Some studies involving the use of chiral amine oxides for the enantioselective preparation of (propargyl alcohol)Co2(CO)5(PR3) complexes

Article abstract of DOI:10.1016/S0277-5387(99)00077-7

The enantioselective preparation of (propargyl alcohol) dicobalt pentacarbonylphosphine complexes (2) via kinetic resolution with an optically active amine oxide was examined. Variations in temperature, solvent, phosphine, amine oxide, and substrate were studied. Diastereoselectivity showed marked improvement over thermal methods, while enantioselectivity was modest. The typical e.e. for product 2a was 20% at 30-40% conversion when the reaction was carried out at -58°C in a 1:1 mixture of tetrahydrofuran/dichloromethane with brucine N-oxide as the promoter. Elsevier Science Ltd.

Full text of DOI:10.1016/S0277-5387(99)00077-7

Polyhedron 18 (1999) 2027–2034
Some studies involving the use of chiral amine oxides for the
enantioselective preparation of (propargyl alcohol)Co2(CO)5(PR3)
complexes
*
Nancy E. Carpenter, Kenneth M. Nicholas
Department of Chemistry and Biochemistry, University of Oklahoma, Norman, OK 73019, USA
Received 2 September 1998; accepted 16 March 1999
Abstract
The enantioselective preparation of (propargyl alcohol) dicobalt pentacarbonylphosphine complexes (2) via kinetic resolution with an
optically active amine oxide was examined. Variations in temperature, solvent, phosphine, amine oxide, and substrate were studied.
Diastereoselectivity showed marked improvement over thermal methods, while enantioselectivity was modest. The typical e.e. for product
2a was 20% at 30–40% conversion when the reaction was carried out at 2588C in a 1:1 mixture of tetrahydrofuran/dichloromethane with
brucine N-oxide as the promoter.
1999 Elsevier Science Ltd. All rights reserved.
Keywords: Chiral amine oxide; Enantioselective preparation; (Propargyl alcohol)Co2(CO)5(PR3) complexes
1. Introduction
pounds, but classical methods for the preparation of
homochiral organometallic complexes are typically lengthy
and often require a tedious resolution step. Generation of a
chiral metal center frequently requires ligand substitution,
most often replacement of CO by a phosphine or similar
ligand. The standard methodology for ligand substitution,
e.g. thermal or photochemical labilization of CO, frequent-
ly requires stringent reaction conditions, potentially caus-
ing epimerization of metal stereocenters. In contrast, the
amine oxide-induced replacement of carbonyl ligands is a
remarkably mild transformation [8] which has found use in
the acceleration of the Pauson–Khand reaction [9–11].
Kerr et al. have recently demonstrated that the use of
enantiomerically enriched chiral amine oxides with proch-
iral (alkyne)Co2(CO)6 complexes in this reaction results in
low to moderate asymmetric induction [12]. These re-
searchers have also reported an elegant achiral amine
oxide-promoted preparation of optically pure dicobalt
(alkyne) pentacarbonylphosphine complexes using a chiral
phosphine ligand [13,14]; Brunner and coworkers had
prepared such complexes earlier by direct thermal substitu-
tion [13,14].
The distinctive reactivity and structure of dicobalt
propargyl alcohol complexes (1 and 2, Eq. (1)) and the
derived cations 3 has led to considerable exploitation of
their synthetic utility by our research group and others
[1,2]. It was anticipated that an asymmetric cobalt cluster
(as in 2) should provide facial selectivity in the reactions
of cobalt-stabilized propargyl cations or radicals, thereby
leading to enantiomerically-enriched complexes and, ulti-
mately, optically active organic products [3]. Given
stereocenters at both C1 [4] and at the cobalt cluster, four
stereoisomers (two diastereomers, each as a pair of en-
antiomers) can be formed in the ligand substitution re-
action, as shown in eq. 1 (the configuration of only one of
the stereogenic cluster carbons is shown since the second
carbon’s configuration is dependent on the first [5–7]).
The ongoing need for stereoselective methods for the
preparation of optically pure organic molecules has moti-
vated our continuing pursuit of efficient enantioselective
routes to chiral cobalt complexes of type 2. Numerous
powerful methodologies in organic synthesis take advan-
tage of asymmetric induction by organometallic com-
In order to prepare 2 in optically pure form, stereocon-
trol at both C1 and the cobalt cluster must be considered.
Our previous work [5–7] demonstrated that control at the
cluster is possible with moderate to high diastereoselectiv-
ity in thermal reactions of 1 with L5PPh3 and that the
*Corresponding author. Tel.: 11-405-325-4811; fax: 11-405-325-
6111.
E-mail address: knicholas@ou.edu (K.M. Nicholas)
0277-5387/99/$ – see front matter
PII: S0277-5387(99)00077-7
1999 Elsevier Science Ltd. All rights reserved.

2028
N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
(1)
cluster is configurationally stable, an important considera-
tion in the development of enantioselective variants of
subsequent reactions. Furthermore, use of optically active
propargyl alcohol complex 4 as shown in Scheme 1 [3]
resulted in the formation of product 5 with the same e.e. as
starting material 4, providing evidence for the stability of
the cations 3 to cluster epimerization, and showing prom-
ise for enantioselective propargylation.
the Kerr group we sought to evaluate the potential of
optically pure amine oxides for the enantioselective and
diastereoselective preparation of phosphine-substituted
cobalt propargyl alcohol complexes as shown in Eq. (1).
This novel approach to optically active cobalt propargyl
alcohol complexes has as its key feature the kinetic
resolution of racemic 1 by reaction with a homochiral
amine oxide in the presence of an achiral phosphine. The
basic premise of the resolution is straightforward. Let kr be
Stimulated by these findings and in communication with
Scheme 1. Enantioselective propargylation by a chiral (propargyl alcohol)Co2(CO)5(PR3) complex.

N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
2029
the rate constant for conversion of the R enantiomer of 1
into product, and ks the rate constant for conversion of the
S enantiomer. If kr4ks, stopping the reaction prior to
100% conversion allows recovery of starting material
enriched in the S isomer and, ideally, optically pure
product. A similar approach to the preparation of optically
active organometallic complexes was first demonstrated by
von Phillipsborn and co-workers [15] in the preparation of
optically active enone–iron carbonyl complexes. The
present system is exceptional in that a new stereocenter at
the metal cluster is created in the course of the kinetic
resolution. Few other kinetic resolutions of chiral or-
ganometallic complexes can be found in the literature, and
these typically involve reaction of the resolving reagent at
the organic ligand rather than at the metal center(s)
[16,17].
1.77 (d, J54.5 Hz, 1 H), 3.60 (m, 1H), 7.00–7.40 (m, 20
H).
3. Results and discussion
The effects of variations in temperature, solvent, sub-
strate, phosphine and amine oxide on the diastereoselectiv-
ity, enantioselectivity and rate of the reaction were investi-
gated. Additionally, variations in both stoichiometry and
procedure were examined in an attempt to optimize the
enantioselectivity of the reaction. The standard reaction
protocol consisted of adding a slurry containing the
phosphine and the amine oxide to a stirred, cooled solution
of racemic hexacarbonyl cobalt complex (1). The reaction
progress (specifically, the change in e.e. of both 1 and 2
with time) was monitored by analyzing aliquots from the
reaction mixture by HPLC using a chiral column. The
products were identified by isolation and comparison with
authentic (racemic) samples. Most experiments were con-
ducted on the amine oxide-promoted reaction of 1a with
PPh3. HPLC analysis allowed determination of the en-
antiomeric composition of both the unreacted alcohol
complex 1a and the major diastereomer of product 2a; the
minor diastereomer of 2a, however, could not be resolved.
The major diastereomer of 2a produced in the reaction was
found to be the R,S/S,R isomer, which also dominates in
the purely thermal reaction [5].
2. Experimental
All reactions were carried out under a positive pressure
of dry high-purity nitrogen. Diethyl ether, dioxane and
tetrahydrofuran were dried and distilled from sodium/
benzophenone ketyl. Dichloromethane and acetonitrile
were dried and distilled from calcium hydride under an
atmosphere of nitrogen. Methanol was dried and distilled
˚
from
4A
molecular
sieves.
The
and
(propargyl
(propargyl
alcohol)Co2(CO)6
(1a–d)
alcohol)Co2(CO)5(PPh3) (2a–d) complexes were prepared
as reported previously [5]. The amine oxides 6–9 were
prepared by oxidation of the corresponding amine with
hydrogen peroxide (Brucine N-oxide, W.J. Kerr, personel
communication; [31–33]). Analysis of e.e. was carried out
using an ISCO HPLC system equipped with a Chiralpak
AD column, eluted with a 2:98 mixture of isopropanol in
hexanes, and detected at l254 nm.
3.1. Temperature effects.
Several initial experiments were carried out to determine
the effect of temperature on diastereoselectivity in re-
actions of 1a with PPh3 and to find the optimum tempera-
ture at which to carry out subsequent reactions. As
expected, diastereoselectivity increased as the temperature
decreased (from 38% d.e. at 2158C to 52% d.e. at 2308C,
76% d.e. at 2608C, and 78% d.e. at 2958C). Because
there appeared to be little improvement in diastereoselec-
tivity below 2608C, a standard temperature of 258648C
was used for all subsequent reactions. As would be
expected for a typical kinetic resolution, the e.e. of the
unreacted starting material slowly increased from its initial
value of zero to as high as 26% e.e. at moderately high
conversion (63%). Fig. 1 shows the change in e.e. of the
unreacted starting material as a function of percent conver-
sion, as well as the change in e.e. and d.e. of the major
diasteromer of 2a for a representative reaction. Interesting-
ly, the e.e. of the product did not decrease steadily with
time, as would be expected in the case of a simple kinetic
resolution [18,19]. Unfortunately, because we could not
assess the e.e. of the minor diastereomer (op cit), it is
impossible to account for this curious behavior. This effect
could be the result of epimerization of the minor to the
major diastereomer.
2.1. General procedure for kinetic resolution of
phosphine substituted complexes (2)
To a stirred and cooled solution (258648C) of racemic
hexacarbonyl cobalt complex (1, 1.0 mmol)) under nitro-
gen in a 1:1 mixture of tetrahydrofuran/dichloromethane
was added slowly a slurry of amine oxide (0.5 mmol) and
triphenylphosphine (1.0 mmol), premixed in the same
solvent mixture. The reaction progress, over a period of up
to several days, was monitored by removing aliquots of the
reaction mixture, flashing them through a small pad of
silica gel to remove the amine/amine oxide, and injecting
the sample on the HPLC.
Mono-phosphinated products were characterized by
comparison to known samples. New complexes 1d and 2d
were characterized by 1H NMR; complex 2e was not
characterized further because of its rapid decomposition.
1d (1H NMR, C6D6) 1.00 (s, 9H), 1.78 (d, 1H, J54.5 Hz),
4.62 (d, 1H, J54.8 Hz), 7.04 (m, 1 H), 7.11 (app t, 2H),
7.58 (d, 2H, J57.5 Hz); 2d (1H NMR, C6D6) 1.02 (s, 9H),

2030
N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
other amine oxide-promoted substitutions [9–12]. Both
protic and aprotic solvents with a range of coordinating
ability were tested. Use of acetone/CH2Cl2 or acetonitrile/
CH2Cl2 increased the rate (Table 1, cf. entries 1 versus 2,
1 versus 5), suggesting the involvement of a coordinatively
unsaturated intermediate [8–11]. However, no enhance-
ment of stereoselectivity was seen with these solvents.
Variations in the ratio of the tetrahydrofuran/dichlorome-
thane co-solvent system similarly had little impact on the
selectivity of the reaction (cf. entries 1 and 6, Table 1), and
was complicated due to the insolubility of the reagents.
Similar solubility difficulties were encountered when using
ether or dioxane in place of tetrahydrofuran and lower
enantioselectivity also resulted (entries 7 and 8, Table 1).
The use of methanol resulted in a greatly diminished
reaction rate (no reaction at 2558C) despite a homogenous
reaction mixture, possibly the result of decreased nu-
cleophilicity of the amine oxide oxygen due to hydrogen
bonding with the solvent (entry 3, Table 1) [21].
3.3. Substrate effects
Based on these findings, the influence of the remaining
variables (the nature of the propargyl substrate, the phos-
phine ligand, and the amine oxide promoter) on
stereoselectivity was tested under the standard conditions
of 258648C and in a 1:1 ratio of THF:dichloromethane.
Only modest enantioselectivity was observed with complex
1a, with a typical maximum enantioselectivity for the
phosphinated product 2a (major diastereomer) of about
20% e.e. at approximately 30–40% conversion when
brucine N-oxide was used as the promoter. As a result, we
briefly turned our attention to complexes 1b–d which bore
Fig. 1. Stereoselectivity versus conversion for the reaction of 1a with
PPh3 promoted by brucine N-oxide at 2588C in 1:1 THF:CH2Cl2.
Percentage e.e. of 2a is that of the major (RS, SR) diastereomer.
3.2. Solvent effects
The nature of the solvent system had a substantial
influence on the rate of the reaction, as has been seen in
Table 1
Effect of solvent on brucine oxide-promoted substitutions
Entry
Solvent
Time
(h)
Conv.
(%)
d.e. (%)a
2a
e.e. (%)b
1
e.e. (%)
2ac
1
2
3
1:1 THF:CH2Cl2
1:1 acetone:CH2Cl2
1:1 CH3OH:CH2Cl2
64
24
18.5
20d
1.5
24
48
43.5
48
33
51
0
54
19
22
24
44
28
58
40
–
54
56
62
–
10
4
–
20
0
–
0
16
0
14
4
4
4
4
5
6
7
8
1:1 CH3CN:CH2Cl2
1:9 CH3CN:CH2Cl2
1:19 THF:CH2Cl2
1:3 dioxane:CH2Cl2
1:4 ether:CH2Cl2
–
4
–
6
4
52
58
a Percentage major isomer–percentage minor isomer.
b Percentage major enantiomer–percentage minor enantiomer.
c Percentage e.e. of major diastereomer.
d After initial time at 2588C, reaction was warmed to room temperature.

N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
2031
isopropyl or tert-butyl groups. We anticipated that the
phosphine substitution could be carried out diastereospeci-
fically due to the increased steric demands of the propargyl
R2 group [5], allowing us to focus on enantioselectivity.
Unfortunately, brucine N-oxide-promoted triphenylphos-
phine substitution of complex 1b stubbornly yielded a
mixture of diastereomers (26% d.e. after 48 h at 2558C).
Complex 1c underwent the brucine N-oxide-promoted
substitution reaction diastereospecifically, but the mono-
substituted product was contaminated with a small amount
of the bis-phosphinated complex. This contaminant was
inseparable by chromatography, thus rendering the HPLC
analysis of enantioselectivity unreliable. Further inves-
tigations with both 1b and c were therefore abandoned.
Triphenylphosphine substitution of complex 1d proceeded
smoothly to yield a single diastereomeric product, but the
complex failed to resolve on the chiral HPLC column. The
decomplexed alcohol, however, did resolve, hence enantio-
selectivity was monitored by removing an aliquot from the
reaction mixture, separating the starting material from the
product by flash silica gel chromatography, decomplexing
the alcohol from product 2d with ceric ammonium nitrate,
and injecting the crude alcohol on the HPLC. A promising
e.e. of 40% was observed after 72 h at 2558C, but the e.e.
of the decomplexed alcohol varied erratically with time,
possibly due to racemization or incomplete separation of
the starting complex from the product during workup.
reaction conditions. This finding, in conjunction with the
expense of tri(o-tolyl) phosphine and its relatively lacklus-
ter selectivity (the system peaked at 12% e.e. for product
2e), resulted in its rejection as a viable alternative.
3.5. Amine oxide effects
Brucine N-oxide (6), sparteine N-oxide (7), quinine
N-oxide (8), and two proline derivatives (9a and b), all
prepared by oxidation of the corresponding amines with
hydrogen peroxide, were studied to determine their effect
on the stereoselectivity of the reaction.
Brucine N-oxide (BNO) was the most effective amine
oxide in terms of ease of preparation, convenience (solu-
bility), and consistency of results: the e.e. for the prepara-
tion of complex 2a was typically 20% when BNO was
used. Furthermore, brucine N-oxide consistently gave the
highest e.e. Use of sparteine N-oxide (in place of BNO)
ultimately resulted in a large decrease in the diastereo-
selectivity (8% d.e. at 97.5 h and 47% conversion), with
the amount of minor diastereomer steadily increasing with
time [20]. Quinine N-oxide initially showed a similar
negative impact on the diastereoselectivity (18% d.e. after
30 h and 5% conversion), but in this case, the dia-
stereoselectivity increased with increasing reaction time
(eventually reaching 64% d.e. at 52% conversion; entry 3,
Table 2). The choice of amine oxide also profoundly
affected the rate of the reaction. Reaction of 1a with the
proline-derived amine oxides 9a–b was dramatically
slower as was the reaction with quinine N-oxide (entries 4
and 5, Table 2). This phenomenon was attributed in part to
the reduced solubility of the amine oxide in the reaction
medium, but the reduced nucleophilicity of the amine
oxide oxygen (due to intramolecular hydrogen bonding)
may play a role as well. Even in a homogenous aprotic
reaction medium, the reaction rate was excruciatingly
slow.
3.4. Phosphine effects
The encouraging results seen with complex 1d sug-
gested that the selectivity of the reaction could be im-
proved by increasing the steric demands of the other
species presumably involved in the transition states (e.g.
the amine oxide and the phosphine). Accordingly, tri(o-
tolyl) phosphine was employed in place of triphenylphos-
phine. Unfortunately, the resulting monophosphinated
complex 2e was found to decompose readily under the

2032
N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
Table 2
Effect of amine oxide on substitutions
Entry
Amine
Oxide
% conv.
d.e. (%)
e.e. (%) of
2aa
2ab
1
2
3
4
6
7
8
9b
33
43
52
11
66
22
64
72
20
4
2
0
a Percentage major isomer–percentage minor isomer.
b Of major (RS/SR) diastereomer.
3.6. Other effects
takes place only very slowly at 208C in the absence of
added ligands [5–7], the amine oxide and/or the derived
amine apparently facilitates the epimerization process.
Although the levels of enantioselection observed in
these reactions are not practically useful, this study does
provide the first examples of kinetic resolution of racemic
(alkyne)Co2(CO)5L complexes. The origin of the enantio-
selectivity in this system, however, appears to be complex,
as indicated by the unexpected increase in the e.e. of the
product 2 with increasing conversion (Fig. 1) as well as
some of the effects of solvent change and the addition of
brucine to the reactions. Clearly this system does not
adhere to the description of a simple kinetic resolution.
Kagan and others [18,19] have demonstrated that, in
systems where an asymmetric center is created in the
resolving process (as is the case here), the e.e. of the
diastereomeric products is interdependent upon a number
of factors, including the e.e. and fractional amount of each
chiral species, as well as the extent of conversion. The
complex interplay of these effects can cause dramatic
deviations from the typical e.e. versus conversion be-
havior.
In the present system, the enantioselectivity must result
from discrimination between diastereomeric transition
states (or intermediates) involving the homochiral amine
oxide and a complexed propargyl alcohol species. A
number of such species may be playing a role in determin-
ing enantioselectivity. The proposed mechanism for amine
oxide-promoted substitutions begins with initial nu-
cleophilic attack at a coordinated carbonyl ligand by the
amine oxide (10, Scheme 2), [8,21–26]. It is generally
accepted that this initial step is followed by loss of CO2 to
generate a coordinatively unsaturated intermediate (11)
[8–11]; however, experimental evidence for this inter-
mediate and subsequent steps is limited. Intermediate 11
may be stabilized by coordination of the amine generated
in situ from the decarboxylation (giving 12) [22–27] but is
likely that the phosphinated products 2 derive from
Given the modest enantioselectivity described above, we
turned our attention to other modifications of the pro-
cedure. Interestingly, the enantioselectivity was considera-
bly improved by the addition of one-half equivalent of
brucine (in addition to the brucine N-oxide). In this trial,
the e.e. of 2a was increased to 32% at 20% conversion.
The presence and amount of the additional amine was
apparently critical, as addition of two equivalents of
brucine returned the e.e. to typical values (18% at 50%
conversion with a d.e. of 66%). Addition of four equiva-
lents of brucine N-oxide yielded similarly lackluster results
(12% e.e. at 20% conversion).
While the enantioselectivities observed in these re-
actions were modest, amine oxide promotion did provide
several advantages over the thermal preparation. The
formation
of
the
bis-phosphinated
complex
(PPh3)2(CO)4Co2(HC≡CCHOHPh) was completely sup-
pressed, for example (as also observed by Kerr et al.
[13,14]). Furthermore, a dramatic increase in rate and a
significant increase in diastereoselectivity was observed.
Earlier studies demonstrated that, under thermal conditions
(508C), triphenylphosphine substitution on complex 1a
takes place with 60% d.e. [5–7]; both the thermal and the
amine oxide-promoted conditions yielding the same major
diastereomer (2aSR /RS ). The observed diastereomeric ex-
cess for the amine oxide-promoted substitution reactions of
this complex varied with the extent of reaction (Fig. 1),
typically reaching a maximum in the first several hours of
the reaction, then slowly leveling off. The maximum
diastereomeric excess for 2a was 74%, achieved after a
reaction time of 4 h (21% conversion) in a solvent mixture
of 25:75 diethyl ether/dichloromethane at 2628C. The
variation of percentage d.e. with time suggests that even at
low temperatures equilibration of the minor diastereomer
to the major diastereomer may be taking place. Since
previous studies have demonstrated that this equilibration

N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
2033
Scheme 2. Proposed kinetic resolution mechanism for ligand substitution of (propargyl alcohol)Co2(CO)6 complexes.
phosphine trapping of 11, given the dissociative mecha-
nism implicated for thermal ligand substitution of
(alkyne)Co2(CO)6 complexes [28,29].
Formation of the chiral species 10 in step one (Scheme
2) could be the primary stereodifferentiating step, with
subsequent steps having no further impact on the e.e. of
the product. Evidence that this is not the sole source of
stereodifferentiation, however, is provided by: (1) the
observed percentage e.e. of 2 versus conversion behavior
and (2) the fact that when 1a was treated with one
equivalent of BNO but no added phosphine over a period
of 217 h at 2628C, unreacted 1a remained racemic (as
monitored by HPLC). It is possible that the breakdown of
intermediate 10 also could be enantioselective: i.e. if k2R
±

2034
N.E. Carpenter, K.M. Nicholas / Polyhedron 18 (1999) 2027–2034
[6] D.H. Bradley, M.A. Khan, K.M. Nicholas, Organometallics 11
(1992) 2598.
[7] D.H. Bradley, Ph.D. dissertation, University of Oklahoma, 1992.
[8] M.O. Albers, N.J. Coville, Coord. Chem. Rev. 53 (1984) 227.
[9] S. Shambayati, W.E. Crowe, S.L. Schreiber, Tetrahedron Lett. 31
(1990) 5289.
[10] N. Jeong, Y.K. Chung, B.Y. Lee, H.L. Lee, S.-E. Yoo, Synlett (1991)
204.
[11] M.E. Krafft, I.L. Scott, R.H. Romero, S. Feibelmann, C.E. Van Pelt,
J. Am. Chem. Soc. 115 (1993) 7199.
[12] W.J. Kerr, G.G. Kirk, D. Middlemiss, Synlett (1995) 1085.
[13] W.J. Kerr, G.G. Kirk, D. Middlemiss, J. Organomet. Chem. 519
(1996) 93.
[14] H. Brunner, A. Niedernhuber, Tetrahedron: Asymmetry 1 (1990)
711.
[15] B.R. Bender, M. Koller, A. Linden, A. Marcuzzi, W. vonPhillip-
sborn, Organometallics 11 (1992) 4268.
[16] S. Top, G. Jaouen, C. Baldoli, P. Del Buttero, S. Maiorana, J.
Organomet. Chem. 413 (1991) 125.
[17] N.W. Alcock, D.H. Crout, C.M. Henderson, S.E. Thomas, J. Chem.
Soc., Chem. Commun. (1988) 746.
[18] H.B. Kagan, J.C. Fiaud, Topics in Stereochemistry 18 (1988) 249,
Review.
[19] V.S. Martin, S.S. Woodward, T. Katsuki, Y. Yamada, M. Ikeda, K.B.
Sharpless, J. Am. Chem. Soc. 103 (1981) 6237.
[20] The percentage conversion changed very little while the d.e. steadily
decreased, suggesting that the presence of either sparteine N-oxide
or sparteine may have facilitated selective decomposition of the
major diastereomer.
k2S and, if step 2 is rate-limiting (or step 1 is reversible),
then it could impact the overall stereoselectivity. More-
over, our finding that addition of one-half equivalent of
brucine to the reaction mixture improves enantioselectivity
suggests that step 3 (formation of the amine-coordinated
complex 12) also plays a role in the developing enantio-
selectivity. Thus, stereoselectivity could arise at a number
of steps, and each step could, conceivably, either reinforce
the selectivity seen in a previous step or counteract it. It is
also important to note that, if the enantioselectivity is
established en route to the formation of 11, epimerization
of the cobalt cluster at this stage (e.g. by formation of
symmetrical bridging carbonyl intermediates) would obvi-
ously compromise any enantioselectivity achieved up to
this point.
The modest enantioselectivities seen in this reaction, the
critical dependence on reaction conditions to maintain this
fragile enantioselectivity, and the unusual trends in selec-
tivity seen over the course of the reaction all indicate that
this is a complex system. In conclusion, we have found
that the use of chiral amine oxides – in particular, brucine
N-oxide – results in a greatly accelerated carbonyl substi-
tution reaction with an increase in diastereoselectivity over
thermal processes, and with significant but modest enantio-
selectivity.
[21] J.-K. Shen, Y.-C. Gao, Q.-Z. Shi, A.L. Rheingold, F. Basolo, Inorg.
Chem. 30 (1991) 1868.
[22] Y.-C. Gao, Q.-Z. Shi, D.L. Kershner, F. Basolo, Inorg. Chem. 27
(1988) 188.
Acknowledgements
[23] J.K. Shen, Y.-C. Gao, Q.-Z. Shi, F. Basolo, J. Organomet. Chem.
401 (1991) 295.
[24] J.K. Shen, Y.-C. Gao, Q.-Z. Shi, F. Basolo, Organometallics 8
(1989) 2144.
[25] Y. Shvo, E. Hazum, J. Chem. Soc., Chem. Commun. (1975) 829.
[26] J.H. Eekhof, H. Hogeveen, R.M. Kellogg, J. Chem. Soc., Chem.
Commun. (1976) 657.
We thank the University of Minnesota (Morris) for the
sabbatical leave for NEC, Professor D. Glatzhofer for use
of the chiral HPLC column, and Dr. W.J. Kerr for helpful
discussions, some experimental assistance, and communi-
cation of unpublished results.
[27] J. Elzinga, H. Hogeveen, J. Chem. Soc., Chem. Commun. (1977)
705.
References
[28] R.F. Heck, J. Am. Chem. Soc. 85 (1963) 657.
[29] F. Basolo, A. Wojcicki, J. Am. Chem. Soc. 83 (1961) 520.
[30] Other researchers have similarly searched for such evidence; see ref.
[7].
[1] K.M. Nicholas, Acc. Chem. Res. 20 (1987) 207.
[2] A.J.M. Caffyn, K.M. Nicholas, in: L.S. Hegedus (Ed.), Comprehen-
sive Organometallic Chemistry II, Vol. 12, Elsevier, Oxford, 1995,
Ch. 7.1.
¨
[31] F. Galinovsky, W. Fischer, Monatshefte fur Chemie 87 (1959) 763,
sparteine N-oxide.
[32] H. Diaz-Arauzo, J.M. Cook, J. Nat. Prod 53 (1990) 112, quinine
N-oxide.
[33] I.A. O’Neil, N.D. Miller, J.V. Barkley, C.M.R. Low, S.B. Kalindjian,
Synlett (1995) 617, proline N-oxide (9a/b).
[3] A.J.M. Caffyn, K.M. Nicholas, J. Am. Chem. Soc. 115 (1993) 6438.
[4] Numbering C1 as the hydroxyl-bearing carbon.
[5] D.H. Bradley, M.A. Khan, K.M. Nicholas, Organometallics 8 (1989)
554.