Comparative chemistry of η3-oxaallyl and η3-allyl rhodium(I) complexes in the hydrosilylation of cyclopropyl ketones: Observation of an unprecedented rearrangement

Article abstract of DOI:10.1016/S0022-328X(99)00041-8

Addition of η³-oxaallyl bis(triphenylphosphine)rhodium(I) (1) or η³-allyl bis(triphenylphosphine)rhodium(I) (2) to a mixture of cyclopropyl phenyl ketone and triethylsilane promotes two competitive catalytic reactions. Complex 1 give

Full text of DOI:10.1016/S0022-328X(99)00041-8

www.elsevier.nl/locate/jorganchem
Journal of Organometallic Chemistry 582 (1999) 235–245
Comparative chemistry of h3-oxaallyl and h3-allyl rhodium(I)
complexes in the hydrosilylation of cyclopropyl ketones:
observation of an unprecedented rearrangement
Carrie Sun a, Ambrose Tu a, Greg A. Slough b,*
a Department of Chemistry, Case Western Reser6e Uni6ersity, Cle6eland, OH, USA
b Chemistry Department, Kalamazoo College, Kalamazoo, MI 49006, USA
Received 10 September 1998
Abstract
Addition of h3-oxaallyl bis(triphenylphosphine)rhodium(I) (1) or h3-allyl bis(triphenylphosphine)rhodium(I) (2) to a mixture of
cyclopropyl phenyl ketone and triethylsilane promotes two competitive catalytic reactions. Complex 1 gives high yields of the silyl
ether 13, while complex 2 favors an unprecedented cyclopropyl carbinyl ring-opening reaction providing good yields of enol silane
(Z)- and (E)-14 (4.1 Z/E ratio). Studies of catalyst turnover provide evidence that two monomeric rhodium complexes account
for the competing reactions. The cyclopropyl carbinyl catalysis occurs under a narrow reaction regime with a specific set of ketone
substrates suitable for the reaction. Chemicals such as triethylsilane activate the catalysis leading to 13, while triphenylphosphine
strongly inhibits the formation of 14. A mechanism involving a polar transition state for the cyclopropyl carbinyl rearrangement
is proposed. © 1999 Elsevier Science S.A. All rights reserved.
Keywords: Kinetics of hydrosilylation; Cyclopropyl carbinyl rearrangement; h3-Allyl rhodium; h3-Oxaallyl rhodium
The effect of s-bound ligands on rhodium catalysts is
only beginning to emerge. Wilkinson’s catalyst
((PPh3)3RhCl) and bis(phosphine)rhodium cations are
classic catalysts for the hydrosilylation of organic car-
bonyls, recently however, hydrido rhodium(I) com-
plexes were shown to be very active catalysts in the
same reaction [4]. Particularly noteworthy is the obser-
vations of Itoh, that organosilanes having two tethered
Si–H groups react with Wilkinson’s catalyst giving a
Rh(V) trihydrido-bis(silyl) complex [4c]. Itoh proposed
that the s-bound chloro ligand exchanges during cata-
lyst activation, and that the resulting hydrido rhodium
material is orders of magnitude more reactive. Further-
more, both putative mechanisms suggested in Itoh’s
paper involve bis(hydrido)-silyl rhodium(III) reaction
intermediates.
1. Introduction
The development of an homogeneous catalyst for
selective organic transformations is a major objective of
chemical research. The hydrosilylation of organic car-
bonyls is an interesting case study that illustrates many
of these goals. The quest for selective and efficient
hydrosilylation catalysts remains open, and recent stud-
ies have taken two different approaches [1]. One ap-
proach examined the influence of ancillary phosphine
ligand basicity on catalyst activity, while the second
approach began investigating the affect of s-bound
ligands on rhodium(I) catalysts. Wrighton [2] demon-
strated that oxidation of the 1,1%-bis(diphenylphos-
phino)cobaltocene ancillary ligand enhanced the
catalytic properties of rhodium(I) complexes towards
the reductive hydrosilylation of ketones. This result fits
into a larger class of reactions in which the donicity of
‘electronically tuned’ ligands affects the reactivity and
selectivity of catalyzed reactions [3].
Taken together, these new catalyses revolve around
the exchange of an Rh–Cl bond for an Rh–H bond,
with the hydrido–rhodium complex being a more active
catalyst. We proposed that the Cl for H-ligand ex-
change can be circumvented through the use of h3-allyl
or h3-oxaallyl rhodium complexes as catalyst precur-
* Corresponding author.
0022-328X/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.
PII: S0022-328X(99)00041-8

236
C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
2. Results and discussion
sors. Marder recently demonstrated that allylrhodium
complexes were highly selective catalyst precursors in
the mechanistically related hydroboration of alkenes
[5]. Work in our laboratory showed that a variety of
h3-oxaallyl rhodium(I) complexes (1) reacted under a
wide range of conditions, including organosilane, and
that a rhodium(I) hydride (4) is available for catalysis
[6]. We reasoned that h3-ligand elimination could be
probed by conducting parallel experiments with the
isoelectronic h3-allyl rhodium complex (2) [7]. Elec-
tronic features of the h3-allyl ligand relative to the
h3-oxaallyl ligand would alter the reaction with
organosilanes, and the production of a trialkylsilyl-
rhodium(I) (3) catalyst would be anticipated. Work by
Milstein [8], Thorn [9], and Wrighton [10], strongly
supported this plan, in that, Group VIII alkyl–metal
complexes reacted with organosilane preferentially
eliminating alkane (R–H) relative to tetraalkylsilane
(R–Si).
Addition of the h3-oxaallyl rhodium complex 1
(2.5×10−3 mmol) to a concentrated solution of cyclo-
propyl phenyl ketone (3.5 M) and triethylsilane (3.5 M)
resulted in consumption of cyclopropyl phenyl ketone
when heated to 60.0°C. The reaction mixture showed
formation of the expected triethylsilyl ether (13) and a
combined 3% yield of E- and Z-enol silane ((E)-14,
(Z)-14) ([13,14], Fig. 1). The same reaction conducted
with the h3-allyl rhodium complex 2 (2.5×10−3 mmol)
significantly enhanced the production of 14 (Z/E ratio
4:1) to ca. 53% of the mixture.
We wish to report that isoelectronic h3-oxaallyl and
h3-allyl rhodium(I) complexes react with triethylsilane
and act as catalyst precursors for the hydrosilylation of
phenyl cyclopropyl ketone. The incipient rhodium com-
plex from h3-allyl bis(triphenylphosphine)rhodium(I)
catalyzes an unprecedented reductive arrangement of
cyclopropyl ketones [11,12], while the h3-oxaallyl com-
plex favors 1,2-carbonyl reduction.
These data indicate that two active catalysts formed
in the initial phase of the reaction (see Scheme 1).
Polarization of the oxaallyl ligand kinetically favored
the elimination of enolsilane, 7, while complex 2 gave a
higher fraction of propene. To test these competing
processes we examined the reaction of rhodium com-
plexes 1 and 2 with triethylsilane. In each experiment
Fig. 1. The hydrosilylation of cyclopropyl phenyl ketone catalyzed by complex 1 (chromatogram A) and complex 2 (chromatogram B),
respectively. Both chromatograms show ca. 20% of the catalyzed reaction.

C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
237
Scheme 1.
we analyzed the organic products [15]. Treatment of 1
with two equivalents of triethylsilane in C6D6 followed
by heating to 60°C produced two sets of signals in the
1H-NMR. The major set, integrating to 73% of the
mixture, was a pair of olefinic doublets (J=1.6 Hz) at
4.87 and 4.45 ppm. These NMR signals matched enol
silane 7. The second organic product showed a diagnos-
tic quartet centered at 4.78 ppm. The chemical shift and
multiplicity of this signal were not anticipated; how-
ever, the reductive elimination of acetophenone gener-
ated a carbonyl substrate suitable for hydrosilylation
reduction. Reduction of the acetophenone under the
ambient conditions resulted in 1-phenyl-1-triethyl-
siloxyethane [16]. The reaction of complex 2 with tri-
ethylsilane (two equivalents) under the same conditions
produced both propene and allyltriethylsilane. Gas–liq-
uid partitioning made GC–MS analyses quite variable,
however, we concluded that similar amounts of each
chemical arose in the reaction. This conclusion is sup-
ported by evidence found in the reaction of
tris(trimethylphosphine)(methyl)rhodium(I) and tri-
ethylsilane in which nearly equal amounts of methane
and methyltriethylsilane were generated [9]. Taken to-
gether, complex 1 proceeded through complex 4 selec-
tively, while complex 2 gave similar quantities of 3 and
4.
14 was linear. Competitive production of silyl ether 13
in each reaction never showed linearity (see Fig. 2),
therefore its production can only be described
qualitatively.
Three sets of experiments were examined to evaluate
the competing catalyses. First, we established that the
production rate of 14 followed directly with the concen-
tration of complex 2 (see entries 1–3 in Table 1). When
normalized for complex
2 concentration, similar
turnover rates were observed for each reaction. This
observation implied that the active catalyst leading to
14 was monomeric. Low catalyst concentrations de-
crease the possibility of dimerization or oligomeriza-
tion, yet the normalized turnover rate of each
experiment remained constant. Linearity of the
turnover rate decayed after consumption of the first
20% of ketone, and distinct curvature was observed
[18]. Below 2.75 M ketone, the rhodium catalyst is no
longer saturated and at least two catalytic steps affected
the turnover rate. By looking at the total turnover
numbers at any time, one can compare the relative rates
of Path A and Path B. For example, in the experiment
leading to Entry 1 in Table 1, the turnover number for
the production of 13 was slightly smaller than that
leading to 14 (see Fig. 2).
Similar results were observed with oxaallyl complex 1
as the catalyst precursor except that the production of
silylether 13 was enhanced (see Fig. 1 and Entries 4,5 in
Table 1). Analysis of enol silane data revealed that the
production of 14 decreased ca. 13 times relative to
reactions catalyzed by complex 2. Two factors con-
tributed to this decrease. First, the selective generation
of hydridorhodium 4 led to an active five-coordinate
catalyst, 6, which promoted carbonyl reduction. The
enhanced production of 13 efficiently drained cyclo-
3. Catalysis study
To further characterize the difference in catalyses
brought about by complexes 1 and 2, we examined the
pseudo-zero order turnover rates for the reaction at
60.0°C. At high ketone concentration (3.5 M) [17], the
maximal initial turnover rate for the total production of

238
C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
A second set of experiments focused on triethylsilane
concentration effects. Prior studies demonstrated that
silane concentration affected both reaction rates and
partitioning of products in the catalyzed hydrosilylation
of ketones [19]. It has been postulated that an associa-
propyl phenyl ketone away from Path B lowering its
reaction rate. Since both factors contributed to the
relative production of 13 and 14, the amount of
rhodium complexes 3 and 4 (subsequently 5 and 6)
present during catalysis could not be evaluated.
Fig. 2. Initial turnover rate studies for the reaction of ketone 9 with triethylsilane catalyzed by complex 2. The reaction additive is highlighted for
each set of experiments.

C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
239
Table 1
Pseudo-zero-order turnover rates for the production of enol silane
Entry
Complex (M×102)
Ketone (M)
Triethylsilane (M)
Enol silane (14) turnover mmol (mmol cat.*min)−1
1
2
3
4
5
6
7
2 (2.50)
2 (1.26)
2 (0.63)
1 (2.50)
1 (1.27)
2 (2.50)
2 (2.50)
9 (3.50)
9 (3.50)
9 (3.50)
9 (3.50)
9 (3.50)
9 (3.50)
10 (3.49)
(3.50)
(3.50)
(3.50)
(3.50)
(3.50)
(1.74)
(3.50)
0.17790.005
0.16990.005
0.17890.005
0.01190.003
0.01490.002
0.14090.011
0.06090.004
this agent. Conversely, a cationic rearrangement re-
quires both a polar reaction media and a coordination
site at the metal center to accommodate the incoming
alkyl ligand. We anticipated that a less polar solvent
such as benzene and a strong coordinating ligand such
as triphenylphosphine might decrease the production of
14. The solvent choice was fortuitous since both
galvinoxyl radical and triphenylphosphine dissolved
sparingly in the neat ketone/triethylsilane mixture. Ben-
zene completely dissolved the catalytic mixture. The
dilute reaction mixture was examined and the maximal
turnover rate for the formation of 14 with no additives
was linear through the first 5–6 turnovers (see Entry 1,
Table 2). Interestingly, the turnover rate for the produc-
tion of 13 in this modified reaction media decreased
slightly, while the turnover rate for 14 decreased ca. 3.5
times.
Introduction of 0.0040 M galvinoxyl radical (0.33
equivalents relative to 2) slowed the turnover rate
leading to 14 by ca. 17%. Higher concentrations of
galvinoxyl radical decreased the turnover rate
marginally to a point such that at one equivalent, no
further decrease was observed (see Entries 2–5, Table
2). At the two highest concentrations of galvinoxyl
radical, the turnover rate remained ca. 60% of the
original rate. The catalysis leading to 13 showed a
similar galvinoxyl radical effect (see Fig. 2). For
instance, the turnover rate decreased when 0.0040 M
galvinoxyl radical was added and then decreased more
until nearly a stoichiometric amount relative to 2 was
tive mechanism for the final reductive elimination step
leading to trialkylsilyl ethers is important for proper
catalyst function. Although the association of silane in
the final step is not well understood, such a process
prevents formation of a three-coordinate rhodium com-
plex. In our study, we found that triethylsilane signifi-
cantly favored the production of 13, but negligibly
impacted the formation of 14. For example, decreasing
the concentration of triethylsilane by one-half resulted
in a turnover rate for 14 only slightly less than the
maximal turnover (see Entry 6, Table 1). However, the
same concentration change significantly slowed the pro-
duction of 13. For instance, the time required for the
formation of 0.2 mmol of 13 increased from 60 to 138
min. These results indicated that triethylsilane was in-
volved in the rate-limiting step leading to 13, while the
rate-limiting reaction producing 14 was independent of
triethylsilane. We tested the synthetic importance of this
observation by infusing triethylsilane into the reaction
mixture with a syringe pump. Slow addition of triethyl-
silane increased the chromatographic yield of 14 to 81%.
Inhibition studies of the competing catalyses com-
prised the third aspect of our study. We reasoned that
the cyclopropyl ring-opening reaction (Path B) occurred
by either a free radical pathway or through a cyclo-
propyl carbinyl cation rearrangement. To distinguish
these mechanistic alternatives, a radical trapping agent
such as galvinoxyl radical should decrease the turnover
rate of pathway B if the reaction operates by way of a
free radical. An ionic reaction should be impervious to
Table 2
Inhibition of the hydrosilylation reaction catalyzed by complex 2
Entry Complex (M× Ketone (M)
Triethylsilane
(M)
Additive (M×102) Enol silane (14) turnover mmol (mmol cat.*min)−1
102)
1
2
3
4
5
6
7
8
2 (1.25)
2 (1.25)
2 (1.25)
2 (1.25)
2 (1.25)
2 (1.25)
2 (1.25)
2 (1.25)
9 (1.75)
9 (1.75)
9 (1.75)
9 (1.75)
9 (1.75)
9 (1.75)
9 (1.75)
9 (1.75)
(1.75)
(1.75)
(1.75)
(1.75)
(1.75)
(1.75)
(1.75)
(1.75)
–
0.04890.0002
Galvinoxyl (0.40) 0.04090.0003
Galvinoxyl (0.95) 0.03390.0002
Galvinoxyl (1.25) 0.03090.0003
Galvinoxyl (2.50) 0.03090.0002
PPh3 (0.19)
PPh3 (0.63)
PPh3 (0.95)
0.02790.0003
0.01190.0002
0.005990.0003

240
C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
added. These results suggested that galvinoxyl radical
reacted with rhodium complexes in the reaction mixture
generating materials that were less efficient at catalyz-
ing the reduction and ring-opening reactions. However,
we cannot exclude the possibility that the galvinoxyl
radical–rhodium complex interaction is reversible and
that these experiments follow saturation kinetics. The
latter explanation is appealing since the production of
13 and 14 continued even in the presence of radical
trapping agent. It is surprising however that saturation
occurs close to the stoichiometric balance. Nevertheless,
galvinoxyl radical is not a strong inhibitor of either
reaction Path A or B [20c,e].
By contrast, triphenylphosphine had a powerful in-
hibitory affect on the product of 14. Addition of 1.0 mg
of triphenylphosphine to a catalytic mixture decreased
the turnover rate by ca. 44% (see Table 2). Adding 5.0
mg of the triphenylphosphine to a similar mixture
decreased the catalytic turnover rate to ca. 12% of the
maximal. Interestingly, triphenylphosphine showed lit-
tle or no effect on the formation of 13. At the highest
phosphine concentration, there appeared to be a brief
induction period followed by a rapid reaction (see Fig.
2). This behavior is similar to hydrosilylation reactions
catalyzed by Wilkinson’s catalyst and hydridotetrakis(t-
riphenylphosphine)rhodium(I) in that dissociation of
triphenylphosphine must occur to open a coordination
site at the catalytic center [5,20a].
5. Summary and conclusions
The competitive nature of reaction Paths A and B
suggested that two similar catalytic species formed
initially from complexes 1 and 2. The strength of the Si–O
(786 kJ mol−1) bond facilitated the reductive elimination
of 1-phenyl-1-triethylsilyloxyethene from complex 1.
When complex 1 was used, a high yield of silyl ether 13
resulted. We propose that a predominance of hydrido
rhodium(I) (4) formed and this complex catalyzed the
reduction of the ketone carbonyl [7a]. Reaction mixtures
catalyzed by complex 2 generated nearly equal amounts
of silyl ether 13 and 14 (E and Z) suggesting that a second
catalytic species was responsible for the ring-opened
products. We believe that the second catalytic species is
a triethylsilyl rhodium(I) complex (3). The difference in
size and electronic character of the hydride ligand and
the triethylsilyl ligand accounts for the bifurcated
catalytic behavior. Neither species remains coordinately
unsaturated in the reaction mixture and rapidly converts
to reactive rhodium(III)-bis(hydride)triethylsilyl complex
or a rhodium(III)-hydride-bis(triethylsilyl) complex.
Once the two respective rhodium(III) species form,
each inserts the ketone carbonyl through a silicon-first
mechanism. Such an addition is believed to be the first
step in the hydrosilylation reaction catalyzed by
Wilkinson’s catalyst, [20] and this step explains the rate
effect observed with ketones 9, 10 and 11. Substituents
in the para position of the phenyl group also influence
the rhodium–carbon bond strength in the a-trie-
thylsiloxy alkyl rhodium intermediates (21 and 22). This
situation is manifest in the noted product ratios observed
with cyclopropyl p-methoxyphenyl ketone and cyclo-
propyl phenyl ketone. The ring-opened product became
dominant with ketone 10. Other mechanisms such as
direct rhodium insertion into the cyclopropane ring fail
to account for the observed rate effects [11d].
The differentiating step in the competing catalyses in-
volves the transformation of a-triethylsiloxy alkyl rhod-
ium intermediates 21 and 22. As depicted in Scheme 2,
reaction Path A transforms directly into the silyl ether
13. High triethylsilane concentration facilitates loss of 13
by consuming coordination sites generated during the
reductive elimination step [20c,22]. Since no coordinative
unsaturation is exposed, this catalysis should not be in-
hibited by triphenylphosphine. Similarly, galvinoxyl
radical may have a modest inhibitory effect since complex
6 is its most likely reaction partner. Complex 6 is
competitively consumed in the carbonyl insertion
reaction, therefore, saturation kinetics should be
anticipated.
Reaction Path B, on the other hand, shows no
triethylsilane dependence since the rate-determining step
is presumably the molecular rearrangement. The key
cyclopropyl carbinyl rearrangement has been observed in
a cyclopropylmethylcobalamin, however, the mechanis-
tic course could not be identified [23]. The cobalamin
4. Substrates for competitive catalysis
Other cyclopropyl ketones were investigated using
complex 2 as catalyst, however, aryl-cyclopropyl ke-
tones occupied a unique niche in this catalytic scheme.
Three aryl-cyclopropyl ketones were examined and all
gave different results. For example, cyclopropyl p-
methoxy-phenyl ketone (10) reacted slowly relative to
ketone 9 but gave both ring-opening and reduction
products (see Table 1). With ketone 10, the silyl ether
16 yield was ca. half that of enol silane 15 (Z/E ratio
4:2). On the other hand, cyclopropyl p-trifl-
uoromethylphenyl ketone reacted rapidly within 2 h
and ultimately gave a 66% isolated yield of silyl ether
17. A trace amount of the ring-opened enol silanes were
observed in the gas chromatogram. These three aryl
ketones followed a general trend in which electron-
withdrawing groups on the ketone increased the rate of
hydrosilylation [21].
Finally, cyclopropyl methyl ketone proved to be a
very poor substrate for either catalysis. The aliphatic
ketone reacted slowly, and after 10 h of heating only
about 20% of the ketone was consumed. The dominant
product, albeit in low yield, was silyl ether 18. A trace
amount of ring-opened enol silane was observed by gas
chromatography.

C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
241
complex rearranged at ca. the same rate in solvents of
different dielectric constants, and the authors concluded
that the reaction did not occur by an ionic pathway. In
our case, there was a measurable solvent effect when
benzene was added to the reaction mixture.
or an ion-pair mechanism. In each, charge develops at
the transition state, and the reaction media imposes some
influence on the reaction rate. Excess electron density at
rhodium makes possible an internal nucleophilic attack
at the g-carbon of the cyclopropane ring [25,26]. As
shown in Scheme 2, a conformer of the 22 brings a
g-carbon into proximity for nucleophilic attack. As the
rearrangement occurs, steric interactions decrease by
moving the bulky rhodium complex to a primary carbon
and torsional strain in the cyclopropane ring is released.
Despite the numerous advantages of the molecular
rearrangement, this transformation is rate-determining.
Recall that Path B is independent of triethylsilane
concentration, is seems unlikely that the final C–H bond
forming step would limit the turnover rate.
Scheme 2 also provides a sensible rationale for the
production of excess Z-isomer in the ring-opening reac-
tion path. Cyclopropane is a powerful stabilizing sub-
stituent for carbocations when a C–C bond of the ring
can align with the vacant p-orbital on carbon [27]. Two
conformations are possible for intermediate 22 such that
a C–C bond in the cyclopropane is colinear with the
Rh–C bond. (For simplicity Scheme 2 shows only the
conformer leading to the Z-isomer.) Conformer (E)-22
suffers from steric interactions with the phenyl group and
is presumably less stable. Rearrangement by way of
conformer (Z)-22 places the alkyl group at C2 syn with
respect to the triethylsiloxy group, while (E)-22 delivers
the alkyl group syn relative to the phenyl group. The
relative ground-state population of (Z)- and (E)-22
would account for the observed Z/E ratio in the enol
silane products.
The experimental data are best accommodated by a
reaction mechanism proceeding through a polar transi-
tion state. Two lines of argument in addition to the
solvent effect lead us to this proposal. First, ketones
capable of stabilizing positive charge promote the cyclo-
propyl carbinyl rearrangement, while ketones possessing
weaker electron-donating groups such as methyl- and
p-trifluoromethylphenyl ketones do not submit to the
rearrangement. Second, intermediates 21 and 22 differ
only in the nature of the hydride and triethylsilyl ligands.
Differences between these ligands include (1) steric
interactions in the a-triethylsiloxyalkyl intermediate (22),
and (2) the influence of the Rh–silyl bond on an adjacent
Rh–C bond. NMR data and MO calculations support
this influence. For example in the facial isomer
[Ir(CH3)H(SiR3)(P(CH3)3)3], the chemical shift of the
methyl resonance shifts downfield 0.1 ppm as the silicon
ligand, R, changes from ethyl to methoxy [8]. In the same
iridium complexes the hydride resonance shifts only 0.03
ppm. Both changes in chemical shift are in the predicted
direction, however, the silyl group has a greater influence
on the polarizable methyl group. As for MO predictions,
calculations on transition metal silyl derivatives suggest
that the short M–Si bonds can be attributed to dp–dp
bonding in which the metal donating electron density to
an empty d-orbital on silicon [24]. Taken together, these
data indicate that the Rh–C bond in complex 22 is
polarized more toward rhodium than in complex 21.
Both the aryl-substituent effect and the Rh–Si effect
work in concert shifting electron density from carbon to
rhodium. We propose that the transformation of 22 to
23 occurs by either an asynchronous concerted reaction
Scheme 2.

242
C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
reaction. A simple reductive elimination/oxidative addi-
tion sequence can, however, interchange the triethylsilyl
ligand in 5 with the hydride ligand in 6 (see Scheme 3).
Reductive elimination of hexaethyldisilane from 5, or
loss of H2 from 6, interchanges the constituency of
complexes 3 and 4. Oxidative addition of triethylsilane
to the reconstituted complexes completes the inter-
change of active catalysts. Eisenberg characterized a
similar s-bound ligand exchange in the complex
[IrX(CO)(dppe)] (X=Br, CN) [28,29]. In the iridium
series, ligand exchange was considered a secondary
reaction with considerably longer reaction times than
the initial oxidative addition. Since we observe no
significant drift in the catalysis towards production of
13 or 14 over many hours, we conclude that s-bound
ligand exchange does not occur to an appreciable
extent.
In conclusion, we have found a unique axis of
reagents and reaction conditions that catalyzed an un-
precedented cyclopropyl carbinyl rearrangement. Under
all conditions the cyclopropyl carbinyl rearrangement
competed against the expected 1,2-carbonyl addition
reaction. The two catalyses showed different character-
istics when exposed to excess triethylsilane and stoichio-
metric quantities of triphenylphosphine, thereby
demonstrating that two rhodium catalysts are responsi-
ble for the product mixture.
Scheme 3.
Our data can not unequivocally identify the source of
profound triphenylphosphine inhibition. Green re-
ported that trimethylphosphite inhibited triethylsilane
addition to [Rh(h3-C3H5)(P(OCH3)3)] [7a], however,
our catalytic system is different since complex 2 is
doubly coordinately unsaturated and
2 mol of
triphenylphosphine must add to 2 to prevent oxidative
addition of triethylsilane. We suggest, on the other
hand, that inhibition stems from phosphine addition to
complex 3. The conversion of 23 into 5 may occur by
two routes: (1) triethylsilane assisted reductive elimina-
tion, or, (2) direct C–H reductive elimination. The
latter route produces complex 3 during every turnover
of catalyst thereby maintaining a level of 3 that could
be trapped by triphenylphosphine. We cannot rule out
an assisted reductive elimination process, however, this
mechanism would generate an active catalyst that en-
ters directly into the reaction stream. We anticipate that
5 exists in a steady-state concentration, and trapping
with triphenylphosphine would be ineffective.
Finally, we are unable to address two issues associ-
ated with this unique bifurcated catalysis. Inspection of
intermediate 22 reveals that it could eliminate 13 in-
stead of rearrange to 23. Formation of 13 from 22
could occur slowly compared to rearrangement yet be
proportional to catalyst concentration and triethylsi-
lane concentration. Our analytical method cannot dis-
tinguish the contribution of this alternative elimination.
One method which we investigated involved preparing
complex 3 in very high yield, thereby all reaction prod-
ucts would funnel through the same catalyst. Based on
an observation of Thorn [9], addition of triphenylsilane
to a benzene solution of 2 should generate complex 3
very selectively. After heating a mixture of 2 and
triphenylsilane (two equivalents) for 10 min, ketone and
triethylsilane were added but unfortunately the mixture
showed no catalytic activity.
6. Experimental
6.1. General considerations
All manipulations of air-sensitive materials were per-
formed under nitrogen by vacuum-line techniques or in
a Vacuum Atmosphere drybox equipped with an inert
gas purifier. Air-sensitive compounds were exposed to
benzene, or benzene-d6 (C6D6) which were dried over
sodium/benzophenone before use. Cyclopropyl phenyl
ketone (Aldrich), cyclopropyl methyl ketone (Aldrich),
and triethylsilane (Aldrich) were doubly distilled and
degassed by freeze–pump–thaw cycles prior to use in a
drybox. Oxaallyl complex (Ph3P)2Rh(h3-CH2C(Ph)O)
(1) was prepared by a previously reported method [6],
while rhodium allyl complex (Ph3P)2Rh(h3-CH2CH-
CH2) (2) was prepared by the method of Muetterties
[30].
1H-NMR spectra were acquired on either a Varian
Unity 300 (300 MHz) or Jeol FX-90Q (90 MHz) spec-
trometer. Broadband decoupled and gated 13C-NMR
spectra were obtained on a JEOL FX-90Q (22.5 MHz)
spectrometer, while proton decoupled 31P{1H}-NMR
spectra were obtained on a Jeol FX-90Q (36.3 MHz)
spectrometer. Chemical shifts were recorded relative to
the residual benzene signal in benzene-d6 (l 7.16) in the
1H-NMR spectra and relative to the central carbon
A second issue is concerned with the integrity of
complexes 5 and 6 during the catalysis. We found
compelling evidence in this study that complexes 5 and
6 remain active, competitive catalysts throughout the

C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
243
resonance of benzene (l 128.0) in the carbon spectra.
All 31P{1H}-NMR resonances were measured relative
to an external standard of 85% H3PO4. IR spectra
were measured on a Perkin–Elmer 16 (FTIR) spec-
trometer. Gas chromatographic analysis was con-
tion rate was adjusted to 135 ml h−1. After 3.5 h the
addition of triethylsilane was complete. The mixture
was heated for an additional 10 h. An aliquot of
reaction mixture (5.0 ml) was removed by a syringe
and placed in a 2.00 ml volumetric flask. The oil was
diluted to 2.00 ml with benzene. A 1.0 ml injection
volume was used for GC analysis. The total integra-
tion of enol silane products ((E)10.29 min, (Z) 10.94
min) and silyl ether (10.64 min) were measured and
the yield of each product based on standard curve
data was 81:15%, respectively.
ducted on
a
Hewlett Packard 5890 IIA gas
chromatograph equipped with a flame ionization de-
tector. All analyses were done with a 25 m OV-1
capillary column.
6.2. General catalytic study (complex 2 or complex 1
catalyst)
6.4. Inhibition studies with gal6inoxyl radical (or
triphenylphosphine)
Cyclopropyl phenyl ketone was weighed (510.5 mg,
3.50 mmol) into an oven-dried 1.00 ml volumetric
flask, and triethylsilane (405 ml, 3.50 mmol) was
added by syringe. Solid complex 2 (16.8 mg, 0.025
mmol) was added in one portion. The volume was
adjusted to 1.00 ml with benzene. The mixture was
thoroughly mixed and the mixture was divided
equally into six 1-ml ampules, each equipped with a
small stir bar. All ampules were sealed with a septum
cap. Outside the glovebox, the six ampules were set
into a rack in a 60.090.2°C oil bath. At appropriate
times an ampule was withdrawn from the oil bath
and frozen in N2. After thawing, 5.0 ml of the mix-
ture is removed by a syringe and placed in a 2.00 ml
volumetric flask. The oil was diluted to 2.00 ml with
benzene. A 1.0 ml injection volume was used for GC
analysis. A second sample was prepared by the same
dilution procedure and the sample was analyzed.
In the case of complex 1, 18.8 mg of the orange
solid was weighed and added to the reaction mixture.
Area counts for the remaining ketone and each re-
action product were averaged for the two GC analy-
ses. Using standard curves generated for cyclopropyl
phenyl ketone, 13, and (Z)-14 the area counts were
converted to molarity. The ketone (510.5 mg, 3.50
mmol), triethylsilane (10 ml) was added by syringe,
and then solid complex molarities of 13 and the sum
of (Z+E)-14 were divided by the concentration of
complex 2 (or 1). These normalized values were plot-
ted vs. time (min) to determine the initial turnover
rate.
Because of solubility issues the general reaction
method was modified. Into an oven-dried 2.00 ml vol-
umetric flask cyclopropyl phenyl ketone was weighed
(510.5 mg, 3.50 mmol) and triethylsilane (405 ml, 3.50
mmol) was added by syringe. Solid galvinoxyl radical
(each of the following weights were used 3.5, 8.2, 10.4
and 20.8 mg) was weighed and added to the volumet-
ric flask. Finally, solid complex 2 (16.8 mg, 0.025
mmol) was added in one portion. The volume was
adjusted to 2.00 ml with benzene. The mixture was
thoroughly mixed, and the mixture was divided
equally into six 1 ml ampules, each equipped with a
small stir bar. All ampules were sealed with a septum
cap. Outside the glovebox, the six ampules were set
into a rack in a 60.090.2°C oil bath. At appropriate
times an ampule was withdrawn from the oil bath
and frozen in N2(1). After thawing, 5.0 ml of the mix-
ture is removed by a syringe and placed in a 2.00 ml
volumetric flask. The oil was diluted to 2.00 ml with
benzene. A 1.0 ml injection volume was used for GC
analysis. Each sample was analyzed in duplicate.
In the case of triphenylphosphine, the following
weights were used: 1.0 mg, 3.3 mg, 5.0 mg.
6.5. Preparation of silyl ether 17
Into an oven-dried 1.00 ml ampule, equipped with
a small stir bar, cyclopropyl p-trifluoromethylphenyl
ketone was weighed (510.5 mg, 3.50 mmol). Triethyl-
silane (405 ml, 3.50 mmol) was added by syringe, and
then solid complex 2 (16.8 mg, 0.025 mmol) was
added in one portion. The ampule was sealed with a
septum cap. Outside the glovebox, the ampule was set
into a rack in a 60.090.5°C oil bath. The mixture
was heated for 10 h. The septum cap was removed
and red/orange solution was added to a 25 ml pear
flask. Approximately 5 ml of the mixture was removed
and used for GC analysis. The remaining solution
was concentrated in vacuo. The residue was separated
by flash chromatography using 10:1 hexane: ethyl ac-
etate as eluant (Rf=0.88). Compound 17 (762 mg,
6.3. Syringe pump addition of triethylsilane
Cyclopropyl phenyl 2 was weighed (16.8 mg, 0.025
mmol) into an oven-dried 1.00 ml ampule, equipped
with a small stir bar, and was added in one portion.
The ampule was sealed with a septum cap. A 6 inch
needle attached to a 0.5 ml syringe filled with tri-
ethylsilane (400 ml) was inserted through the septum
cap. Outside the glovebox, the ampule was set into a
rack in a 60.090.5°C oil bath. The syringe was
placed in a single stage syringe pump and the addi-

244
C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
1
Org. Chem. 47 (1982) 5326. (e) M. Sohn, J. Blum, J. Am.
Chem. Soc. 101 (1979) 2694. (f) J.M. Brown, B.T. Golding,
J.J. Stofko Jr., J. Chem. Soc. Perkin Trans. 2 (1978) 436. (g)
R.G. Salomon, M.F. Salomon, J.L.C. Kachinski, J. Am.
Chem. Soc. 99 (1977) 1043. (h) T.H. Johnson, T.F. Baldwin, J.
Org. Chem. 45 (1980) 140. (i) P.G. Gassman, T.J. Atkins, J.
Am. Chem. Soc. 94 (1972) 7748. See other examples for catal-
ysis involving strained cyclopropane systems. (j) H.C. Volger,
H. Hogeveen, M.M.P. Gaasbeek, J. Am. Chem. Soc. 91 (1969)
2138. (k) H. Hogeveen, H.C. Volger, J. Am. Chem. Soc. 89
(1967) 2486. (l) T.J. Katz, S. Cerefice, J. Am. Chem. Soc. 91
(1969) 2405.
66%) was isolated as a faint yellow oil. H-NMR (300
MHz, CDCl3): 7.57 (d, J=9.0 Hz, H-Ar), 7.47 (d,
J=9.0 Hz, 1H, H-Ar), 4.25 (d, J=6.0 Hz, 1H, H–
C–O), 1.09 (m, 1H, cyclopropyl), 0.92 (dt, J=2.9,
8.0 Hz, 9H, CH3), 0.55 (m, 10H, Si–CH2 and cyclo-
propyl), 0.46 (m, 2H, cyclopropyl), 0.37 (m, 1H, cy-
clopropyl). 13C-NMR (75 MHz, CDCl3): 149.6 (Ar),
129.2 (q, JC–F=31.2 Hz, Ar), 126.1 (Ar), 125.0 (Ar),
125.0 (Ar), 76.8 (C-O), 20.1 (cyclopropyl), 6.7 (CH3),
4.9 (Si–C), 3.2 (cyclopropyl), 2.5 (cyclopropyl). IR
(neat): 3084(m), 3009(m), 2957(s), 1619(m), 1321(vs),
1164(s), 1128(vs).
[12] Voigt differentiated the reactivity of vinylcyclopropanes and
acyl-substituted cyclopropanes based on the indifference of
acyl cyclopropanes towards rearrangement. H.M. Voigt, J.A.
Roth, J. Catal. (1974) 91.
[13] (Z)-1-phenyl-1-triethylsiloxy-1-butene (98%) and (E)-1-phenyl-
1-triethylsiloxy-1-butene (2%) were prepared by enolization of
butrophenone with LDA and trapping the enolate with chloro-
triethylsilane. See: C.H. Heathcock, S.K. Davidsen, K.T. Hug,
L.A. Flippin, J. Org. Chem. 51 (1986) 3027.
[14] Asymmetric hydrosilylation of cyclopropyl phenyl ketone has
been achieved with 1,5-cyclooctadienerhodium chloride dimer-
pyridinethiazolidene catalyst. No ring-opened products were
identified in the reaction. H. Brunner, A. Kuerzinger, J.
Organomet. Chem. 346 (1988) 413.
[15] The 31P-NMR spectrum in these non-catalytic mixtures gave
several resonances and the identity of each product could not
be asserted.
Acknowledgements
The authors gratefully acknowlede the generous
support of the National Science Foundation (Grant
No. CHE-9200699) and also Johnson Matthey for a
generous loan of rhodium(III) chloride hydrate.
References
[1] I. Ojima, Z. Li, J. Zhu, Recent advances in the hydrosilylation
and related reactions, in: Z. Rappoport, Y. Apeloig (Eds.),
The Chemistry of Organic Silicon Compounds, Wiley, New
York, 1998, p. 1687.
[2] I.A. Lorkovic, R.R. Duff Jr., M.S. Wrighton, J. Am. Chem.
Soc. 117 (1995) 3617.
[3] (a) E.N. Jacobsen, W. Zhang, M.L. Guler, J. Am. Chem. Soc.
113 (1991) 6703. (b) H. Nishiyama, S. Yanagushi, M. Kondo,
K. Itoh, J. Org. Chem. 57 (1992) 4306. (c) T.V. Rajanbabu,
A.L. Casalnuovo, J. Am. Chem. Soc. 114 (1992) 6265. (d)
T.V. Rajanbabu, T.A. Ayers, A.L. Casalnuovo, J. Am. Chem.
Soc. 116 (1994) 4101.
[4] (a) T.H. Chan, Z.G. Zheng, Tetrahedron Lett. 34 (1993) 3095.
(b) G.Z. Zheng, T.H. Chan, Organometallics 14 (1995) 70. (c)
H. Nagashima, K. Tatebe, T. Ishibashi, A. Nakaoka, J.
Sakakibara, K. Itoh, Organometallics 14 (1995) 2868.
[5] S.A. Westcott, H.P. Blom, T.B. Marder, R.T. Baker, J. Am.
Chem. Soc. 114 (1992) 8863.
[16] Authentic samples of ether 9 were prepared and the methine
hydrogen signal matched the resonance frequency.
[17] Three different concentrations of ketone (3.50, 1.73 and 1.12
M) were investigated. Each showed an initial linear region
followed by a region of logarithmic decay.
[18] Ketone coordination with silylhydridorhodium has been shown
to be the rate-determining step at low ketone concentration. I.
Kolb, J. Hetflejs, Collect. Czech. Chem. Commun. 45 (1980)
2224.
[19] I. Ojima, M. Kumagai, J. Organomet. Chem. 111 (1976) 43.
[20] (a) A.J. Chalk, J.F. Harrod, J. Am. Chem. Soc. 87 (1965) 16.
(b) I. Ojima, M. Nihonyanagi, Y. Nagai, Bull. Chem. Soc.
Jpn. 45 (1972) 3722. (c) I. Ojima, T. Kogure, M. Kumagai, S.
Horiuchi, T. Sato, J. Organomet. Chem. 122 (1976) 83. (d) K.
Yamamoto, T. Hayashi, M. Kumada, J. Organomet. Chem.
113 (1976) 124. (e) J.F. Peyronel, H.B. Kagan, Nouv. J.
Chem. 2 (1978) 211.
[6] G.A Slough, J.R Ashbaugh, L.A. Zannoni, Organometallics 13
(1994) 3587.
[21] L.A. Zannoni, G.A. Slough, unpublished results.
[22] (a) C.P. Casey, P.C. Vosejpka, J.A. Gavney, Jr., J. Am. Chem.
Soc. 112 (1990) 4083. (b) C.P. Casey, P.C. Vosejpka, T.L.
Underinger, G.A. Slough, J.A. Gavney Jr., J. Am. Chem. Soc.
115 (1993) 6680. (c) C.P. Casey, T.L. Underinger, P.C. Vose-
jpka, G.A. Slough, J.A. Gavney Jr., R.K. Hayashi,
Organometallics 16 (1997) 2189.
[7] Two previous papers used h3-allyl rhodium(I) complexes as a
catalyst precursor. Neither paper identified the rhodium mate-
rial produced when reacted with organosilane. (a) M. Bottrill,
M.J. Green, Organomet. Chem. 111 (1976) C6. (b) J. Rejhon,
J. Hetflejs, Collect. Czech. Chem. Commun. 40 (1975) 3190.
[8] M. Aizenberg, D. Milstein, Angew. Chem. Int. Engl. 33 (1994)
317.
[23] S. Chemaly, J.M. Pratt, J. Chem. Soc. Chem. Commun. (1976)
988.
[9] D.L. Thorn, R.L. Harlow, Inorg. Chem. 29 (1990) 2017.
[10] M.S. Eitz, M.S. Wrighton, Angew. Chem. Int. Ed. Engl. 27
(1988) 289.
[11] Several homogenous rhodium(I), rhodium(II), platinum(II),
copper, and ruthenium(II) complexes efficiently catalyze struc-
tural rearrangements of cyclopropane derivatives. (a) J.
Halpern, in: I. Wender, P. Pino (Eds.), Organic Syntheses Via
Metal Carbonyls, vol. II, Wiley, New York, 1977, p. 705. (b)
K.C. Bishop III, Chem. Rev. 76 (1976) 461. (c) S. Sarel, Acc.
[24] A.D. Berry, E.R. Corey, A.P. Hagan, A.G. MacDiarmid, F.E.
Saalfeld, B.B. Wayland, J. Am. Chem. Soc. 92 (1970) 1940.
Other explanations regarding bonding of electropositive ele-
ments to metals have also appeared. For example: B.Y.K. Ho,
J.J. Zuckerman, J. Organomet. Chem. 49 (1973) 1.
[25] (a) Z. Majerski, P.R. Von Schleyer, J. Am. Chem. Soc. 93
(1971) 665. (b) K.B. Wiberg, G. Szeimies, J. Am. Chem. Soc.
92 (1970) 572.
[26] S. Danishefsky, Acc. Chem. Res. 18 (1979) 12.
Chem. Res. 11 (1978) 204. (d) M.P. Doyle, D. van Leusen, J.
.

C. Sun et al. / Journal of Organometallic Chemistry 582 (1999) 235–245
245
[27] For a review of cyclopropylmethyl cations, see: H.C. Richey
Jr., in: G.A. Olah, P.R. Von Schleyer, (Eds.), Carbonium Ions,
vol. III, Wiley-Interscience, New York, 1972, Ch. 25.
[28] C.N. Johnson, R. Eisenberg, J. Am. Chem. Soc. 107 (1985)
6531.
[29] Glocking isolated a germyl Ir(I) and a germyl Ir(III) complex
which model the exchange process. F. Glocking, M.D. Wilbey,
J. Chem. Soc. A. (1970) 1675.
[30] A.J. Sizak, E.L. Muetterties, J. Am. Chem. Soc. 101 (1979)
4878.
.