Homo- and cross-[2+2]-cycloaddition of 1,1-diphenylsilene and 1,1-diphenylgermene. Absolute rate constants for dimerization and the molecular structures and photochemistry of the resulting 1,3-dimetallacyclobutanes

Article abstract of DOI:10.1021/om990447k

Absolute rate constants for the head-to-tail [2+2]-dimerization of 1,1-diphenylsilene and 1,1-diphenylgermene have been determined in hexane and isooctane solution at 23 °C by laser flash photolysis, using the corresponding 1,1-diphenylmetallacyclobutanes

Full text of DOI:10.1021/om990447k

Organometallics 1999, 18, 5643-5652
5643
Hom o- a n d Cr oss-[2+2]-Cycloa d d ition of
,1-Dip h en ylsilen e a n d 1,1-Dip h en ylger m en e. Absolu te
Ra te Con sta n ts for Dim er iza tion a n d th e Molecu la r
Str u ctu r es a n d P h otoch em istr y of th e Resu ltin g
1
1
,3-Dim eta lla cyclobu ta n es
Nicholas P. Toltl, Mark Stradiotto, Tracy L. Morkin, and William J . Leigh*
Department of Chemistry, McMaster University, 1280 Main Street West,
Hamilton, Ontario, L8S 4M1 Canada
Received J une 10, 1999
Absolute rate constants for the head-to-tail [2+2]-dimerization of 1,1-diphenylsilene and
,1-diphenylgermene have been determined in hexane and isooctane solution at 23 °C by
1
laser flash photolysis, using the corresponding 1,1-diphenylmetallacyclobutanes as precur-
sors. This requires knowledge of the molar extinction coefficients at the UV absorption
maxima of the two transient species, which have been determined by a transient actinometric
procedure. The rate constants for dimerization of the two compounds are similar and within
a factor of about 2 of the diffusional rate constant in both cases. The 1,3-dimetallacyclobutane
dimerization products have been prepared by direct photolysis of the corresponding 1,1-
diphenylmetallacyclobutanes in dry isooctane. In addition, 1,1,3,3-tetraphenyl-1-germa-3-
silacyclobutane has been prepared by photolysis of a 1:1 mixture of the two metallacyclo-
butanes. The solid-state structures of the three 1,3-dimetallacyclobutanes have been
determined by single-crystal X-ray crystallography, and their photochemistry has been
studied. Laser flash and/or steady-state photolysis experiments indicate that all three
compounds cyclorevert to the corresponding 1,1-diphenylmetallaenes upon photolysis in
•
•
hydrocarbon solvents, most likely via the same ( M-C-M′-C ) 1,4-biradical intermediates
which link the metallaenes with their corresponding dimers via stepwise dimerization
mechanism. The quantum yield for photocycloreversion of the digermacyclobutane is roughly
one-fourth that of 1,1-diphenylgermacyclobutane. However, it is about 3 times higher than
that for the silagermacyclobutane and roughly 15 times higher than that for the disila-
cyclobutane. Singlet lifetimes have been determined for the metallacyclobutanes and the
disila- and digermacyclobutanes using single photon counting techniques. The implications
of these results for the mechanisms of the head-to-tail [2+2]-dimerization of silenes and
germenes are discussed.
In tr od u ction
cleophiles are carefully excluded. Of course, a number
of “stable” derivatives are known in both cases, but to
the extent that these compounds always contain one or
more sterically bulky substituents at one or both sites
of the MdC bond, their stability is often more a
reflection of a kinetic persistence toward dimerization
than to actual thermodynamic stabilization effects.
Head-to-tail [2+2]-cycloaddition to yield the corre-
sponding 1,3-dimetallacyclobutane derivative is the
most common mode of dimerization for both silenes and
germenes, in keeping with the substantial natural
The chemistry of multiple bonds between carbon and
one of the heavier group 14 elements such as silicon or
germanium has been of considerable interest over the
1
-3
past few decades.
The combined effects of a low
π-bond energy and the differences in electronegativity
between carbon and the heteroatom make such bonding
situations unfavorable thermodynamically, and as a
result, most known silenes and germenes exist only as
transient species in the gas phase and in solution.
Bimolecular reaction with oxygen-, nitrogen-, or even
carbon-centered nucleophiles is virtually always rela-
δ+
δ-
1,2
polarity of the
MdC
bond.
The head-to-head
regiochemistry is common with derivatives bearing
sterically bulky substituents at carbon, particularly
when they also reduce the natural MdC bond polarity
through electronic effects. Silenes of this type show
enhanced kinetic stability compared to naturally polar-
7
-1
tively facile and occurs at rates in excess of ca. 10 M
-1
s
in derivatives bearing simple alkyl or aryl substit-
4
-6
uents at the double bond.
This reactivity is sup-
planted by self-reaction (i.e., dimerization) when nu-
(
(
(
1) Brook, A. G.; Brook, M. A. Adv. Organomet. Chem. 1996, 39, 71.
2) Barrau, J .; Escudi e´ , J .; Satg e´ , J . Chem. Rev. 1990, 90, 283.
3) Escudi e´ , J .; Couret, C.; Ranaivonjatovo, H.; Satg e´ , J . Coord.
(5) Leigh, W. J .; Kerst, C.; Boukherroub, R.; Morkin, T. L.; J enkins,
S.; Sung, K.; Tidwell, T. T. J . Am. Chem. Soc. 1999, 121, 4744.
(6) Leigh, W. J .; Boukherroub, R.; Kerst, C. J . Am. Chem. Soc. 1998,
120, 9504.
Chem. Rev. 1994, 130, 427.
4) Leigh, W. J . Pure Appl. Chem. 1999, 71, 453.
(
1
0.1021/om990447k CCC: $18.00 © 1999 American Chemical Society
Publication on Web 12/01/1999

5
644 Organometallics, Vol. 18, No. 26, 1999
Toltl et al.
ized derivatives and often dimerize reversibly; when
simple alkyl substituents are present at carbon, the
main dimeric product is usually that corresponding to
formal ene-addition rather than [2+2]-cycloaddition.
Until recently, the commonly accepted view was that
the head-to-tail dimerization of silenes proceeds via a
concerted π2s + π2s cycloaddition mechanism, in which
the normal symmetry-imposed barrier for such a process
is reduced due to the strong polarization of the CdSi
bond. Head-to-head dimerization, on the other hand, is
germanium analogue 2b. We have previously shown
that these two transient compounds can be conveniently
generated and detected by laser flash photolysis of the
4
corresponding 1,1-diphenylmetallacyclobutanes 1; both
decay with predominant second-order kinetics when
generated in solution in the absence of nucleophiles,
consistent with the fact that they are both known to
dimerize to the corresponding 1,1,3,3-tetraphenyl-1,3-
1
9-21
dimetallacyclobutanes (3) (eq 1).
In several previous
•
thought to proceed in stepwise fashion involving a C-
•
Si-Si-C 1,4-biradical intermediate resulting from
initial Si-Si bond formation.7-9 Recent high-level ab
initio theoretical calculations have raised serious ques-
tions regarding the mechanism for the head-to-tail
dimerization of naturally polarized silenes. Depending
on the level of theory employed, the lowest energy
pathway varies between concerted10 and stepwise, the
latter involving the initial formation of either an anti
•
•
11,12
•
•
Si-C-Si-C biradical
or an anti C-Si-Si-C
biradical, which undergoes concerted rearrangement to
the head-to-tail dimer.13 Much less is known about
germene dimerization, although both head-to-tail and
head-to-head dimers have been reported,2 and the
effects of substituents on the preferred regiochemistry
appear to be the same as with silenes.14 As far as we
are aware, germene dimerization has not yet been
examined computationally.
So far only one solution-phase kinetic study has been
reported, on the head-to-head ene-type dimerization of
papers, we have studied the kinetics and mechanisms
1
5
a stabilized silene. It is interesting to note that the
reported Arrhenius parameters (Ea ) 0.2 ( 0.1 kcal/
mol and log A ) 7 ( 1) are indeed consistent with a
stepwise reaction mechanism, involving some interme-
diate (a 1,4-biradical, for example) which partitions
between product (via intramolecular disproportionation)
and starting materials at similar rates. The gas-phase
dimerization of the naturally polarized silene Me2Sid
CH2, which is well-known to proceed in head-to-tail
fashion,16 is very rapid and also proceeds with a
negligible Arrhenius activation energy over the 370-
of the bimolecular reactions of these potent electrophiles
21-26
with various nucleophilic trapping agents.
These
studies suggest that germenes can generally be expected
to show 10-1000 times lower reactivity than silenes of
corresponding structure, at least in polar addition
reactions in which the first step is attack of the
nucleophile at the metal end of the MdC bond. In the
present paper, we probe whether this generalization
applies as well to [2+2]-dimerization, through the
determination of absolute rate constants for the dimer-
ization of 2a and 2b in hydrocarbon solution at ambient
temperature.
1
7
5
00 °C temperature range. As has been recognized for
1
8
some time, such behavior is consistent with either
concerted or stepwise reaction mechanisms.
We have also isolated and structurally characterized
the 1,3-dimetallacyclobutane dimers of the silene and
germene (3a and 3b, respectively) and of the 1-germa-
We now wish to report the initial results of a detailed
experimental study of the head-to-tail dimerization of
3
-silacyclobutane (3c) derived from cross-[2+2]-cyclo-
“naturally polarized” silenes and germenes in solution,
addition of 2a and 2b (eq 2). To our knowledge, this is
the first reported example of a silene-germene cross-
cycloaddition reaction, and we had hoped to provide
some qualitative information on its rate relative to those
of the dimerization reactions of these two substrates.
The X-ray crystallographic study of the three 1,3-
dimetallacyclobutanes was prompted by the fact that
our original structural assignment of the 1,1-diphenyl-
germene dimer was based only on spectroscopic and
represented here by 1,1-diphenylsilene (2a ) and its
(
(
(
(
7) Gusel’nikov, L. E.; Nametkin, N. S. Chem. Rev. 1979, 79, 529.
8) Raabe, G.; Michl, J . Chem. Rev. 1985, 85, 419.
9) Brook, A. G.; Baines, K. M. Adv. Organomet. Chem. 1986, 25, 1.
10) Seidl, E. T.; Grev, R. S.; Schaefer, H. F., III. J . Am. Chem. Soc.
1
992, 114, 3643.
11) Bernardi, F.; Bottoni, A.; Olivucci, M.; Robb, M. A.; Venturini,
A. J . Am. Chem. Soc. 1993, 115, 3322.
12) Bernardi, F.; Bottoni, A.; Olivucci, M.; Venturini, A.; Robb, M.
J . Chem. Soc., Faraday Trans. 1994, 90, 1617.
13) Venturini, A.; Bernardi, F.; Olivucci, M.; Robb, M. A.; Rossi, A.
J . Am. Chem. Soc. 1998, 120, 1912.
14) Bravo-Zhivotovskii, D.; Zharov, I.; Kapon, M.; Apeloig, Y. J .
Chem. Soc., Chem. Commun. 1995, 1625.
15) Zhang, S.; Conlin, R. T.; McGarry, P. F.; Scaiano, J . C.
Organometallics 1992, 11, 2317.
16) Gusel’nikov, L. E.; Flowers, M. C. J . Chem. Soc., Chem.
Commun. 1967, 864.
17) Brix, Th.; Arthur, N. L.; Potzinger, P. J . Phys. Chem. 1989, 93,
193.
(
(
2
1
analytical data, which are not entirely unambiguous
(
(
(19) J utzi, P.; Langer, P. J . Organomet. Chem. 1980, 202, 401.
(20) Auner, N.; Grobe, J . J . Organomet. Chem. 1980, 197, 147.
(21) Toltl, N. P.; Leigh, W. J . J . Am. Chem. Soc. 1998, 120, 1172.
(22) Leigh, W. J .; Bradaric, C. J .; Sluggett, G. W. J . Am. Chem. Soc.
1993, 115, 5332.
(23) Leigh, W. J .; Bradaric, C. J .; Kerst, C.; Banisch, J . H. Organo-
metallics 1996, 15, 2246.
(24) Bradaric, C. J .; Leigh, W. J . J . Am. Chem. Soc. 1996, 118, 8971.
(25) Bradaric, C. J .; Leigh, W. J . Can. J . Chem. 1997, 75, 1393.
(26) Bradaric, C. J .; Leigh, W. J . Organometallics 1998, 17, 645.
(
(
(
8
1
(18) Houk, K. N.; Rondan, N. G.; Mareda, J . Tetrahedron 1985, 41,
555.

Homo- and Cross-[2+2]-Cycloaddition
Organometallics, Vol. 18, No. 26, 1999 5645
considering the symmetry possessed by both possible
isomeric structures. The structure of the silene dimer
3
a was originally established on the basis of compari-
sons to an authentic sample made by an independent
route,1
9,20
but its crystal structure has evidently not
been determined. We felt that it would be useful to do
so, as it provides an opportunity to compare the absolute
structures of a homologous series of 1,3-dimetalla-
cyclobutanes which differ only in the identity of the ring
heteroatoms.
Finally, we have studied the photochemistry of the
three 1,3-dimetallacyclobutanes in some detail, with the
intent of establishing their relative propensities toward
F igu r e 1. Transient decays for 2a (a) and 2b (b) in hexane
at 23 °C. The solid lines represent the fit of the data to eqs
7
and 8, respectively.
taining 0.5 M methanol, which leads to quantitative
trapping of the germene as methyldiphenylgermanium
[2+2]-cycloreversion to regenerate the silene and/or
germene. The photolyses of these compounds provide a
possible entry point to the potential energy surface
2
1
methoxide (4b). The formation of methoxymethyl-
diphenylsilane (4a ) from photolysis of 1,1-diphenyl-
silacyclobutane (1a ) under the same conditions (Φ4a )
•
corresponding to one of the nonconcerted ( M-C-M-
•
C biradical) mechanisms for silene and germene [2+2]-
2
3
0
.21 ( 0.02 ) was employed as actinometer. The pho-
tolyses were monitored at several conversions below
10%, and the relative quantum yields for formation
dimerization, since there is good evidence to suggest,
at least for 1,3-disilacyclobutanes, that photolysis re-
sults in initial Si-C bond cleavage to yield a 1,4-
∼
2
7,28
of 4a and 4b were taken as the relative slopes of product
concentration versus time plots. The experiment af-
forded a value of Φ4b ) 0.22 ( 0.02.
biradical intermediate.
Photolysis of 3a in solution
purportedly does not lead to cycloreversion,19 but noth-
ing appears to be known of the behavior of 1,3-sila-
germacyclobutanes or -digermacyclobutanes in this
regard. Our results afford a few potentially relevant
pieces of information on the mechanisms of silene and
germene dimerization and allow a comparison of at least
one aspect of the reactivity of 1,3-dimetallacyclobutane
derivatives as a function of the heteroatom.
Laser flash photolysis of continuously flowing, air- or
nitrogen-saturated solutions of 1a and 1b in hexane or
isooctane gave rise to the readily detectable transient
absorptions which we have previously assigned to 1,1-
diphenylsilene (2a ) and 1,1-diphenylgermene (2b), re-
4
,21
spectively.
The decay of the germene followed strict
second-order kinetics under these conditions, while the
silene exhibited mixed second- and pseudo-first-order
decay kinetics due to competing dimerization and reac-
tion with traces of water in the solvent. Representative
decay traces (in the form ∆ODt/∆OD0 vs time) are shown
in Figure 1. The second-order rate coefficients (2kdim/
ꢀ325-nm) were obtained by nonlinear least-squares fitting
of data from several experiments with each compound
(see Experimental Section for details); the values meas-
ured in hexane were identical to those measured in
isooctane within experimental error for both compounds,
and so the averages were calculated using data from
both solvents. The results are 2kdim/ꢀ325-nm ) (2.45 (
Resu lts a n d Discu ssion
Hom o- a n d Cr oss-[2+2]-Cycloa d d ition of 1,1-
Dip h en ylsilen e a n d 1,1-Dip h en ylger m en e. As has
been reported previously,1
9-21
steady-state photolysis of
deoxygenated 0.01 M solutions of 1,1-diphenylsila-
cyclobutane (1a ) or 1,1-diphenylgermacyclobutane (1b)
in cyclohexane or cyclohexane-d12 led in each case to
the formation of a single major product. The two
compounds were each isolated in analytically pure form
by radial chromatography and identified as the 1,3-
dimetallacyclobutanes 3a and 3b, respectively (eq 1).
The quantum yield for formation of 1,1-diphenyl-
germene (2b) from photolysis of 1b was determined by
merry-go-round photolysis of a deoxygenated 0.01 M
solution of the germacyclobutane in cyclohexane con-
6
-1
0.34) × 10 cm s for 2a (average of 4 determinations)
6
-1
and 2kdim/ꢀ325-nm ) (1.56 ( 0.42) × 10 cm s for 2b
(average of 7 determinations). The errors are quoted as
the standard deviations from the mean values and
(
27) Yoo, B. R.; Lee, M. E.; J ung, I. N. Organometallics 1992, 11,
626.
28) J ung, I. N.; Pae, D. H.; Yoo, B. R.; Lee, M. E.; J ones, P. R.
Organometallics 1989, 8, 2017.
cr oe sr rp ee sc pt io vn e dl yt.o confidence intervals of ∼85% and ∼95%,
1
Extinction coefficients (ꢀ) were determined for 2a ,b
in hexane solution at their absorption maxima (325 nm)
(

5
646 Organometallics, Vol. 18, No. 26, 1999
Toltl et al.
using the so-called intensity variation method.29,30 This
method involves comparing the ∆OD value at the very
beginning of the decay of the transient absorption
such a procedure might prove to be useful for the
synthesis of mixed 1,3-dimetallacyclobutanes, of which
relatively few examples are known. A statistical distri-
bution of homo- and cross-cycloaddition products is most
reasonably expected considering the absolute rate con-
stants, yet photolysis of an equimolar mixture of 1a and
1b (0.05 M each) in sodium-dried hexane solution
afforded the three dimetallacyclobutanes in relative
yields of 3a :3b:3c ) 1.0:4.8:2.1 at conversions less than
∼15%. This changed continuously over the course of the
photolysis to a value of 1.0:3.0:2.0 at ∼40% conversion.
Isolation of 3c was accomplished by reverse-phase
high-performance liquid chromatography, and its struc-
ture could be established unambiguously on the basis
(∆OD0) to the ∆OD0 of an actinometer, whose photolysis
yields a transient product of known ꢀ with a known
quantum yield, as a function of excitation laser inten-
sity. The actinometer employed was benzophenone,
whose irradiation yields the triplet state (λmax ) 525 nm,
-
1
-1 31
ꢀ
525-nm ) 6250 ( 940 M cm
)
with unit quantum
yield. The substrate and actinometer solutions were
optically matched at the laser wavelength (248 nm) to
ensure equal light absorption, and the laser intensity
was controlled using neutral density filters. Plots of
∆OD0 versus laser intensity, examples of which are
+
1
13
included as Supporting Information, were linear for all
three compounds, although that for the actinometer
showed some deviation at higher laser intensities (this
of its mass (M ) 434) and H NMR and C NMR
1
13
spectra. The high-field portions of the H and C NMR
H
C
spectra show single resonances at δ 1.35 and δ 3.85,
respectively, consistent with the D2h symmetry of the
head-to-tail cycloadduct. The H and C chemical shifts
are intermediate between those in the corresponding
spectra of 3a ( δ 1.10 and δ 0.85) and 3b ( δ 1.62 and
δ 7.71), a trend which supports our original structural
3
0
is normal ). Extinction coefficients for 2a ,b at their
absorption maxima were calculated from the relative
slopes of the plots and the transient product quantum
yields determined above. Duplicate determinations
1
13
H
C
H
-
1
C
yielded average values of ꢀ325-nm ) 8900 ( 1800 M
-
1
-1
-1
cm for 2a and ꢀ325-nm ) 11 400 ( 2700 M cm for
b. These values should be compared to that at the long-
wavelength absorption maximum of 1,1-diphenylethyl-
assignment for the germene dimer 3b. The same trend
2
in proton chemical shifts exists for the set of 1,1,3,3-
H
35
tetramethyl-1,3-dimetallacyclobutanes 5a ( δ 0.03), 5b
( δ 0.64), and 5c ( δ 0.23).
-
1
-1 32
H
36
H
36
ene, ꢀ250-nm ) 7680 M cm
.
Correction of the experimentally determined rate
coefficients by these values affords estimates of kdim )
1
0
-1 -1
9
-1 -1
(
1.1 ( 0.3) × 10
M
s
and (9.0 ( 3.2) × 10 M
s
for the absolute rate constants for dimerization of 2a
and 2b, respectively, in hexane/isooctane solution at 23
°C. The results indicate that in both cases dimerization
is the most rapid reaction that these species undergo.
Most importantly, the data indicate that the two species
dimerize at essentially equal rates in hydrocarbon
solvents and within a factor of about 2 of the diffusional
limit. This is in sharp contrast to their relative reac-
tivities toward nucleophiles such as amines, alcohols,
ketones, or carboxylic acids, where the germene is 10-
The distribution of the three 1,3-dimetallacyclo-
butanes obtained in this experiment at low conversions
(3a :3b:3c ≈ 1.0:4.8:2.1) indicates that photolysis of 1b
is roughly 3 times more efficient than that of 1a under
these conditions. This is supported by the observation
that the disappearance of 1b proceeded significantly
faster than 1a during the initial stages of the photolysis.
We were initially surprised by this result, since the
quantum yields for photolysis and the extinction coef-
ficients of the two compounds at the excitation wave-
length employed (254 nm) are the same. Photolysis of
mixtures containing higher (equimolar) concentrations
of 1a and 1b led to product distributions that were even
more dramatically skewed in favor of 3b; for example,
a distribution of 3a :3b:3c ≈ 1:10:2.5 was obtained upon
photolysis of a hexane solution of 1a (0.2 M) and 1b (0.2
M) to ∼25% conversion. These results suggest that there
is a second mechanism for excitation of 1b in addition
to direct light absorption, which results in the relative
steady-state concentrations of 3b and 3a being signifi-
cantly greater than 1 during the early stages of the
photolysis. This second mechanism can most likely be
identified as singlet energy transfer, which would be
expected to favor excitation of 1b if this compound has
a lower excited singlet state energy and/or longer excited
singlet state lifetime than 1a . Indeed, while the static
UV absorption spectra of 1a and 1b are superimposable,
the fluorescence emission spectrum of 1b is broader and
1
000 times less reactive than the silene under similar
4
conditions. This presumably indicates that dimeriza-
tion is substantially more exothermic than addition of
oxygen- or nitrogen-centered nucleophiles in the case
of the germene. These differences are almost certainly
less pronounced in the case of the silene,33,34 in agree-
ment with the much reduced spread in the absolute rate
constants for reaction of 2a with itself and the various
nucleophiles listed above.
With the indication that both 2a and 2b dimerize at
close to the diffusion-controlled rate in solution at
ambient temperatures, it seemed very likely that they
would also undergo cross-cycloaddition if they could be
generated together in the absence of nucleophiles. We
were encouraged to attempt such an experiment, since
(29) Carmichael, I.; Hug, G. L. In CRC Handbook of Organic
Photochemistry, Vol. I; Scaiano, J . C., Ed. CRC Press: Boca Raton,
1
1
1
989; pp 369-404.
30) Wintgens, V.; J ohnston, L. J .; Scaiano, J . C. J . Am. Chem. Soc.
988, 110, 511.
(
(31) Carmichael, I.; Hug, G. L. J . Phys. Chem. Ref. Data 1986, 15,
.
(32) The Sadtler Standard Spectra, 6401UV; Sadtler Research
F
shifted to lower energies (λmax ) 305 nm; see Support-
Laboratories, Inc., 1966.
(
(
33) Walsh, R. Acc. Chem. Res. 1981, 14, 246.
34) Walsh, R. In The Chemistry of Organic Silicon Compounds;
(35) Seyferth, D.; Attridge, C. J . J . Organomet. Chem. 1970, 21, 103.
(36) Mironov, V. F.; Gar, T. K.; Mikhailyants, S. A. Dokl. Akad. Nauk
SSSR 1969, 188, 718.
Patai, S., Rappoport, Z., Eds.; J ohn Wiley & Sons Ltd.: New York,
989; pp 371-391.
1

Homo- and Cross-[2+2]-Cycloaddition
Organometallics, Vol. 18, No. 26, 1999 5647
F igu r e 2. X-ray structures of 3a (left) and 3b (right), showing the atomic numbering schemes. Thermal ellipsoids are
shown at the 50% probability level. For 3b, only one of the two independent molecules is presented.
F
ing Information) compared to that of 1a (λmax ) 285
3
7
nm), indicating a significant difference in the geom-
etries and energies of the relaxed excited singlet states
of the two compounds.
Photolysis of solutions containing less than 0.05 M
each of 1a ,b led to initial product distributions similar
to those obtained in the 0.05 M experiments, but in
these cases, the reaction mixtures also contained ap-
preciable amounts of methyldiphenylsilanol and the
corresponding disiloxane due to preferential trapping
of 2a by adventitious water in the solvent. The Si:Ge
material balances in these experiments approached 1:1
as the concentration decreased however, indicating the
expected falloff in singlet energy transfer processes with
decreasing bulk substrate concentrations. Unfortu-
nately, we were unable to dry the solvent to a sufficient
level to allow a meaningful determination of the 3a :3b:
3
c product distribution at low substrate concentrations;
hence we are ultimately unable to estimate the relative
rates of homo- versus cross-[2+2]-cycloaddition of 3a
and 3b on the basis of these experiments.
F igu r e 3. X-ray structure of 3c, showing the atomic
numbering scheme. Only the complete independent mol-
ecule is presented with thermal ellipsoids shown at the 50%
probability level.
In all of these experiments, the variation in the
product distribution with increased photolysis times is
partially due to the fact that the relative concentrations
of the 1-metallacyclobutane starting materials (and
hence the importance of singlet energy transfer proc-
esses) changes with photolysis time. A second contribu-
tion to this effect arises from the fact that the three
products also undergo photocycloreversion to 2a and/or
three lattice packings, despite the fact that only the
identity of the heteroatoms differs in the three com-
pounds.
The planar four-membered ring in 3a (Figure 2, left)
contains endocyclic angles of approximately 90°, com-
parable to those reported for other crystallographically
3
8-42
characterized 1,3-disilacyclobutanes.
Perhaps sur-
2
b, but with different efficiencies (vide infra). This
prisingly, these geometric constraints result in only a
modest expansion of the associated exocyclic angles at
the carbon and silicon tetrahedral centers, with no
significant elongation of the Si-C ring bonds.43 The
transannular Si-Si distance (∼2.62 Å) in 3a is typical
results in equilibration of the three dimetallacyclo-
butanes at longer photolysis times, where the concen-
trations of the products are high enough to compete with
the starting materials for light absorption.
Molecu la r Str u ctu r es of th e 1,1,3,3-Tetr a p h en yl-
1
,3-d im eta lla cyclobu ta n es 3a-c. The crystallograph-
(
38) Braddock-Wilking, J .; Chiang, M. Y.; Gaspar, P. P. Organo-
ically determined structures of 3a , 3b, and 3c are shown
in Figures 2 and 3, while the relevant refinement
information and selected metrical parameters are col-
lected in Tables 1-3. It is particularly interesting to
note that no two of these structures are isomorphous,
and very different symmetry relationships exist in the
metallics 1993, 12, 197.
(39) Ishikawa, M.; Kovar, D.; Fuchikami, T.; Nishimura, K.; Ku-
mada, M.; Higuchi, T.; Miyamoto, S. J . Am. Chem. Soc. 1981, 103,
2
9
4
324.
(40) Boetze, B.; Herrschaft, B.; Auner, N. Chem. Eur. J . 1997, 3,
48.
(41) Auner, N.; Ziche, W.; Herdtweck, E. J . Organomet. Chem. 1992,
26, 1.
(
42) Krempner, C.; Reinke, H.; Oehme, H. Chem. Ber. 1995, 128,
1083.
(43) Klebe, G. J . Organomet. Chem. 1985, 293, 147.
(37) Leigh, W. J .; Boukherroub, R.; Bradaric, C. J .; Cserti, C.;
Schmeisser, J . M. Can. J . Chem. 1999, 77, 1136.

5
648 Organometallics, Vol. 18, No. 26, 1999
Ta ble 1. Cr ysta llogr a p h ic Collection a n d Refin em en t P a r a m eter s for 3a , 3b, a n d 3c
Toltl et al.
3
a
3b
3c
empirical formula
molecular weight
description
cryst dimens, mm
temp, K
C26H24Si2
392.63
colorless plate
0.34 × 0.11 × 0.06
300(2)
C26H24Ge2
481.63
colorless prism
0.20 × 0.16 × 0.10
300(2)
C26H24SiGe
437.13
colorless prism
0.26 × 0.12 × 0.07
300(2)
3
wavelength, Å
cryst syst
space group
a, Å
b, Å
c, Å
(Mo KR) 0.71073
monoclinic
C2/ c
20.5430(2)
6.1256(2)
17.8378(3)
90.000
97.206(1)
90.000
2226.95(8)
4
(Mo KR) 0.71073
(Mo KR) 0.71073
triclinic
triclinic
P( 1h )
P( 1h )
10.2432(2)
10.5455(1)
11.3542(1)
103.452(1)
102.52(2)
93.074(1)
1157.42(3)
2
6.1718(9)
14.406(3)
25.481(4)
81.446(2)
85.695(3)
88.502(3)
2233.7(6)
4
R, deg.
â, deg.
γ, deg.
volume, Å
3
Z
3
calcd density, g/cm
1.171
0.168
ω-scans
832
1.382
2.605
ω-scans
488
1.300
1.433
ω-scans
904
-
1
absorp coeff, mm
scan mode
F(000)
θ-range for collection, deg
2.00-22.49
1.90-22.50
-13 e h e 13
-13 e k e 13
-14 e l e 14
8478
1.43-22.50
-6 e h e 6
-15 e k e 15
-27 e l e 27
13941
index ranges
-26 e h e 26
-
5 e k e 7
-23 e l e 23
no. of reflns collected
no. of indep reflns
R(int)
6265
1461
0.0511
2982
0.0284
5856
0.0708
no. of data/restraints/params
1456/0/135
1.033
2976/0/269
1.037
5856/0/523
0.843
2
goodness-of-fit on F
final R indices (I > 2σ(I))
a
R1 ) 0.0396;
wR2 ) 0.0916
R1 ) 0.0573;
wR2 ) 0.1012
0.978, 0.453
0.152
R1 ) 0.0290;
wR2 ) 0.0693
R1 ) 0.0370;
wR2 ) 0.0737
0.704, 0.603
0.273
R1 ) 0.0450;
wR2 ) 0.0751
R1 ) 0.1286;
wR2 ) 0.0937
0.921, 0.681
0.292
R indices (all data)a
transmission (max, min)
3
largest diff peak, e/Å
largest diff hole, e/Å
3
-0.207
-0.237
-0.277
a
R1 ) ∑(|Fo| - |Fc|)/∑|Fo|; wR2 ) [∑[w(Fo - Fc ) ]/∑[w(Fo ) ]]0.5.
2
2
2
2 2
Ta ble 2. Selected Bon d Len gth s (Å) a n d An gles
d eg) for 3a a n d 3b
Ta ble 3. Selected Bon d Len gth s (Å) a n d An gles
a ,b
(
(d eg) for 3c
Si(1)-Si(1a)
2.620(1)
Ge(1)-C(1a)
Ge(1)-C(3)
Ge(1)-C(9)
Ge(2)-C(2)
Ge(2)-C(2a)
Ge(2)-C(15)
Ge(2)-C(21)
1.970(3)
1.954(3)
1.945(3)
1.978(4)
1.978(4)
1.949(3)
1.945(4)
90.9(2)
M(1)-M(2)
M(3)-M(3x)
M(4)-M(4x)
M(1)-C(1a)
M(1)-C(2a)
M(2)-C(1a)
M(2)-C(2a)
M(3)-C(27)
M(3)-C(27a)
M(4)-C(40)
M(4)-C(40a)
2.694(1)
2.689(2)
2.691(2)
1.909(5)
1.913(5)
1.924(5)
1.935(5)
1.912(6)
1.928(5)
1.910(6)
1.912(6)
C(3)-M(1)-C(9)
110.1(2)
115.5(2)
112.8(2)
113.6(2)
112.5(2)
91.3(2)
112.3(2)
113.0(2)
114.6(2)
114.9(2)
90.2(2)
Si(1)-C(1)
Si(1)-C(1a)
Si(1)-C(2)
Si(1)-C(8)
1.888(3)
1.881(3)
1.876(2)
1.871(3)
C(3)-M(1)-C(1A)
C(9)-M(1)-C(1A)
C(3)-M(1)-C(2A)
C(9)-M(1)-C(2A)
C(1A)-M(1)-C(2A)
C(15)-M(2)-C(1A)
C(21)-M(2)-C(1A)
C(15)-M(2)-C(2A)
C(21)-M(2)-C(2A)
C(1A)-M(2)-C(2A)
C(1a)-Si(1)-C(1) 92.0(12)
Si(1a)-C(1)-Si(1) 88.1(12)
C(1)-Si(1)-C(2)
C(1)-Si(1)-C(8)
C(2)-Si(1)-C(8)
C(1a)-Si(1)-C(8) 112.7(13) Ge(2)-C(2)-Ge(2a) 90.0(2)
C(1a)-Si(1)-C(2) 113.8(13) C(3)-Ge(1)-C(9)
Ge(1)-Ge(1a)
Ge(2)-Ge(2a)
Ge(1)-C(1)
114.3(13) C(1)-Ge(1)-C(1a)
113.0(13) C(2)-Ge(2)-C(2a)
110.1(10) Ge(1)-C(1)-Ge(1a) 89.1(2)
90.0(2)
110.2(14)
a
The heavy atom positions of the complete independent mol-
ecule are denoted as M(1) and M(2), while the heavy atom positions
of the two independent half-molecules are denoted as M(3) and
M(4) (the corresponding symmetry-generated heavy atom positions
are M(3x) and M(4x), respectively). b The refined occupancy of
silicon at the heavy atom positions M(1) and M(2) are ap-
proximately 0.60 and 0.40, respectively (germanium contributing
the remainder).
2.7704(6) C(1a)-Ge(1)-C(3)
2.7980(7) C(1)-Ge(1)-C(9)
1.980(4)
114.4(2)
112.8(2)
113.7(2)
C(1)-Ge(1)-C(3)
for these systems and is consistent with a negligible
bonding interaction between the silicon centers.
4
4
The head-to-tail connectivity in 3b (Figure 2, right)
parallels that observed for 3a and is in full agreement
with the NMR and mass spectral data obtained for this
compound.21 Compounds 3a and 3b are not isomor-
phous, as evinced by the fact that 3b crystallizes with
two independent molecules per unit cell. As in the case
of 3a , both of the planar, square dimetallacyclic frag-
ments in 3b contain ring atoms possessing distorted
tetrahedral geometries. The transannular Ge-Ge dis-
tances in both of the independent molecules of 3b (dav
∼
2.78 Å) are similar to those previously reported in the
literature for other 1,3-digermacyclobutane deriva-
45,46
tives.
Even though the structure of 3c could be unambigu-
ously assigned on the basis of NMR evidence (vide
supra), it was also characterized by use of X-ray crystal-
lography (Figure 3). To our knowledge, 3c represents
the first example of a crystallographically characterized
(45) Gusev, A. I.; Gar, T. K.; Los', M. G.; Alekseev, N. V. J . Struct.
Chem. 1976, 17, 635.
(46) Zemlyanskii, N. N.; Borisova, I. V.; Bel’skii, V. K.; Kolosova,
N. D.; Beletskaya, I. P. Izv. Akad. Nauk SSSR, Ser. Khim. 1983, 869.
(
44) Savin, A.; Flad, H.-J .; Flad, J .; Preuss, H.; von Schnering, H.
G. Angew. Chem., Int. Ed. Engl. 1992, 31, 185.

Homo- and Cross-[2+2]-Cycloaddition
Organometallics, Vol. 18, No. 26, 1999 5649
molecule containing the 1-germa-3-silacyclobutane struc-
ture. Not unexpectedly, the data show a nearly statisti-
cal occupation disorder at the heavy atom sites in the
three crystallographically independent molecules of 3c
present in the unit cell. However, the remainder of the
structure is well-ordered and yields geometrical data
that are consistent with an average of the related values
in 3a and 3b. In 3c, the average transannular silicon-
germanium distance (∼2.69 Å) agrees well with the
corresponding average obtained from the related dis-
tances in 3a and 3b (dav ∼2.70 Å), but is clearly outside
4
7,48
of the range of a typical Si-Ge σ-bond (∼2.43 Å).
Similarly, this disorder prevents the differentiation of
unique Si-CH2 and Ge-CH2 bond lengths and instead
leads to the observation of an M-CH2 distance (∼1.92
Å) that is close to the average of the Si-CH2 (∼1.88 Å)
and Ge-CH2 (∼1.98 Å) distances found in 3a and 3b,
respectively.
P h otoch em istr y of th e 1,1,3,3-Tetr a p h en yl-1,3-
d im eta lla cyclobu ta n es 3a -c. Steady-state irradia-
tion of deoxygenated solutions of 3a -c in cyclohexane
or cyclohexane-d12 containing 0.5 M methanol led to the
fairly clean formation of methoxymethyldiphenylsilane
F igu r e 4. Transient absorption spectra recorded by 248-
nm laser flash photolysis of optically matched, air-
saturated hexane solutions of germacyclobutane 1b (O) and
digermacyclobutane 3b (0), 0.1-1.0 µs after the laser pulse.
The inset shows a representative decay trace for 3b,
recorded at the absorption maximum.
(4a ) and/or methyldiphenylgermanium methoxide (4b),
indicative of the formation of 2a and/or 2b as the
initially formed products. Qualitatively, the photolysis
of digermacyclobutane 3b appeared to be roughly 3
times more efficient than that of 3c and about 15 times
more efficient than that of disilacyclobutane 3a . The
photolyses of 3b and 3c were clean, producing only the
methoxysilane and/or -germane within the limits of our
GC detection method. In the case of 3a , on the other
hand, the material balance was poor (∼10%), and GC/
MS analysis indicated the coformation of 1,1′-bicyclo-
hexyl in the early stages of the photolysis. This product
is presumably derived from cyclohexyl radicals, further
evidence of which was revealed by the additional forma-
tion of cyclohexanol and cyclohexanone in less com-
pletely deoxygenated solutions. These products are not
formed in detectable yields in the photolyses of the other
two compounds. It is not yet clear how cyclohexyl
radicals are formed in this case, but their formation
would be consistent with the involvement of a radical
or biradical intermediate which is formed in higher
steady-state concentrations in the photolysis of 3a than
in those of the other two compounds.
merry-go-round photolysis of deoxygenated 0.01 M
solutions in cyclohexane containing 0.5 M methanol,
using the procedure described above for 1b (vide supra).
These experiments afforded values of Φ4a ) 0.007 (
0.003 for the disilacyclobutane (3a ), Φ4a +4b ) 0.04 (
0.01 for the germasilacyclobutane (3c), and Φ4b ) 0.11
( 0.02 for the digermacyclobutane (3b).
Similar indications of the relative reactivities of 3a
and 3b toward photoinduced cycloreversion could be
obtained by laser flash photolysis methods. Thus, 248-
nm laser flash photolysis of a continuously flowing, air-
saturated 0.0025 M solution of the digermacyclobutane
in hexane gave rise to transient absorptions that were
less intense but otherwise identical to those obtained
2
1
from similar experiments with 1b and can thus be
assigned to 1,1-diphenylgermene (2b). As was found
with 1b, saturation of the solution with nitrogen had
no discernible effect on the spectrum or second-order
decay kinetics of the transient. Figure 4 shows transient
absorption spectra obtained from optically matched
solutions of 3b and 1b, which illustrate both the
similarities in the transient spectra obtained from the
two compounds and the relative transient quantum
yields. The intensity of the signal obtained from 3b is
roughly half that from 1b, in good agreement with the
relative quantum yields for formation of 2b determined
(as the methanol adduct 4b) by steady-state photolysis
of the two compounds in the presence of methanol.
Similar experiments with the disilacyclobutane (3a )
did not yield detectable transient absorptions assignable
to silene 2a , not surprisingly in light of the quantum
yield determined above by steady-state trapping meth-
ods. Interestingly, flash photolysis of this compound did
yield a transient species whose formation and decay
followed the laser pulse and which exhibited an absorp-
tion spectrum consisting of a single broad band centered
at 540 ( 20 nm. Though it is tempting to speculate that
this species might be the 1,4-biradical derived from Si-
CH2 cleavage in 3a , we have no evidence to support this
Quantum yields for formation of 4a /b from photolysis
of the three dimetallacyclobutanes were determined by
(
47) Pannell, K. H.; Kapoor, R. N.; Raptis, R.; Parkanyi, L.; Fulop,
V. J . Organomet. Chem. 1990, 384, 41.
48) Suzuki, H.; Tanaka, K.; Yoshizoe, B.; Yamamoto, T.; Kenmotsu,
N.; Matuura, S.; Akabane, T.; Watanabe, H.; Goto, M. Organometallics
998, 17, 5091.
(
1

5
650 Organometallics, Vol. 18, No. 26, 1999
Toltl et al.
rived from 1a due to â-silyl interactions.49 One possible
mechanism for such an effect is rapid equilibration of
the reactive excited singlet state of 3a with a much
longer-lived, nonreactive (intramolecular) excimer state,
the presence of which is suggested by the broad emission
to the red of the benzenoid emission band in the
fluorescence spectrum of this compound. This would also
explain the similar fluorescence lifetimes measured for
the two emitting species. Thus, we tentatively conclude
that the substantially lower quantum yield for 1,1-
diphenylsilene (2a ) formation from photolysis of 3a
compared to that from 1a is due to both a reduction in
the partitioning of the excited singlet state between the
corresponding 1,4-biradical (probably small, but cer-
tainly significant) and a reduction in the partitioning
of the biradical between ring cleavage and coupling to
regenerate the starting material. Thermodynamic fac-
tors dictate that the latter is likely to be quite large,
since cycloreversion of 1a yields 2a + ethylene, while
that of 3a yields two molecules of 2a .
Sch em e 1
assignment and hence must defer a full report of the
transient spectroscopic behavior of this compound to a
later date.
Assuming that photocycloreversion of the three di-
metallacyclobutanes proceeds in all three cases by a
stepwise mechanism involving a 1,4-biradical interme-
diate (6; see Scheme 1), then the overall quantum yield
for formation of 2 (ΦMdC) will be given by the product
of two efficiency factors: that for biradical formation
from the reactive excited state of the dimetallacyclo-
butane relative to other excited-state decay processes
We should also expect to find significant differences
in the photophysical behavior of the mono- and diger-
macyclobutanes 1b and 3b. The fluorescence spectra of
the two compounds suggest that indeed there are and,
furthermore, that there are pronounced differences
between the silicon and germanium homologues of both
types of compound. In the case of the monometalla-
cyclobutanes, the spectrum of the germanium compound
is shifted to significantly lower energies than that of
the silicon compound, indicating that a much larger
change in geometry results from excitation of 1b to its
lowest excited singlet state than occurs with 1a , whose
fluorescence spectrum shows almost no Stokes shift
whatsoever. The fluorescence decay of 1b is more
complex than that of 1a , however; it analyses to two
exponentials (τ1 ) 3.0 ( 0.2 ns; τ2 ) 12.8 ( 0.1 ns),
which contribute almost equally to the overall decay.
The emission properties of 3b are even more difficult
to interpret. The spectrum shows no similarity what-
soever to that of 1b, and again, its decay consists of two
components, although in this case there is clearly a
major one which exhibits a lifetime τ ) 1.8 ( 0.1 ns.
Unfortunately, this behavior makes it difficult to be
completely confident of the reliability of our fluorescence
data for the germanium compounds, although they (as
well as those for 1a and 3a ) were obtained from samples
that were >99% pure (by GC), under conditions that
led to <2% substrate photodecomposition during the
course of the experiments. Thus, our data do not provide
(
φbir) and the efficiency with which the biradical cleaves
to metallene relative to regenerating the starting ma-
terial (φMdC; see eqs 4-6). While it is not possible to
determine either of these state efficiencies indepen-
dently in the present cases, it is most likely that
variations in both contribute to the overall variation in
ΦMdC throughout the series. One indication of the likely
variation in the photophysical term (φbir) is provided by
the fluorescence properties of 1a ,b and their dimetalla-
counterparts 3a ,b.
d
i
d
bir
φbir ) kbir/(kbir
+
k ) ) k τs
(4)
(5)
(6)
∑
φMdC ) kMdC/(kMdC + ks.m.
)
ΦMdC ) φ φ
bir MdC
Fluorescence emission spectra were measured for
a ,b at 25 °C in argon-saturated isooctane solution and
3
are shown in the Supporting Information along with
those for 1a ,b . The spectra of the mono- and disila-
cyclobutane both show benzenoid emission bands cen-
tered at 285 nm. In the spectrum of 3a , this is super-
imposed on a broad, less intense band at longer wave-
lengths, which might be ascribable to intramolecular
excimer emission. Fluorescence lifetimes were deter-
mined by time-correlated single photon counting to be
τ ) 2.6 ( 0.1 ns for 1a , in good agreement with our
previously reported value,37 and τ ) 7.5 ( 0.1 ns for
•
any really reliable clues as to how ( MPh2CH2M′Ph2-
•
CH2 ) 1,4-biradical behavior might vary as a function
3
a . Both decays fit acceptably to a single exponential,
of the heteroatom, a question which is of central
importance in our attempts to ascertain the mechanisms
of the head-to-tail dimerization of 2a and 2b.
and lifetimes measured for 3a at emission wavelengths
of 290 and 325 nm did not differ within experimental
error. In the case of 1a , comparison of its singlet lifetime
to that of dimethyldiphenylsilane suggests that Si-C
bond cleavage in the lowest excited singlet state occurs
at a comparable rate to those of the other processes that
contribute to excited-state decay (fluorescence, internal
Su m m a r y a n d Con clu sion s
1,1-Diphenylsilene (2a ) and 1,1-diphenylgermene (2b)
undergo head-to-tail [2+2]-dimerization in hydrocarbon
solution with absolute rate constants that are surpris-
ingly similar to one another and are within a factor of
∼2 of the diffusional rate. The data suggest that in both
cases the reaction proceeds with a very low activation
3
7
conversion, etc.). The significantly longer singlet life-
time observed for 3a suggests that the second silicon
atom slows this process down (hence reducing φbir),
•
•
although the resulting ( SiPh2CH2SiPh2CH2 ) 1,4-bi-
radical might be expected to enjoy enhanced stabiliza-
•
•
tion relative to the ( SiPh2CH2CH2CH2 ) biradical de-
(49) J arvie, A. W. P. Organomet. Chem. Rev. A 1970, 6, 153.

Homo- and Cross-[2+2]-Cycloaddition
Organometallics, Vol. 18, No. 26, 1999 5651
barrier, the magnitude of which cannot be significantly
greater than the activation energy for diffusion in these
solvents (ca. 2 kcal/mol). For 2a , this agrees well with
for carbon. Infrared spectra were recorded on a BioRad FTS-
4
0 FTIR spectrometer and are reported in wavenumbers.
Fluorescence emission spectra and singlet lifetimes were
determined on a Photon Technologies Inc. LS-100 spectrom-
eter, which provides both steady-state and time-resolved (time-
correlated single photon counting) capabilities. Samples for
fluorescence experiments were contained in Suprasil quartz
cuvettes, deoxygenated with a stream of dry argon, and sealed
with Teflon tape.
Reverse phase HPLC separations were carried out using a
Hewlett-Packard 1090 liquid chromatograph equipped with a
built-in diode-array UV detector and a 5 µm Vydac reversed
phase column (25 cm × 4.6 mm i.d., Separations Group,
Heseria, CA). The HPLC was operated with a column tem-
perature of 40 °C, a flow rate of 1.0 mL/min, and a monitoring
wavelength range of 250-274 nm. The separation employed
a gradient elution with water-acetonitrile mixtures (30%
water f 81% water at 16 min f 90% water at 20 min). Under
these conditions, ca. 1-mg samples of a mixture of the three
dimetallacyclobutanes could be separated cleanly into the
three components.
Analytical gas chromatographic analyses were carried out
using a Hewlett-Packard 5890 gas chromatograph equipped
with a flame ionization detector, a Hewlett-Packard 3396A
recording integrator, conventional heated splitless injector, and
a DB-1 fused silica capillary column (15 m × 0.20 mm;
Chromatographic Specialties, Inc.). Semipreparative GC sepa-
rations employed a Varian 3300 gas chromatograph equipped
with a thermal conductivity detector and a 6 ft × 0.25 in.
stainless steel OV-101 packed column (Chromatographic
Specialties). Radial chromatographic separations employed a
Chromatotron (Harrison Research, Inc.), 2- or 4-mm silica gel
60 thick-layer plates, and hexane/dichloromethane mixtures
as eluant.
recent theoretical predictions for the dimerization of the
parent molecule1
0-13
and is consistent with a previous
estimate of the Arrhenius activation energy for dimer-
1
7
ization of 1,1-dimethylsilene in the gas phase. Given
the similarity in the rate constants for dimerization of
2
a and 2b, it seems very likely that the dimerization of
naturally polarized silenes and germenes proceeds by
the same mechanism. Mechanistic commonalities have
been documented previously in the reactions of 2a and
2
b with nucleophilic reagents, although in these reac-
tions the germene is generally much (10-1000 times)
less reactive than the silene.4
The head-to-tail dimer of 2a sdisilacyclobutane 3a s
undergoes inefficient cycloreversion to the silene upon
photolysis in solution in the presence of methanol. The
quantum yield for photocycloreversion is 4-5 times
lower than that of the homologous 1-sila-3-germa-
cyclobutane (3c) and about 15 times lower than that of
1
,3-digermacyclobutane 3b . This variation in photo-
reactivity is quite remarkable, considering that the
monometallacyclobutanes 1a ,b cyclorevert with equal
quantum efficiencies under similar conditions. These
reactions most likely proceed via excited-state M-C
•
•
bond cleavage to yield the corresponding M-C-M′-C
biradical, which undergoes competing reclosure and
fragmentation to the metallaene(s). They thus provide
an entry point into the potential energy surface for
nonconcerted silene/germene dimerization. Unfortu-
nately however, the overall quantum yields for photo-
cycloreversion (ΦMdC) give an incomplete picture of the
variation in biradical reactivity throughout the series.
This is because ΦMdC is given by the product of two
efficiency factors: that for biradical formation from the
lowest excited singlet state (φbit) and that for cleavage
of the biradical (φMdC). These efficiency factors cannot
be extracted on the basis of the data we have available,
and thus we are not yet able to rule out any of the three
distinct mechanisms for the head-to-tail dimerization
of silenes that have been previously suggested: con-
Acetonitrile (Caledon HPLC) was used as received from the
supplier. Hexane and isoctane (BDH Omnisolv) were used as
received or distilled from sodium. 1,1-Diphenylsilacyclobutane
(
1b) and 1,1-diphenylgermacyclobutane (1b) were prepared
2
1,25
and purified as previously described.
Preparative irradiations of 1a ,b employed a Rayonet Pho-
tochemical Reactor (Southern New England Ultraviolet Co.)
equipped with 8-10 RPR-2537 lamps. Solutions of the met-
allacyclobutanes (0.25 g) were dissolved in hexane (15 mL),
sealed in a quartz photolysis tube with a rubber septum,
deoxygenated with dry argon, and irradiated to 20-50%
conversion of starting material (6-10 h). The dimers (3a ,b)
were separated from residual starting material by radial
chromatography and recrystallized from pentane. Their melt-
ing points (3a , mp 135-136 °C; 3b, mp 120-121 °C) and
spectroscopic data matched the previously reported data in
1
0
certed [π2s+π2s]-cycloaddition, stepwise involving a
•
•
11,12
•
Si-C-Si-C biradical,
or stepwise involving a C-
•
13
Si-Si-C biradical.
Future work in this area will be directed at defining
the effects of temperature, solvent, and substituents on
the absolute rate constants for dimerization of 1a and
1
9,21
both cases.
The following spectroscopic data for 3a have
not been reported previously: 1H NMR, δ ) 1.10 (s, 4H), 7.31-
.36 (m, 12H), 7.50-7.56 (m, 8H); C NMR, δ ) 0.85, 127.88,
29.42, 134.29, 137.37; IR (neat), 3065.7 (m), 3018.0 (m),
999.5 (m), 2924.2 (m), 1428.0 (s), 1375.6 (m), 1344.4 (m),
13
7
1
2
1
b in solution, as well as investigating in a more direct
•
•
way the behavior of M-C-M-C biradicals as a func-
tion of the heteroatom.
1262.6 (m), 1110.4 (s), 1064.9 (s), 1024.7 (s), 936.6 (m); MS,
m/e (I) ) 392 (5), 377 (7), 314 (40), 301 (40), 257 (20), 237 (20),
2
23 (20), 181 (30), 157 (25),105 (100), 91 (10),79 (10); HRMS,
Exp er im en ta l Section
2
calcd for C26H24Si , 392.1408; found 392.1416.
1H and 13C NMR spectra were recorded on Bruker AC200
or DRX500 NMR spectrometers in deuterated chloroform
solution and are referenced to tetramethylsilane. Ultraviolet
absorption spectra were recorded on Hewlett-Packard HP8451
or Perkin-Elmer Lambda 9 spectrometers. Low-resolution
mass spectra and GC/MS analyses were determined using a
Hewlett-Packard 5890 gas chromatograph equipped with a
HP-5971A mass selective detector and a DB-5 fused silica
capillary column (30m × 0.25 mm; Chromatographic Special-
ties, Inc.). High-resolution desorption electron impact (DEI)
mass spectra and exact masses were recorded on a VGH ZABE
mass spectrometer. Exact masses employed a mass of 12.000000
1,1,3,3-Tet r a p h en yl-(1-sila -3-ger m a )cyclob u t a n e (3c)
was prepared by photolysis of a deoxygenated solution of 1a
(0.24 g, 1.07 mmol) and 1b (0.27 g, 1.0 mmol) in hexane (20
mL), under conditions similar to those described above. The
photolysis was stopped after ∼40% conversion of the two
starting materials, at which time capillary GC analysis
indicated the mixture consisted of (in order of increasing
retention time) 1a (50%), 1b (19%), 3a (2.4%), 3c (7.7%), and
3b (6.4%). Radial chromatography of the crude reaction
mixture after evaporation of the residual solvent allowed
separation of 1a ,b from the three dimetallacyclobutanes, which
eluted together and were hence isolated as a mixture. The

5
652 Organometallics, Vol. 18, No. 26, 1999
Toltl et al.
three compounds were successfully separated by reverse phase
HPLC, after which 3c was recrystallized from pentane to yield
colorless prisms (mp 98-99 °C). The compound exhibited the
following spectroscopic and analytical data: 1H NMR, δ ) 1.31
the program SMART) and a rotating anode using graphite-
monochromated Mo KR radiation (λ ) 0.71073 Å). Data
processing was carried out by use of the program SAINT, while
the program SADABS was utilized for the scaling of diffraction
data, the application of a decay correction, and an empirical
absorption correction based on redundant reflections. The
structures were solved by using the direct methods procedure
in the Siemens SHELXTL program library and refined by full-
(
s, 4H), 7.31-7.36 (m, 12H), 7.46-7.51 (m, 4H), 7.52-7.59 (m,
13
4
1
2
1
H); C NMR, δ ) 3.75, 127.86, 128.22, 128.99, 129.37, 133.72,
34.19, 137.82, 138.97; IR (neat), 3070.2 (m), 2965.5 (m),
858.3 (m), 1464.1 (m), 1425.8 (m), 1382.1 (m), 1259.1 (m),
112.2 (s), 1069.8 (s), 1022.5 (s), 846.7 (w); MS, m/e (I) ) 438
2
matrix least-squares methods on F with anisotropic thermal
parameters for all non-hydrogen atoms. Hydrogen atoms were
added as fixed contributors at calculated positions, with
isotropic thermal parameters based on the carbon atom to
which they are bonded. In the crystal, 3a is situated on a
crystallographic inversion center, resulting in the observation
of only half a molecule per asymmetric unit. A similar scenario
exists for 3b; however the asymmetric unit consists of two
crystallographically independent half-molecules. For 3c, the
asymmetric unit was found to contain two half-molecules and
one complete molecule, giving rise to a total of four molecules
per unit cell. Given the asymmetry in the 3c molecular
framework, equal (50:50) occupancy of silicon and germanium
in the heavy atom positions of the final refined structure for
each of the two crystallographically independent half-mol-
ecules was assigned. However, when the distribution of silicon
and germanium over the two heavy atom sites in the complete
independent molecule was allowed to refine as a free variable
(with the coordinates and thermal parameters of the disor-
dered heavy atoms at each site constrained to refine to common
values), an occupancy ratio of approximately 60:40 was found
and subsequently utilized in the final refined model.
(
1
5), 423 (5), 360 (35), 347 (50), 259 (100), 226 (10), 209 (10),
95 (15), 181 (15), 151 (15), 105 (20), 91 (10),79 (5); HRMS,
calcd for C26 24SiGe, 438.0834; found 438.0859.
H
Nanosecond laser flash photolysis experiments employed the
pulses (248 nm; ca. 16 ns; 120-160 mJ ) from a Lambda-Physik
Compex 120 excimer laser filled with F
a microcomputer-controlled detection system.
were prepared at concentrations such that the absorbance at
the excitation wavelength (248 nm) was ca. 0.7 (2.5 × 10 M)
and were flowed continuously through a 3 × 7 mm Suprasil
flow cell connected to a calibrated 100-mL reservoir.
2
/Kr/He mixtures and
2
1,50
Solutions
-
3
Dimerization rate coefficients were calculated by nonlinear
least-squares fitting of ∆OD versus time data for 2a using eq
7
, the standard expression for mixed pseudo-first-order and
5
1
second-order decay, in which l is the path length (3 mm) and
is the pseudo-first-order component of the decay. The
k
1
4
-1
analyses afforded k
traces for 2b were more predominantly second order, but plots
of 1/(∆OD) versus time deviated significantly from linearity.
Fitting the data to eq 7 gave statistically acceptable results,
but k values were negative, i.e., characteristic of a pseudo-
first-order growth to a relatively long-lived residual absorption.
While the residual absorption is real, it is formed within the
laser pulse, and hence the indication of a first-order growth
1
values on the order of 10 s . The decay
t
1
Ack n ow led gm en t. We thank D. Hughes for his
assistance with high-resolution NMR spectra, the Mc-
Master Regional Centre for Mass Spectrometry for high-
resolution mass spectra and exact mass determinations,
Professor Brian McCarry for his advice and assistance
with chromatographic separations, and Professor Mark
Steinmetz for helpful comments. The financial support
of the Natural Sciences and Engineering Research
Council (NSERC) of Canada is also gratefully acknowl-
edged. M.S. and T.L.M. thank NSERC for postgraduate
scholarships.
2
1
can be considered an artifiact. Thus, the data were finally
analyzed according to the more appropriate expression given
in eq 8, where ∆OD
OD did not exceed 10% of the initial signal intensity in any
case, consistent with prior experience.
∞
is the value of the residual absorption.
∆
∞
2
1
(
∆OD )/(∆OD ) ) k exp(-k t)/[k + (2k (∆OD )/
t 0 1 1 1 dim 0
lꢀ3
25-nm
)(1 - exp(-k t))] (7)
1
Su p p or tin g In for m a tion Ava ila ble: Tables of atomic
coordinates, bond lengths and angles, anisotropic displacement
parameters, and hydrogen atom coordinates for 3a -c, UV
absorption and fluorescence emission spectra of 1a ,b and 3a ,b,
(
∆OD )/(∆OD ) ) (∆OD )/(∆OD ) +
t
0
∞
0
1
/[1 + (2kdim(∆OD )/lꢀ325-nm)t] (8)
0
0
and plots of ∆OD versus laser intensity from flash photolysis
of optically matched solutions of benzophenone, 1a and 1b,
from which the extinction coefficients of 2a ,b were calculated.
This material is available free of charge via the Internet at
http://pubs.acs.org.
X-r a y Cr ysta llogr a p h y. Crystallographic data52 for 3a , 3b,
and 3c were collected from suitable crystals which were grown
from pentane and mounted on a glass fiber. The instrument
used for the data collection was a P4 Siemens diffractometer
equipped with a Siemens SMART 1K CCD area detector (using
OM990447K
(52) (a) SMART, Release 4.05; Siemens Energy and Automation
Inc.: Madison, WI, 1996. (b) SAINT, Release 4.05; Siemens Energy
And Automation Inc.: Madison, WI, 1966. (c) Sheldrick, G. M. SADABS
(Siemens Area Detector Absorption Corrections); 1996. (d) Sheldrick,
G. M. Siemens SHELXTL, Version 5.03; Siemens Crystallographic
Research Systems: Madison, WI, 1994.
(
50) Leigh, W. J .; Workentin, M. S.; Andrew, D. J . Photochem.
Photobiol. A: Chem. 1991, 57, 97.
51) Espenson, J . H. Chemical Kinetics and Reaction Mechanisms,
nd ed.; McGraw-Hill: New York, 1995.
(
2