disappearance of
this line in the process of degradation of the
disilene in air with formation of Mes2Si(O)2SiMes2
(Fig. 3) [16].
n(SiÃ
/
C) line at 720 cmꢂ1 in the four spectra. All
Precise frequency measurements performed in
these observations are in good accord with the
NCA results, mentioned above. The differences in
the Raman spectra of 1aꢀ
/
d (with accuracy 9
/
0.5
cmꢂ1, using spectral slit width of 1 cmꢂ1 and
the Raman spectra of 1aꢀd reflect their different
/

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1981
molecular structures and allow their unambiguous
identification by the Raman method.
The reason for the intensity redistributions in
the Raman spectra under consideration is not only
kinematic (different ring orientations in the mole-
cules of 1aꢀd leading to a change in the corre-
/
sponding eigenvector elements) but also electronic.
The change in the aromatic ring arrangements
affects the degree of pꢀp conjugation in the
/
molecule, which in its turn affects the Raman
intensity of the n1, n2 and, possibly, n3 lines, since
the Raman intensity of the appropriate lines is
known to be very sensitive to conjugation [17ꢀ19].
/
There are also notable differences between the
four forms of 1 in the low-frequency region of the
Raman spectrum where the crystal lattice modes
are situated (Fig. 4). These differences agree well
Fig. 6. UVꢀ
temperatures. The sample was obtained by sublimation on a
quartz window in vacuum cryostat. (a) ꢂ160 8C; (b) ꢂ70 8C;
(c) 20 8C.
/
vis absorption spectrum of solid 1d at various
/
/
with the X-ray data [6ꢀ
/
8], which point to different
crystal structures of 1aꢀ/c. From Fig. 4 it also
follows that the crystal structure of 1d (which has
not been characterized by the X-ray data) differs
from those of 1aꢀ
allow the conclusion that its crystal and molecular
structure is distinct from those of 1aꢀc.
In the IR spectra of 1aꢀd, in contrast to the
Raman spectra, the symmetric n1ꢀn4 modes man-
/
c. Thus, the Raman data for 1d
/
/
/
ifest themselves as very weak features, if at all (Fig.
5). The IR spectra are strongly dominated by
mesityl group vibrations and hence are not so
informative. Anyway, there are some differences
between the IR spectra of 1a and 1d, also in the
diagnostic region below 750 cmꢂ1
.
2.3. Electronic absorption and fluorescence
emission spectra
The UVꢀvis absorption spectrum of 1 was
/
investigated in 3-Me-pentane at 77 K and in
hexane solutions and was reported to contain a
lowest-energy band at ꢀ
first pꢀp* transition localized in a SiÄ
phore [2ꢀ5]. We obtained a band with a similar
lmax at about 425 nm for the solid sample of
yellow unsolvated form 1d prepared as a thin film
deposited on a cold target of the cryostat in vacuo
(Fig. 6) or as a nujol mull (Fig. 7). The lmax value
of this band varies slightly with temperature, being
/
420 nm, assigned to the
/
/
Si chromo-
Fig. 7. UVꢀvis absorption spectra of solid 1d and 1a as nujol
mulls. All the intermediate curves belong to various mixtures of
1a and 1d obtained as a result of 1a grinding (see text).
/
/
As was said above, our first attempt to obtain the
UVꢀ/vis spectrum of the solid 1a by subliming the
orange crystals in vacuo onto a cold window of the
424 nm at ꢂ160 8C and 428 nm at 20 8C (Fig. 6).
/

1982
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
cryostat, resulted in an excellent spectrum, but
which was surprisingly that of 1d. However,
different colors of 1a and 1d suggested that they
should differ also in electronic absorption. Indeed,
slowly transformed into 1a, this is clearly seen
from the lower curves of Fig. 8.
2.4. Pre-resonance enhancement in the Raman
spectra
the UVꢀ
/
vis spectrum of real 1a exhibited a band
40 nm red-shifted
at about 465 nm which is ꢀ
/
As the disilenes absorb in the visible region 420ꢀ
460 nm and as their lowest-energy electronic
absorption band is largely due to a SiÄSi pꢀp*
/
compared to that of 1d (Fig. 7). To obtain this
result, it appeared necessary to prepare the sample
without exposing the compound 1a to any me-
chanical or thermal stress. The best way was to put
small crystals in nujol, apiezon or silicone grease in
a dry box and just to slightly press this suspension
between the quartz windows almost without
grinding. Otherwise spectra of mixtures of 1a
and 1d in various proportions are obtained,
corresponding examples being given in Fig. 7.
As one might expect, we could not obtain
reliable electronic absorption spectra of the sol-
vates 1b and 1c because it was impossible to
determine whether or not or to what extent the
compounds have lost the solvents during sample
preparation. It should be noted, however, that the
compounds 1b, 1c and 1d seem to be of the same
color.
/
/
transition, the Raman spectra of disilenes, pro-
vided they are excited by visible light, should
exhibit pre-resonance enhancement of some lines.
It could be expected that, as the excited light
frequency approaches the disilene absorption band
(Fig. 9), those normal modes which are vibroni-
cally active in this electronic transition should be
most enhanced [17]. These are, above all, the n1
and n2 modes, involving the SiÄ/Si stretching
coordinate. However, if the aromatic rings are in
conjugation with the double bond, the totally
symmetric modes of the rings could be also
enhanced to some extent, depending on ring
orientation [18]. The dependence of the Raman
spectra of 1aꢀ
investigated, using the laser lines with vꢃ
514.5; 632.8; 647.1 and 1064 nm. The results
obtained for 1aꢀc are similar; the examples are
/
c on the exciting wavelength was
/457.9;
The fluorescence emission and excitation spectra
for a solution of 1 were reported [2ꢀ5]. We
/
/
recorded the fluorescence emission spectra excited
by the 457.9 nm laser line for both 1a and 1d as
solids (Fig. 8). The lmax values of their broad
presented in Figs. 10 and 11.
Excitation dependence of the Raman spectrum
of the THF solvate 1c is given in Fig. 10. To
eliminate the intensity increase due to the v4
scattering law, the spectra excited by different v
were normalized so as to make the intensity of the
line at 1380 cmꢂ1 approximately equal. The latter
line corresponds to a dCH3 mode, hardly sensitive
to resonance. But the intensity of this feature is too
small (Fig. 10) and therefore this normalization
procedure does not allow strict quantitative mea-
surements of line intensities, but only qualitative
comparisons between the spectra.
bands differ significantly (ꢀ
/
560 and ꢀ515 nm,
/
respectively) in good accord with the position of
electronic absorption bands. A large Stokes shift
points to a substantial geometry change upon
excitation. Under prolonged illumination, 1d
If the Raman spectrum of 1c is excited by the
1064 nm line in the near IR region, away from
resonance (Fig. 9), then the most intense is the line
of the aromatic ring stretching mode at 1600 cmꢂ1
(Fig. 10, curve 1). The intensity of the n2 line at
approximately 680 cmꢂ1 in this spectrum is
comparable with that of a mesityl mode at 1290
cmꢂ1, while the diagnostic triplet in the region of
Fig. 8. Fluorescence emission spectra of solid 1a and 1d excited
by 457.9 nm line. Two lower curves demonstrate gradual
transformation of 1d into 1a upon laser illumination.

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1983
Fig. 9. Location of the absorption bands of 1a and 1d and of the laser lines applied for Raman excitation.
500ꢀ
550 cmꢂ1, involving n1, is of modest inten-
sity. On approaching resonance, already with red
/
Fig. 10. Excitation dependence of the Raman spectrum of 1c.
(1) 1064 nm; (2) 647.1 nm; (3) 514.5 nm; (4) 457.9 nm.
Fig. 11. Excitation dependence of the Raman spectrum of 1a.
(1) 1064 nm; (2) 632.8 nm; (3) 514.5 nm; (4) 457.9 nm.

1984
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
(647.1 nm) excitation, a dramatic redistribution of
the Raman intensities is observed (Fig. 10, curve
2), demonstrating evident pre-resonance enhance-
ment of some lines, which regularly becomes
greater with green and blue excitation (Fig. 10,
curves 3 and 4). The most enhanced is evidently
the line at approximately 680 cmꢂ1 which dom-
inates in the spectrum. This is one more strong
piece of evidence of the complex origin of n2 with a
This assumption is confirmed by polarization
measurements in the Raman spectrum of a solu-
tion of 1 in hexane. Only two Raman lines are
observed in the n(SiÃ
/
C) region, those at approxi-
mately 680 (n2) and 720 cmꢂ1, both being
polarized. This suggests C2h symmetry of the
species predominant in solution, that is, a quasi-
trans structure of this conformer.
substantial SiÄ
spectrum excited in the blue region, close to
resonance, an overtone 2n2 at ꢀ
1350 cmꢂ1
/Si stretch contribution. In the
/
becomes noticeable, as is typical for pre-resonance
Raman spectra [17]. Some other lines correspond-
ing to mesityl group vibrations also exhibit in-
tensity enhancement, but not so great as n2.
Analogous results in a limited region are presented
in Fig. 11 for the orange form 1a. In the triplet at
It is important that the Raman pattern in the
region 500ꢀ
that observed for a solution of 1 in hexane (Fig.
12). Moreover, the UVꢀvis spectrum of solid 1d is
/
550 cmꢂ1 for solid 1d is very close to
/
similar to those of 1 in solutions (the data on
solution in hexane and 3-methylpentane are given
500ꢀ
550 cmꢂ1 the most enhanced for all forms of
/
1 is the line at 540 cmꢂ1 (Figs. 10 and 11). This
gives some reason to assign it to n1.
in Refs. [2ꢀ
/
5]; the UVꢀvis spectrum of solution of
/
1 in liquid Xe obtained at different temperatures is
presented in Fig. 13). As both the Raman and
It is notable that in the Raman spectra of the
solvates 1b and 1c excited in the visible region,
even the most intense lines of toluene and THF are
not seen, because their intensities are thousands of
times smaller than those of the pre-resonantly
enhanced disilene lines. To detect the most intense
lines of toluene or THF in the Raman spectra of
1b or 1c as weak features, excitation in the near
infrared was needed.
UVꢀvis absorption as well as fluorescence emis-
/
sion spectra of solid 1d are similar to those
observed for 1 in solution, we can tentatively
suggest for 1d also a quasi-trans structure shown
in Fig. 1.
2.5. On the structure of the unsolvated yellow form
1d
In spite of the fact that we have no X-ray data
for 1d, there are some reasons to speculate about
its structure. For disilenes of the type RR?SiÄ
/
SiR?R, photochemical cisꢀtrans isomerization in
/
solution was shown to occur [20], the trans isomer
being predominant under equilibrium conditions
at room temperature. Of course, symmetrically
substituted 1 cannot have real cisꢀtrans isomers,
/
but, by analogy, a similar equilibrium with pre-
dominance of a conformer, close to the trans one,
seems likely in solution.
Fig. 12. Comparison of the region 500ꢀ
550 cmꢂ1 of the
Raman spectra of solid 1d (1) and of concentrated solution of
/
1 in hexane (2).

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1985
To study the possible vibrational origin of the
thermochromism of 1a (see above), its Raman
spectrum was obtained at different temperatures.
It is evident that only depopulation of excited low-
lying vibrational modes upon cooling could lead to
the observed change in color. However, we in-
vestigated temperature dependence not only of the
low-frequency spectral region, but also of the
higher frequency region in order to obtain better
resolution of overlapping bands. As could be
expected, no substantial changes were found on
cooling to ꢂ
only the usual band narrowing leading to a better
550
/
180 8C for intramolecular vibrations,
resolution (Fig. 15), especially in the 500ꢀ
/
cmꢂ1 triplet. It became evident that the middle
peak of the triplet is the weakest one and thus
cannot correspond to n1. At low temperature, the
left and the right peaks of the triplet split in two
components each, probably due to Davydov
(correlation field splitting) effect.
Fig. 13. UVꢀvis spectrum of solution of 1 in liquid Xe at
various temperatures.
/
The low-frequency Raman region (Figs. 15 and
16) does exhibit some changes as the temperature
decreases. Apart from regular band narrowing and
shifts to higher frequencies, some redistributions
of line intensities seem to occur below 60 cmꢂ1
that could be possibly related to the presence of
lattice hot bands in the room temperature spec-
trum.
However, all these results do not give a definite
answer about the nature of thermochromism. The
latter is a common phenomenon among disilenes
(see Ref. 4, p. 241) and calls for further special
investigation.
2.6. Thermochromism of 1a
First of all, we have investigated temperature
dependence of the UVꢀ
Comparison of the spectra obtained at 20 and
180 8C is presented in Fig. 14. No real lmax shift
/vis spectrum of orange 1a.
ꢂ
/
was observed (within the limits of experimental
error) but the band narrowing is evident. The
latter could be the origin of the thermochromism,
human eyes being very sensitive to slight changes
in this spectral region [21].
3. Discussion
Thus, solid 1 exists as at least four forms
differing in both crystalline and molecular struc-
ture: two yellow crystalline 1:1 solvates with
toluene and THF (1b and 1c), and two solvent-
free forms: orange crystals 1a and a yellow powder
1d. Bernstein [22] uses a term ‘conformational
polymorphism’ for this phenomenon, that is,
existence of several forms of a conformationally
flexible molecule (with energy difference between
Fig. 14. UVꢀ
silicone grease, without grinding). (1) room temperature; (2)
180 8C.
/
vis absorption spectrum of solid 1a (suspension in
the conformers B/2 kcal/mol) depending on crys-
ꢂ/
tallization conditions. Our experience with a step-

1986
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
Fig. 15. Temperature evolution of the Raman spectrum of 1a in the region of intramolecular vibrations. (514.5 nm excitation). (1) 208;
(2) ꢂ1008; (3) ꢂ1408; (4) ꢂ180 8C.
/
/
/
of the mesityl groups, the tendency of the whole
molecule towards planarity which would maximize
pꢀ
packing.
The facts that both the solvates as well as the
orange form 1a easily convert to 1d clearly show
that forms 1aꢀc are metastable while form 1d is
the most stable thermodynamically. Modifications
1bꢀd can be converted to 1a, but only upon strong
/
p conjugation, and crystal forces favoring close
/
/
illumination.
It is notable that the bands in electronic
absorption and fluorescence emission spectra of
the orange form 1a (lmax at ꢀ
respectively) are significantly red-shifted compared
to those of the yellow 1d (ꢀ425 and ꢀ515 nm). It
seems unreasonable to suppose that pꢀp conjuga-
tion is maximal for 1a. There is some evidence that
the position of the low-energy UV band for
disilenes (as well as for aryl-substituted ethylenes)
/
465 and 560 nm,
/
/
Fig. 16. Temperature evolution of the Raman spectrum of 1a
in the low-frequency region (632.8 nm excitation). (1) 20 8C; (2)
ꢂ/180 8C.
/
is determined not only by degree of pꢀp conjuga-
/
tion with substituents, but also by distortion of the
double bond plane, its twisting leading to a red
shift ([4], p. 241). However, the distortion for
orange 1a is less than for yellow 1b (Table 1), so
this cannot be the explanation. The reason for the
lower-energy absorption and emission of 1a still
remains a mystery.
by-step serendipitous finding of new conforma-
tional polymorphs of 1 (Refs. [1,2,6ꢀ9], and this
paper) closely resembles the classical story about
analogous forms of dimethylbenzylideneaniline,
described vividly in Ref. [22].
/
Conformational polymorphism of 1 is evidently
a result of interplay between the steric hindrance

L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ
/1988
1987
4. Experimental
region were applied, those of an Arꢁ laser SP-
2020, a Krꢁ laser SP-165, and a Heꢀ
Ne laser SP-
124. Fourier-transform Raman spectra of 1aꢀc
were recorded with a Bruker-88 spectrometer
equipped with a FRA 106 FT Raman module.
For temperature measurements, various types of
cryostats were applied, including a Meric 91
Arpagon and an Air Products cryogenic system
DE 220.
/
The synthesis of 1 was accomplished according
to the method reported previously [2]. All experi-
ments were carried out under high vacuum or
strictly inert atmosphere to prevent sample oxida-
tion, which was checked by the absence in the
spectra of the bands corresponding to products of
degradation.
/
To remove the solvents from solvates 1b and 1c,
the substances put in quartz cells were heated in
high vacuum (10ꢂ5 Torr) to ꢀ
/
60 and ꢀ100 8C,
/
Acknowledgements
respectively, because THF was lost more reluc-
tantly. A yellow layer of 1d sublimed on the upper
part of the cell, while caked residue remained on
the bottom.
For Raman and fluorescence studies, the sam-
ples were sealed in capillaries in vacuo or in an
The Russian authors acknowledge financial
support of the Organosilicon Research Center
and Dow Corning Co. and of the Russian
Foundation for Basic Research (grant #01-03-
33057).
inert atmosphere. To obtain the UVꢀvis and IR
/
spectra of solid orange 1a was a difficult task. We
used samples prepared in degased nujol, apiezon
or silicone grease in an inert atmosphere almost
without grinding because the latter procedure was
found to lead to transformation of 1a into 1d.
Analogous spectra of yellow unsolvated 1d were
obtained either for nujol or apiezon mulls or for
thin solid films obtained by slow sublimation of
the sample onto a cool quartz or CsI window of
the cryostat in high vacuum. Sublimation of 1a
also results in 1d.
References
[1] L.A. Leites, S.S. Bukalov, J.E. Mangette, T.A. Schmedake,
R. West, Mendeleev Commun. (1998) 43.
[2] R. West, M.J. Fink, J. Michl, Science 214 (1981) 1343.
[3] R. West, Pure Appl. Chem. 56 (1984) 163.
[4] (a) G. Raabe, J. Michl, Chem. Rev. 85 (1985) 419;
(b) G. Raabe, J. Michl, in: S. Patai, Z. Rappoport (Eds.),
The Chemistry of Organic Silicon Compounds (Chapter
17), Wiley, Chichester, 1989, p. 1015.
[5] R. Okazaki, R. West, Adv. Organomet. Chem. 39 (1995)
234.
The UVꢀvis spectra of solution of 1 in liquid Xe
/
were obtained applying the technique proposed
and elaborated by Bulanin [23]. A specially con-
structed high pressure (up to 60 atm) metal cell
with quartz windows and optical path length 5 cm
was used, which permitted to change the tempera-
[6] B.D. Shepherd, C.F. Campana, R. West, Heteroatom.
Chem. 1 (1990) 1.
[7] M.J. Fink, M.J. Michalczyk, K.J. Haller, R. West, J.
Michl, Organometallics 3 (1984) 793.
[8] M. Wind, D.R. Powell, R. West, Organometallics 15
(1996) 5772.
ture of the solution within the interval from ꢂ
to ꢂ5 8C. The temperature was maintained with
accuracy 91 8C.
The UVꢀvis absorption spectra were recorded
using a computerized Carl Zeiss M-40 and a
PerkinꢀElmer Lambda Array 3280 spectrophot-
ometers. The IR spectra were obtained using a
computerized Carl Zeiss M-82 and an IFS 113V
Bruker FT spectrometers. To obtain the Raman
/
80
[9] L.A. Leites, S.S. Bukalov, R. West, J.E. Mangette, T.A.
Schmedake, in: N. Auner, J. Weiss (Eds.), Organosilicon
IV. From Molecules to Materials, Wiley-VCH, Weinheim,
2000, p. 98.
/
/
/
[10] (a) A.I. Kitaigorodsky, Mixed Crystals (in Russian),
Nauka, Moscow, 1983;
/
(b) A.I. Kitaigorodsky, Molecular Crystals and Molecules,
Academic Press, NY, London, 1973.
[11] (a) K. Krogh-Jespersen, J. Am. Chem. Soc. 107 (1985) 537
(See also relevant recent papers);
(b) V.S. Mastryukov, J.E. Boggs, Struct. Chem. 11 (2000)
97;
and fluorescence spectra, a JobinꢀYvon HG2S, an
/
U-1000 and a T-64000 laser Raman spectrometers
were used. Various exciting lines in the visible
(c) Y.L. Wang, R.A. Poirier, Can. J. Chem. 76 (1998) 477.
[12] E. Fischer, J. Photochem. 17 (1981) 331.

1988
L.A. Leites et al. / Spectrochimica Acta Part A 59 (2003) 1975ꢀ1988
/
[13] T.A. Schmedake, M. Haaf, Y. Apeloig, T. Muller, S.
Bukalov, R. West, J. Am. Chem. Soc. 121 (1999) 9479.
[14] (a) L.A. Leites, S.S. Bukalov, I.A. Garbuzova, R. West,
(d) P.P. Shorygin, L.L. Krushinskij, J. Raman Spectrosc.
28 (1997) 383.
[18] R.J.H. Clark, in: R.J.H. Clark, R.E. Hester (Eds.),
Advances in Infrared and Raman Spectroscopy, vol. 1,
Heyden, London, 1975, p. 143.
J.E. Mangette, H. Spitzner, J. Organomet. Chem. 536ꢀ
/
537
(1997) 425;
(b) I.A. Garbuzova, L.A. Leites, S.S. Bukalov, J. Mol.
Struct. 410 (1997) 467.
[19] (a) E.D. Schmid, B. Brosa, J. Chem. Phys. 56 (1972) 6267;
(b) E.D. Schmid, B. Brosa, J. Chem. Phys. 58 (1973) 3871;
(c) E.D. Schmid, P. Schlenker, R.R.M. Brand, J. Raman
Spectrosc. 6 (1977) 314.
[15] B.V. Lokshin, V.T. Alexanyan, M.G. Ezernitskaya, Izv.
Akad. Nauk SSSR, Ser. Khim. (1982) 2265 (in Russian).
[16] H. Watanabe, K. Takeuchi, N. Fukawa, M. Kato, M.
Goto, Y. Nagai, Chem. Lett. (1987) 1341
[20] (a) M.J. Michalczyk, R. West, J. Michl, J. Am. Chem. Soc.
106 (1984) 821;
Thus, Watanabe et al. were partially right when they
assigned an intense Raman line at 685 cmꢂ1 of (tip)4Si2
(b) M.J. Michalczyk, R. West, J. Michl, Organometallics 4
(1985) 826.
(evidently, the n2 mode) to the n(SiÄSi) mode.
/
[21] J.N. Murrell, The Theory of the Electronic Spectra of
Organic Molecules, Wiley, NY, 1963, p. 3.
[22] J. Bernstein, J. Phys. D. Appl. Phys. 26 (1993) B66
(references cited therein).
[17] (a) P.P. Shorygin, Uspekhi Khimii (Russ. Chem. Rev.) 40
(1971) 694;
(b) P.P. Shorygin, Uspekhi Khimii (Russ. Chem. Rev.) 47
(1978) 1698;
(c) P.P. Shorygin, Pure Appl. Chem. 4 (1962) 87;
[23] M.O. Bulanin, J. Mol. Struct. 19 (1973) 59.