Solvent-dependent dynamic kinetic asymmetric transformation/kinetic resolution in molybdenum-catalyzed asymmetric allylic alkylations

Article abstract of DOI:10.1021/jo015871l

Catalytic asymmetric alkylation reactions of branched racemic carbonates 1a and 1b with sodium dimethyl malonate, promoted by molybdenum and ligand 5, proceed by a kinetic resolution in toluene, THF, tetrahydropyran, i-PrOAc, 1,2-dichloroethane, and MeCN with k_rel of 7-16. In THF, MeCN, tetrahydropyran, and i-PrOAc using the (S,S)-5 ligand, the fast reacting (S)-carbonate enantiomer provides the branched product with high ee (97-99.5%) and branched/linear selectivity, but the ee erodes as the reaction of the slow-reacting (R)-enantiomer takes place. This implies that the rate of equilibration of the oxidative addition complexes in these solvents is competitive with the subsequent malonate displacement step. In toluene and dichloroethane, the ee and branched/linear ratios diminish during the reaction of the slow-reacting (R)-isomer, but not nearly as much as in the other solvents. This is most likely due to either an increase in the rate of equilibration of the oxidative addition complexes relative to the malonate displacement step, or vice versa. Because of the minimal stereochemical memory effect in toluene and 1,2-dichloroethane, the reactions in these solvents can be carried to completion (dynamic kinetic asymmetric transformation) and still provide product with excellent ee (>95%). The anion of dimethyl methylmalonate also reacts via a kinetic resolution, although the ee's, rates, and k_rel values differ from those of the reactions with dimethyl malonate.

Full text of DOI:10.1021/jo015871l

2762
J . Org. Chem. 2002, 67, 2762-2768
Solven t-Dep en d en t Dyn a m ic Kin etic Asym m etr ic Tr a n sfor m a tion /
Kin etic Resolu tion in Molybd en u m -ca ta lyzed Asym m etr ic Allylic
Alk yla tion s
David L. Hughes,* Michael Palucki, Nobuyoshi Yasuda, Robert A. Reamer, and
Paul J . Reider
Department of Process Research, Merck and Company, Inc., Rahway, New J ersey 07065
Received J une 26, 2001
Catalytic asymmetric alkylation reactions of branched racemic carbonates 1a and 1b with sodium
dimethyl malonate, promoted by molybdenum and ligand 5, proceed by a kinetic resolution in
toluene, THF, tetrahydropyran, i-PrOAc, 1,2-dichloroethane, and MeCN with krel of 7-16. In THF,
MeCN, tetrahydropyran, and i-PrOAc using the (S,S)-5 ligand, the fast reacting (S)-carbonate
enantiomer provides the branched product with high ee (97-99.5%) and branched/linear selectivity,
but the ee erodes as the reaction of the slow-reacting (R)-enantiomer takes place. This implies
that the rate of equilibration of the oxidative addition complexes in these solvents is competitive
with the subsequent malonate displacement step. In toluene and dichloroethane, the ee and
branched/linear ratios diminish during the reaction of the slow-reacting (R)-isomer, but not nearly
as much as in the other solvents. This is most likely due to either an increase in the rate of
equilibration of the oxidative addition complexes relative to the malonate displacement step, or
vice versa. Because of the minimal stereochemical memory effect in toluene and 1,2-dichloroethane,
the reactions in these solvents can be carried to completion (dynamic kinetic asymmetric
transformation) and still provide product with excellent ee (>95%). The anion of dimethyl
methylmalonate also reacts via a kinetic resolution, although the ee’s, rates, and krel values differ
from those of the reactions with dimethyl malonate.
In tr od u ction
tially high-value products if an asymmetric variant of the
reaction could be realized. Toward this end, stoichiomet-
ric asymmetric alkylations were described by Faller et
al. for allylmolybdenum complexes during the 1980s.3 The
first catalytic asymmetric alkylation reaction, promoted
by tungsten, was reported by Pfaltz in 1995 for the
reactions of allylic phosphates with malonate nucleo-
philes using chiral P,N-ligands.4 These reactions, while
giving high enantioselectivity, generally provided poor
regioselectivity (4/1 branched/linear ratio for most sub-
strates) and required long reaction times (1 week). A
truly practical variant was reported by Trost in 1998
using molybdenum as the metal source and the bispi-
colinamide ligand 5.5 High yields, ee’s, and regioselec-
tivities were obtained with a variety of substrates and
nucleophiles. More recently, Pfaltz6 and Kocˇovsky´7 have
designed ligands that also gave excellent results with
molybdenum.
Nucleophilic substitution reactions of allylic carboxy-
lates and carbonates, catalyzed by transition metals, have
enjoyed widespread application in organic synthesis over
the past 2 decades.1 While palladium has been the most
widely employed metal, a few studies have been aimed
at identifying other transition metals that could provide
a benefit relative to palladium. During the 1980s Trost
and co-workers pioneered the use of group VI transi-
tion metals (molybdenum and tungsten) in allylic alky-
lations, finding these metals often provided regiochem-
istry complementary to that of palladium.2 Thus, while
palladium-catalyzed reactions generally provided the
linear product, W and Mo gave predominately the
branched product (eq 1). Introduction of a chiral center
The source of molybdenum in all of these reports was
Mo(CO)3(cycloheptatriene) or Mo(CO)3(propionitrile)3, nei-
ther of which is ideal for larger scale applications due to
instability and relative unavailability. We recently com-
(3) Faller, J . W.; Mazzieri, M. R.; Nguyen, J . T.; Parr, J .; Tokunaga,
M. Pure Appl. Chem. 1994, 66, 1463. Faller, J . W.; Chao, K.-H. J . Am.
Chem. Soc. 1983, 105, 3893. Faller, J . W.; Lambert, C.; Mazzieri, M.
R. J . Organomet. Chem. 1990, 383, 161.
(4) Lloyd-J ones, G. C.; Pfaltz, A. Angew. Chem., Int. Ed. Engl. 1995,
34, 462-464.
(5) Trost, B. M.; Hachiya, I. J . Am. Chem. Soc. 1998, 120, 1104-
1105. Trost, B. M.; Hildbrand, S.; Dogra, K. J . Am. Chem. Soc. 1999,
121, 10416-10417.
(6) Glorius, F.; Pfaltz, A. Org. Lett. 1999, 1, 141-144. Glorius, F.;
Neuburger, M.; Pfaltz, A. Helv. Chim. Acta 2001, 84, 3178-3196.
(7) Malkov, A. V.; Spoor, P.; Vinader, V.; Kocˇovsky´, P. Tetrahedron
Lett. 2001, 42, 509-512. Kocˇovsky´, P.; Malkov, A. V.; Vyskocil, S.;
Lloyd-J ones, G. C. Pure Appl. Chem. 1999, 71, 1425-1433.
via the Mo- or W-catalyzed reactions made these poten-
(1) Trost, B. M.; van Vranken, D. L. Chem. Rev. 1996, 96, 395.
Hayashi, T. In Catalytic Asymmetric Synthesis; Ojima, I. Ed.; VCH
Publishers: New York, 1993; pp 325-365. Consigio, G.; Waymouth,
R. M. Chem. Rev. 1989, 89, 257.
(2) Molybdenum: (a) Trost, B. M.; Lautens, M. J . Am. Chem. Soc.
1982, 104, 5543. (b) Trost, B. M.; Lautens, M. J . Am. Chem. Soc. 1983,
105, 3344. (c) Trost, B. M.; Lautens, M. J . Am. Chem. Soc. 1987, 109,
1469. (d) Tetrahedron 1987, 43, 4817. (e) Trost, B. M.; Merlic, C. A. J .
Am. Chem. Soc. 1990, 112, 9590. Tungsten: Trost, B. M.; Hung, M.-
H. J . Am. Chem. Soc. 1983, 105, 7757; 1984, 106, 6837. Trost, B. M.;
Tometzki, G. B.; Hung, M.-H. J . Am. Chem. Soc. 1987, 109, 2176.
10.1021/jo015871l CCC: $22.00 © 2002 American Chemical Society
Published on Web 03/15/2002

Mo-Catalyzed Asymmetric Allylic Alkylations
J . Org. Chem., Vol. 67, No. 9, 2002 2763
Sch em e 1
Sch em e 2
municated that the readily available and stable precata-
lyst Mo(CO)6 was equally effective in the allylic alkylation
if appropriately activated with the ligand.8 Concurrently,
Larhed and co-workers reported that Mo(CO)6 could be
used in these reactions when promoted by microwave
irradiation.9 On the basis of the similar enantio- and
regioselectivities of the three precatalysts employed, we
concluded they all produced the same active catalyst in
the system.8
In recent years a handful of papers have been pub-
lished regarding kinetic resolution and memory effects
on Pd-catalyzed reactions of branched allylic carbonates
or carboxylates. Due to the variety of ligands and con-
ditions employed in these various studies, general mecha-
nistic conclusions cannot be drawn except that ki-
netic resolutions are quite common for these types of
reactions. On the other hand, Lehmann and Lloyd-
J ones10 found a complete stereochemical memory effect
in tungsten-catalyzed alkylations, with chiral allylic
carbonates reacting with malonate anion to give complete
stereochemical retention. This result, along with cross-
over experiments, led the authors to conclude that
catalysis by the group VI metals may involve a mecha-
nism different from that with Pd. Given both the con-
trasts (complementary regiochemistry) and similarities
(comparable substrates and ligands) between the Mo- and
Pd-catalyzed reactions, and the potential differences in
the mechanism, we have undertaken a study of the ki-
netic resolution and memory effects in the Mo-catalyzed
asymmetric alkylation, as reported herein.
Since the racemic branched carbonates 1a and 1b
produce the coupling products 3a and 3b, respectively
(Scheme 1), with high enantioselectivity, the reaction
must proceed via equilibration of a diastereomeric inter-
mediate, most likely the oxidative addition products (II
and II-ep i in Scheme 2). Thus, the overall reaction
is best described as a dynamic kinetic asymmetric
transformation.12 Trost,5 Kocˇovsky´,7 and Pfaltz6 re-
ported that the linear carbonate 2 produced higher ee’s
than the racemic branched carbonate 1a . This suggested
either that the two carbonate regioisomers were not
reacting entirely via a common transition state or that
the two enantiomers of the branched carbonate were not
fully equilibrating after the oxidative addition step. We
have now examined this question in more depth11 and
(8) Palucki, M.; Um, J . M.; Conlon, D. A.; Yasuda, N.; Hughes, D.
L.; Mao, B.; Wang, J .; Reider, P. J . Adv. Synth. Catal. 2001, 343, 46-
50.
(9) Kaiser, N.-F. K.; Bremberg, U.; Larhed, M.; Moberg, C.; Hallberg,
A. Angew. Chem., Int. Ed. 2000, 39, 3596-3598. Belda, O.; Kaiser,
N.-F.; Bremberg, U.; Larhed, M.; Hallberg, A.; Moberg, C. J . Org.
Chem. 2000, 65, 5868-5870.
(10) Lehmann, J .; Lloyd-J ones, G. C. Tetrahedron 1995, 51, 8863-
8874.

2764 J . Org. Chem., Vol. 67, No. 9, 2002
Hughes et al.
Ta ble 1. Rea ctivity a n d Selectivity for Br a n ch ed Allylic
Ca r bon a tes Rea ctin g w ith Sod iu m Dim eth yl Ma lon a tea
branched/
temp
product
linear
entry substrate
solvent
(°C) ligand ee (%) ratio (3/4)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
rac-1a THF
48 (S,S)-5 87 (S)
48 (S,S)-5 83 (S)
90 (S,S)-5 97 (S)
60 (S,S)-5 96 (S)
48 (S,S)-5 90 (S)
25
23
32
46
23
30
18
40
55
12
65
32
36
8
rac-1a CH3CN
rac-1a tolueneb
rac-1a toluene
rac-1a i-PrOAc
rac-1a tetrahydropyran 48 (S,S)-5 92 (S)
rac-1a tetrahydropyran 80 (S,S)-5 90 (S)
rac-1a ClCH2CH2Cl
(R)-1a THF
75 (S,S)-5 97 (S)
48 (R,R)-5 99 (R)
48 (S,S)-5 70 (S)
60 (R,R)-5 99.5 (R)
60 (S,S)-5 90 (S)
48 (R,R)-5 98 (R)
48 (S,S)-5 44 (S)
48 (S,S)-5 88 (S)
75 (S,S)-5 96 (S)
48 (S,S)-5 97 (S)
(R)-1a THF
(R)-1a toluene
(R)-1a toluene
(R)-1a CH3CN
(R)-1a CH3CN
rac-1b THF
15
30
35
rac-1b toluene
2
THF
F igu r e 1. Percent ee vs percent conversion for reaction of 1a
in THF at 48 °C (circles) and in toluene at 60 °C (squares).
a
Conditions: Mo(CO)3C7H8 precatalyst; sodium dimethyl ma-
lonate was prepared in situ from dimethyl malonate and NaH,
combined with the carbonate, and added to the preheated activated
catalyst mixture. Mo(CO)6 precatalyst.
Ta ble 2. Rela tive En a n tiom er ic Rea ctivity of Br a n ch ed
Ca r bon a tes Rea ctin g w ith Sod iu m Dim eth yl Ma lon a tea
b
rate
temp
(°C) ligand
constant
report that the allylic alkylation using ligand 5 proceeds
via a kinetic resolution and that the rate of diastereo-
meric equilibration relative to the subsequent steps in
the reaction is dependent on the solvent and nucleophile.
entry substrate
solvent
(s-1
)
krel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
rac-1a
rac-1a
rac-1a
rac-1a
rac-1a
rac-1a
rac-1a
rac-1a
(R)-1a
(R)-1a
(R)-1a
(R)-1a
rac-1b
rac-1b
THF
48
48
90
60
48
48
80
75
48
48
60
60
48
75
(S,S)-5
(S,S)-5
(S,S)-5
(S,S)-5
(S,S)-5
(S,S)-5
(S,S)-5
(S,S)-5
9c
8c
CH3CN
tolueneb
toluene
11c
13c
8c
i-PrOAc
tetrahydropyran
tetrahydropyran
ClCH2CH2Cl
THF
Resu lts a n d Discu ssion
8c
8c
Kin etic Resolu tion in THF . As a baseline experi-
ment, the reaction of linear carbonate 2 in THF at 48 °C
using 10 mol % Mo(CO)3(cycloheptatriene), 15 mol %
(S,S)-ligand 5, and 1.5 equiv of sodium dimethyl mal-
onate provided product 3a in >90% yield with 3% of the
linear product 4a (Table 1, entry 17). The ee of the
branched product was 97% and did not change over the
course of the reaction. On the basis of HPLC analysis,
no detectable interconversion of linear carbonate 2 and
branched carbonate 1a occurred under the reaction
conditions.
Under identical conditions, racemic branched carbon-
ate 1a was reacted with sodium dimethyl malonate in
THF. Monitoring the disappearance of starting material
via chiral HPLC revealed that one enantiomer reacted
much faster than the other, signifying that a kinetic
resolution was taking place.12 On the basis of an inde-
pendent synthesis of the enantiomerically pure isomers
of 1a , we determined that the (S)-isomer of 1a was the
fast-reacting enantiomer when using the (S,S)-ligand 5.
Due to convention changes, the (R)-product 3a that is
formed is derived from overall retention of stereochem-
10c
(R,R)-5 4.8 × 10-4
(S,S)-5 5.3 × 10-5
(R,R)-5 1.8 × 10-3
(S,S)-5 1.1 × 10-4
(S,S)-5
9d
THF
toluene
toluene
16d
THF
9c
toluene
(S,S)-5
15d
a
b
Conditions the same as for Table 1. Mo(CO)6 precatalyst.
c From eq 2, at 50% conversion. From rate constants.
d
istry (Scheme 2). Monitoring the ee and branched/linear
product ratio (3/4) as the reaction proceeded revealed that
a high product ee (98%) and a high branched/linear ratio
(35/1) was produced during the first half of the reaction
as the (S)-enantiomer reacted, but the ee eroded as the
slower reaction of the (R)-isomer took place such that an
overall 87% ee was obtained (Table 1, entry 1). This is
shown graphically in Figure 1 as a plot of % ee vs
conversion of 1a to 3a . Comparable results were obtained
for the 3-F substrate (1b) (Table 1, entry 15). Using eq
2, where c is the conversion of starting material and ee
ln(1 - c)(1 - ee)
ln(1 - c)(1 + ee)
krel
)
(2)
(11) A very preliminary test of the possibility of differential reactivity
of the two carbonate enantiomers was reported in ref 8, with the
suggestion that very little kinetic resolution was occurring under one
specific experimental condition.
is the enantiomeric excess of starting material, to deter-
mine the relative rate constants (krel) from the ee of the
starting material,13 the krel for the reaction of the two
enantiomers of 1a was found to be 9 (Table 2, entry 1).
Similarly, krel for the 3-F substrate 1b was determined
to be 9 (Table 2, entry 13).
The slow-reacting (R)-enantiomer of 1a was isolated
via a kinetic resolution on a preparative scale, which
allowed for an independent measurement of the rate
(12) The terminology for similar transformations catalyzed by Pd
is inconsistent in the literature. They are often referred to as “kinetic
resolutions” since one substrate enantiomer reacts faster than the
other. However, the goal of a kinetic resolution is to stop the reaction
midcourse to recover enantioenriched starting material as well as
product with high ee. In the current Mo-catalyzed reaction, the goal
is to carry the reaction to high conversion, and this is possible since
both enantiomers give the same product due to a dynamic equilibration
of an intermediate. Thus, the current reaction should most properly
be termed a dynamic kinetic asymmetric transformation when the
reaction is carried to completion. On the other hand, the reaction can
be run in a kinetic resolution mode, wherein enantioenriched substrate
can be recovered after partial reaction.
(13) Kagan, H. B.; Fiaud, J . C. In Topics in Stereochemistry; Allinger,
N. L., Eliel, E. L., Eds.; Interscience: New York, 1987; Vol. 14, p 249.

Mo-Catalyzed Asymmetric Allylic Alkylations
J . Org. Chem., Vol. 67, No. 9, 2002 2765
constants and ee’s for this enantiomer with both the
(R,R)- and (S,S)-ligands 5. Thus, instead of comparing
the rates of each carbonate enantiomer in the presence
of a single ligand enantiomer, the “matched” and “mis-
matched” rates were alternatively determined by use of
one carbonate enantiomer with each of the two ligand
enantiomers. These kinetic experiments verified the krel
of 9 for the reactivity of (R)- and (S)-1a (Table 2, entries
9 and 10). In addition, these experiments confirmed that
the two enantiomers of 1a produce products with quite
different ee’s and branched/linear ratios (Table 1, entries
9 and 10). For the matched case, (R)-1a reacting in the
presence of (R,R)-ligand 5, the ee was >99% and the
branched/linear ratio 55. For the mismatched case, (R)-
1a reacting in the presence of the (S,S)-ligand 5, the ee
was only 70% and the branched/linear ratio 12/1. Thus,
these reactions exhibit a modest stereochemical “memory”
effect, wherein the stereochemistry of the substrate is
partially maintained in the product. On the other hand,
there is no regiochemical memory effect. The linear
carbonate 2 provides a 97/3 ratio of branched/linear
product, while the branched carbonates produce a 98/2
ratio in the matched case and a 92/8 ratio in the
mismatched case. That is, the mismatched branched case
gives more linear product than the linear substrate. The
branched/linear ratio is thus controlled by the inherent
bias of the diastereomeric transition states, not by the
position of the carbonate leaving group in the starting
substrate. To test whether the product had any effect,
the reaction of the (R)-enantiomer of 1a was carried out
in the presence of the m-F product 3b that had an ee of
97%. There was no effect on the product ee (67%),
branched/linear ratio (11:1), or reaction rate (4.9 × 10-5
s-1), which demonstrates that the product has no impact
on these reactions.
An interpretation of these results is outlined below and
in Scheme 2. For the sake of simplicity, we assume that
the major product isomer (III) from the fast-reacting
carbonate enantiomer (I) (the matched case) arises from
an “inversion-inversion” mechanism, for which there is
literature precedence.14 Since both enantiomers give the
same major product stereochemistry (III), diastereomeric
equilibration must occur along the reaction path. How-
ever, since the two carbonate enantiomers do not give
the same ee and branched/linear ratio, the rate of
equilibration (Keq ) kS/kR) of the diastereomeric π-allyl
intermediates (II and II-ep i) must be competitive with
the malonate displacement steps (k1 through k4). If
equilibration were rapid, the ratio of diastereomers would
not be important since the diastereomer having the
pathway with the lowest energy transition state would
react preferentially (Curtin-Hammett postulate). That
is, even if kR > kS, such that the equilibrium favored the
(R)-derived oxidative addition product (II-ep i), the major
are obtained from the two carbonate enantiomers, equili-
bration must be slow and product arising from alkylation
of the nonequilibrated isomer (II-ep i) via k3 is occurring
to some extent.
Kin etic Resolu tion in Tolu en e. In our previous
paper we had determined that reaction of 1a and 1b in
toluene as solvent gave improved ee’s (97% vs 90%) and
branched/linear ratios (20/1 vs 10/1) relative to those in
THF.8 Due to the insolubility of sodium dimethyl mal-
onate in toluene, these reactions were carried out at 90
°C vs the 65 °C temperature used for THF.8 Examination
of the reaction of 1a in toluene at 90 °C using Mo(CO)6
as precatalyst revealed that a kinetic resolution was
again occurring with a krel of 11 (Table 2, entry 3).
Likewise, a kinetic resolution was operable with the m-F
substrate with a krel of 15 (Table 2, entry 14). Most
interestingly, however, for both substrates the ee and
branched/linear ratio eroded only modestly as the reac-
tion continued past the 50% conversion point, with an
ee of 99% during the early phase of the reaction and a
final ee of 96-97% (Table 1, entries 3 and 16). Similar
results were obtained with Mo(CO)3
(cycloheptatriene) as
precatalyst at a temperature of 60 °C, and is shown
graphically in comparison to the reaction in THF in
Figure 1.
To compare the reactivity of each enantiomer, the (R)-
enantiomer of carbonate 1a was reacted in the presence
of the individual (S,S)-5 and (R,R)-5 ligands. These
experiments confirmed that the matched case gave very
high ee (>99%) (Table 1, entry 11), similar to the reaction
in THF, but the mismatched case still gave reasonably
high ee (90%) (Table 1, entry 12), which is in contrast to
that in THF. In addition, the rate constants measured
for the matched and mismatched systems in toluene gave
a krel of 16 vs the value of 13 determined from eq 2 for
the reaction with the racemate (Table 2, entries 11 and
12).
In toluene, the equilibration of the two diastereomeric
complexes II and II-ep i relative to alkylation must be
faster than in THF, thus giving improved selectivity. A
possible explanation for the difference between the two
solvents may be the poor solubility of sodium dimethyl
malonate in toluene, which would cause the malonate
displacement reaction to be slower and allow for the
diastereomeric complexes to equilibrate before malonate
displacement. To test this hypothesis, the reaction in
THF was run with slow addition of sodium dimethyl
malonate, thus keeping the concentration of malonate low
such that the displacement reaction is retarded. In this
case the ee for the mismatched case ((R)-carbonate
reacting in the presence of (S,S)-5 ligand) in THF at 48
°C provided the product with 91% ee vs 70% observed
when the malonate was added at the beginning. Thus,
poor solubility of the malonate salt in toluene may
account for the improved selectivity in this solvent.
Kin etic Resolu tion in Oth er Solven ts. The allylic
alkylation of 1a with malonate in 1,2-dichloroethane,
tetrahydropyran (THP), MeCN, and i-PrOAc occurred
with kinetic resolution with krel in the 8-13 range (Table
2). The reactions in MeCN, i-PrOAc, and THP were
similar to those in THF, with the matched case giving
an ee of 99% or better and the mismatched case giving a
substantially lower ee, such that the overall ee for the
reaction of the racemic substrate was in the 83-92%
range (Table 1). In THP the reactions were studied at
both 48 and 80 °C (entries 6 and 7 of Tables 1 and 2),
product could arise via k1 as long as kR and kS
[Nu]‚k1,2,3,4. Since different ee’s and branched/linear ratios
.
(14) The alternative “retention-retention” argument would lead to
the same conclusions. Another possibility is that each enantiomer
reacts by different mechanisms, for example, the (S)-isomer by
“retention-retention” and the (R)-isomer by “inversion-retention.”
Kocˇovsky´ has demonstrated a retention-retention pathway for Mo-
catalyzed allylic alkylations. Dvorak, D.; Stary, I.; Kocˇovsky´, P. J . Am.
Chem. Soc. 1995, 117, 6130-6131. Liebeskind has demonstrated that
a change in experimental conditions can cause a change in oxidative
addition from retention to inversion, so he cautions that either
retention-retention or inversion-inversion pathways may be viable
in many cases. Ward, Y. D.; Villanueva, L. A.; Allred, G. D.; Liebeskind,
L. S. J . Am. Chem. Soc. 1996, 118, 897-898.

2766 J . Org. Chem., Vol. 67, No. 9, 2002
Hughes et al.
Ta ble 3. Rea ctivity a n d Selectivity for Allylic
Ca r bon a tes 1a a n d 2 Rea ctin g w ith Sod iu m Dim eth yl
Meth ylm a lon a tea
substituted end of the π-allyl complex to form more of
the linear product.
As with dimethyl malonate, a kinetic resolution is
operative, with the (S)-carbonate reacting faster than the
(R)-carbonate when catalyzed by the (S,S)-5 ligand. In
THF the krel is higher for dimethyl methylmalonate vs
dimethyl malonate (18 vs 9), while in toluene it is reduced
(4 vs 11). An interpretation of these results is difficult,
but it is important to note that the nucleophile is involved
in some way in the rate-determining step. This is also
true for the reaction with the linear carbonate 2, where
the dimethyl methylmalonate anion reacts about twice
as fast as dimethyl malonate in THF. These results
suggest that the malonate anion may ligate to Mo and
be a part of the catalytic species, or that the basicity of
the malonate may be important in activating the Mo-
ligand catalytic complex. Relevant to this, Trost2c has
studied the relative reactivity of unsubstituted vs sub-
stituted malonates in achiral molybdenum-catalyzed
allylic alkylations. In the most dramatic case, the un-
substituted malonate was unreactive while methylmal-
onate gave the expected alkylated product. When both
the unsubstituted and substituted malonates were present
together, no reaction occurred. Trost proposed that the
unsubstituted malonate was binding to molybdenum,
greatly attenuating its catalytic abilities, where as the
more hindered, substituted malonate could not bind and
therefore reacted as expected.
Com p a r ison of Mo- a n d P d -Ca ta lyzed Rea ction s.
The molybdenum-catalyzed allylic alkylations described
in this paper occur with kinetic resolution and show
modest stereochemical memory effects. These features
are common in a number of Pd-catalyzed reactions that
have been reported in the past few years. In this section,
we summarize the literature on Pd-catalyzed reactions
to compare and contrast those with molybdenum.
The first kinetic resolution involving Pd catalysis was
with unsymmetrical acyclic allylic acetates using ferro-
cencylphosphine ligands, reported by Hayashi and Ito in
1986.17 These reactions had krel values in the 1-14 range,
depending on the substrate. Since the substrates were
unsymmetrical and produced two regioisomeric products,
it was not clear whether these reactions proceeded via a
dynamic equilibration of the π-allyl intermediate, and the
authors did not invoke equilibration to explain the
outcome. A more straightforward example of a kinetic
resolution/dynamic kinetic asymmetric transformation
was reported by Osborn in 1998.18 In this case, using a
chiral bisphosphine ligand, cyclohexenyl acetate as sub-
strate, malonate as nucleophile, and THF and dichlo-
romethane as solvents, krel values in the 2.4-8.1 range
were measured. This was the first unambiguous case that
demonstrated a selective activation of a racemic allylic
substrate could occur with Pd catalysis. In these reac-
tions, the product ee remained unchanged throughout the
course of the reactions, indicating that rapid dynamic
equilibration of the diastereomeric π-allyl complexes was
occurring prior to reaction with the nucleophile. There-
fore, these reactions exhibited no memory effects. Osborn
found similar results with the acyclic substrate 4-acetoxy-
2-pentene.18
branched/
temp
product
linear
entry substrate solvent (°C) ligand
ee (%) ratio (3/4) krel
1
2
3
4
5
6
7
rac-1a
rac-1a
(R)-1a
(R)-1a
(R)-1a
(R)-1a
2
THF
toluene
THF
48
60
48
48
60
60
48
(S,S)-5
(S,S)-5
(R,R)-5 99 (R)
(S,S)-5 40 (S)
(R,R)-5 99 (R)
80 (S)
89 (S)
9
9
36
5
21
6
20
18b
4b
19c
THF
toluene
toluene
THF
(S,S)-5
(S,S)-5
62 (S)
95 (S)
a
Conditions: Mo(CO)3C7H8 precatalyst; sodium dimethyl mal-
onate was prepared in situ from dimethyl malonate and NaH,
combined with the carbonate, and added to the preheated activated
b
catalyst mixture. From eq 2. c From rate constants.
with minimal differences in the ee and krel at the two
temperatures. The reaction in dichloroethane mirrored
that in toluene, with the overall ee for the racemic
carbonate being 97% (Table 1, entry 8), indicating the
mismatched case was still giving a high ee in the
reaction. Like toluene, the low solubility of sodium
dimethyl malonate anion in this solvent may be respon-
sible for the increased ee.
Kin et ic R esolu t ion w it h Dim et h yl Met h ylm a l-
on a te. The anion of dimethyl methylmalonate reacts
with branched carbonate 1a to provide the corresponding
branched and linear products, as summarized in Table
3. Reaction with racemic carbonate in THF and toluene
(entries 1 and 2) reveals that the ee’s are lower vs those
with dimethyl malonate (80% vs 87% in THF and 89%
vs 96% in toluene). The lower ee’s result from a greater
memory effect with dimethyl methylmalonate, as shown
in entries 3-6 in Table 3. In both solvents, the matched
case of the (R)-carbonate catalyzed by the (R,R)-5 ligand
provides product with 99% ee, but the mismatched case
gives ee’s of only 40% in THF and 62% in toluene. Using
Scheme 2 as a model for the reaction pathway, the results
with dimethyl methylmalonate indicate that the equili-
bration of II-ep i to II (kS) has been slowed relative to
the reaction to form product (k3). Two of the factors which
control k3 are the basicity and steric effects of the anion.
Comparing dimethyl methylmalonate vs dimethyl mal-
onate, addition of the methyl group adds steric hindrance,
which should reduce the nucleophilicity, yet previous
work indicates displacement reactions with carbon nu-
cleophiles are relatively insensitive to small steric ef-
fects.15 On the other hand, addition of a methyl group
increases the basicity of the anion by 1.5-2 pKa units,16a
and this in turn will increase the nucleophilicity (k3),
leading to the increased rate of k3 vs the equilibration
rate. Previous work has shown the more substituted
enolate is more reactive in alkylations.16b
Regarding the branched/linear ratio, the dimethyl
methylmalonate anion gives an increase in the amount
of linear product vs dimethyl malonate for both the
branched carbonate 1a (11% vs 3%) and the linear
carbonate 2 (5% vs 3%). This is likely due to a steric
effect, with the more sterically demanding dimethyl
methylmalonate reacting toward the least hindered un-
(15) J ackman, L. M.; Lange, B. C. Tetrahedron 1977, 33, 2737-2769.
(16) (a) The pKa in Me2SO of R-methylacetylacetone is 15.07 vs 13.33
for acetylacetone. The pKa of ethyl R-methyacetoacetate is 15.75 vs
14.26 for ethyl acetoacetate. (Olmstead, W. N.; Bordwell, F. G. J . Org.
Chem. 1980, 45, 3299-3305.) (b) Caine, D.; Huff, R. Tetrahedron Lett.
1967, 3399.
(17) Hayashi, T.; Yamamoto, A.; Ito, Y. Chem. Commun. 1986,
1090-1092.
(18) Ramedeehul, S.; Dierkes, P.; Aguado, R.; Kamer, P. C. J .; van
Leeuwen, P. W. N. M.; Osborn, J . A. Angew. Chem., Int. Ed. 1998, 37,
3118-3121.

Mo-Catalyzed Asymmetric Allylic Alkylations
J . Org. Chem., Vol. 67, No. 9, 2002 2767
The first suggestion of a memory effect in Pd-catalyzed
alkylations came from the Fiaud laboratories, where
racemic and optically active cyclohexenyl acetate gave
different ee’s, although the overall ee’s were quite low
(<20%).19 The authors invoked an (o-allyl)palladium
complex to account for the results. A careful study by
Trost and Bunt in 1996 conclusively demonstrated a
memory effect was operative in the reactions of cyclo-
pentenyl acetates and carbonates. These authors sug-
gested intimate ion pairs could be the factor behind the
memory effects.20 More recently, the cyclopentenyl system
has been studied in depth by Lloyd-J ones and Stephen.
They reported that chloride ion has a large effect on the
kinetic resolution of Pd-catalyzed reactions of cyclopen-
tenyl pivalate in THF with sodium dimethyl malonate
as nucleophile.21 In the presence of chloride, the mis-
matched rate is enhanced, leading to a diminution of the
krel for this system. They concluded that chloride coor-
dination to Pdo resulted in a more reactive and less
selective catalyst.21 Incisive studies by Lloyd-J ones and
co-workers of memory effects in Pd-catalyzed reactions
using labeled substrates demonstrated that memory
effects were dependent on the nature of the ligand, steric
effects in the allylic substrate, and the counterion.22 The
effect of counterions on memory effects and kinetic
resolution has also been noted by Togni26 and Trost,27
which led them to conclude ion pairs play a key role in
controlling the outcome of these reactions. Finally,
Zhang23 and Gilbertson24 have recently reported kinetic
resolutions and significant memory effects in Pd-cata-
lyzed allylic alkylations, while Trost and co-workers have
demonstrated a near perfect kinetic resolution/kinetic
asymmetric transformation with a cyclohexenyl sub-
strate.25
and can be explained using the conventional π-allyl
mechanism.
Con clu sion
The molybdenum-catalyzed allylicalkylation ofbranched
carbonates 1a and 1b with sodium dimethyl malonate
in the presence of ligand 5 can be carried out either as a
dynamic kinetic asymmetric transformation (100% con-
version of starting material) or as a kinetic resolution
(partial reaction and isolation of enantioenriched starting
material). In all solvents investigated, the reaction
proceeds via a kinetic resolution with krel values in the
7-16 range. While not extremely large, these krel values
are of a magnitude to effectively carry out these reactions
in the resolution mode to obtain recovered starting
material in high ee (<95%) at 60-65% conversions, as
well as produce product with an excellent ee. The
effectiveness of the dynamic kinetic asymmetric trans-
formation is dependent on the solvent. In THF, THP,
i-PrOAc, and MeCN, a significant stereochemical memory
effect is operative, with the slow-reacting enantiomer
providing product in much lower ee than the fast-reacting
enantiomer. Thus, a lower product ee for reactions of the
racemic branched carbonate (80-90%) vs the linear
carbonate (>97%) results when the reactions are carried
to completion. In toluene and 1,2-dichloroethane, even
though the two enantiomers of the carbonate differ in
reactivity by an order of magnitude, the memory effect
is minimal, and therefore the product ee’s in these
solvents are excellent (95%) for reactions carried to
completion. Spectroscopic and kinetic experiments, along
with studies using designed ligand probes, are in progress
to further elucidate the mechanism. These studies will
be disclosed shortly.
It is clear from these examples that kinetic resolu-
tion is a common and perhaps the preferred pathway for
Pd-catalyzed reactions of branched allylic acetates, and
that memory effects are also common and variable.
Kinetic resolution and memory effects are governed by
a variety of factors, including ligand structure, solvent,
counterion effects, nucleophile, and substrate struc-
ture. In the single study in which memory effects were
studied in a non-Pd-catalyzed reaction, a complete memory
effect was observed for the alkylation of allylic carbonates
using malonate anion catalyzed by tungsten.10 This
result, combined with crossover experiments, led the
authors to speculate that these reactions may not be
occurring by the widely accepted π-allyl mechanism that
is invoked in Pd catalysis.10 Our results with the Mo-
catalyzed reactions reported in this paper, wherein
modest memory effects are observed, mirror those of Pd
Exp er im en ta l Section
Gen er a l P r oced u r es. The molybdenum-catalyzed reac-
tions were carried out under an atmosphere of argon after
thorough degassing of the solvent. NMR spectra were recorded
on Bruker 300, 400, and 500 MHz instruments. All solvents
were dried with molecular sieves prior to use. The molybde-
num precatalysts were obtained from Aldrich and Strem. The
synthesis and characterization of the allylic carbonates and
reaction products have been described previously.8
HP LC Assa ys. Rever sed -P h a se Assa y. To determine
conversion, the reactions were monitored by reversed-phase
HPLC using a Zorbax SB-phenyl column with the following
conditions: eluent, 28/72 MeCN/0.1% aqueous H3PO4 isocratic
for 40 min and then gradient to 70/30 over 5 min, column
temperature 45 °C, flow rate 1.5 mL/min, detection at 220 nm.
Elution times: branched phenyl carbonate (1a ), 27 min;
branched m-fluorophenyl carbonate (1b), 32.5 min; linear
phenyl carbonate (2), 28 min; branched phenyl product (3a ),
34.5 min; branched m-fluorophenyl product (3b), 42.5 min;
linear phenyl product (4a ), 45 min; linear m-fluorophenyl
product (4b), 46 min.
Ch ir a l Assa y. A chiral assay was developed which sepa-
rated both enantiomers of the branched phenyl and m-
fluorophenyl carbonates as well as the corresponding malonate
products. Either an (R,R)- or (S,S)-Whelko column could be
used; the enantiomer elution order was reversed on the two
columns and provided a check that the correct enantiomer
peaks were being observed and that there were no interfering
peaks. The eluent was 99% hexanes/1% i-PrOH isocratic, 220
nm detection, 1.5 mL/min flow rate , 25 °C. Reaction samples
for assay were quenched into 10% i-PrOH/90% hexanes and
an equal volume of 0.5 N HCl, and the organic layer was
assayed. Elution times for the (S,S)-Whelko column: branched
(19) Fiaud, J . C.; Malleron, J . L. Tetrahedron Lett. 1981, 22, 1399-
1402.
(20) Trost, B. M.; Bunt, R. C. J . Am. Chem. Soc. 1996, 118, 235-
236.
(21) Lloyd-J ones, G. C.; Stephen, S. C. Chem. Commun. 1998, 2321-
2322.
(22) Lloyd-J ones, G. C.; Stephen, S. C. Chem.sEur. J . 1998, 4,
2539-2549. Lloyd-J ones, G. C.; Stephen, S. C.; Murray, M.; Butts, C.
P.; Vyskocil, S.; Kocˇovsky´, P. Chem.sEur. J . 2000, 6, 4348-4357.
(23) Longmire, J . M.; Wang, B.; Zhang, X. Tetrahedron Lett. 2000,
41, 5435-5439.
(24) Gilbertson, S. R.; Lan, P. Org. Lett. 2001, 3, 2237-2240.
(25) Trost, B. M.; Hembre, E. J . Tetrahedron Lett. 1999, 40, 219-
222. Trost, B. M.; Patterson, D. E.; Hembre, E. J . Chem.sEur. J . 2001,
7, 3768-3775.
(26) Burckhardt, U.; Baumann, M.; Togni, A. Tetrahedron: Asym-
metry 1997, 8, 155-159.
(27) Trost, B. M.; Toste, F. D. J . Am. Chem. Soc. 1999, 121, 4545-
4554.

2768 J . Org. Chem., Vol. 67, No. 9, 2002
Hughes et al.
phenyl carbonate (1a ), (R)-isomer at 4.4 min, (S)-isomer at 6.5
min; branched m-fluorophenyl carbonate (1b), (R)-isomer at
3.9 min, (S)-isomer at 5.5 min; linear phenyl carbonate (2),
7.5 min; branched phenyl product (3a ), (S)-isomer at 8.1 min,
(R)-isomer at 8.9 min; branched m-fluorophenyl product (3b),
(S)-isomer at 7.3 min, (R)-isomer at 8.1 min; linear phenyl
product (4a ), 14 min; linear m-fluorophenyl product (4b), 13
min.
P r ep a r a tive-Sca le Kin etic Resolu tion of 1a . To a 25-
mL Schlenk tube were added (S,S)-ligand 5 (528 mg, 1.62
mmol) and THF (14 mL). While being stirred, the solution was
degassed by two vacuum/argon backfill cycles. Tricarbonyl-
molybdenum cycloheptatriene (406 mg, 1.49 mmol) was added,
and two additional vacuum/argon backfill cycles were carried
out. The mixture was then warmed to 48 °C for 30 min. To a
250 mL flask were added dimethyl malonate (6.24 g, 47.3
mmol) and THF (85 mL). To the stirring solution was added
60% NaH (1.57 g, 39.3 mmol) portionwise over 5 min (vigorous
hydrogen evolution). The homogeneous solution was degassed
by three vacuum/argon backfill cycles and then warmed to 48
°C. The catalyst mixture was added to the sodium dimethyl
malonate solution at 48 °C, the resulting mixture was stirred
for 5 min, and then the branched phenyl carbonate 1a (4.89
g, 25.5 mmol) was added. After 30 min the reaction was
quenched by addition of water (40 mL) and hexanes (30 mL).
Assay at this point indicated 63% conversion. The layers were
separated, and the organic layer was washed with brine,
filtered through a pad of sodium sulfate, and concentrated to
a red oil. The carbonate was separated from product by flash
chromatography, eluting with 7% EtOAc/hexanes. The (R)-
carbonate 1a (1.81 g, 93% purity) was recovered as a colorless
oil, with an ee of 94%. The recovered malonate product, which
eluted after the carbonate, had an ee of 97%.
Typ ica l Rea ction . To a 25 mL Schlenk tube were added
(S,S)-ligand 5 (65.4 mg, 0.20 mmol) and 3.0 mL of THF. While
being stirred, the solution was degassed by three vacuum/
argon backfill cycles. Tricarbonylmolybdenum cycloheptatriene
(55.7 mg, 0.20 mmol) was added, and three additional vacuum/
argon backfill cycles were carried out. The mixture was then
warmed to 48 °C for 30 min. The catalyst solution turned dark
purple with some solids precipitating. To another 25 mL
Schlenk tube were added dimethyl malonate (459 mg, 3.48
mmol) and THF (7.5 mL). With stirring, NaH (60%, 122 mg,
3.05 mmol) was added in one portion, resulting in vigorous
hydrogen evolution. After 5 min when most hydrogen evolution
had ceased, the solution was degassed by two vacuum/argon
backfill cycles, and then the linear phenyl carbonate 2 (391
mg, 2.04 mmol) was added and degassed twice. The mixture
remained homogeneous. The solution was warmed to 48 °C
for 5 min and then transferred to the activated catalyst
mixture via gastight syringe to initiate the reaction. Samples
were taken via syringe to monitor the reaction and assayed
as outlined above. In kinetic runs with carbonate 2 and the
individual enantiomers of 1a , the disappearance of the carbon-
ate followed good first-order kinetics for >3 half-lives. The
details of the kinetics will be described in a future paper.
Ack n ow led gm en t. We thank Dr. Chris Welch for
developing the chiral HPLC assay and Professor Barry
M. Trost for valuable discussions.
J O015871L