Delineating origins of stereocontrol in asymmetric Pd-catalyzed α-hydroxylation of 1,3-ketoesters

Article abstract of DOI:10.1021/jo1002906

Systematic studies of reaction conditions and subsequent optimization led to the identification of important parameters for stereoselectivity in the asymmetric α-hydroxylation reaction of 1,3-ketoesters. Enantioselectivities of up to 98% can be achieved for cyclic substrates and 88% for acyclic ketoesters. Subsequently, the combination of cyclic/acyclic ketoester, catalyst, and oxidant was found to have a profound effect on reaction rates and turnover-limiting steps. The stereochemistry of the reaction contradicts that observed for other similar electrophilic substitution reactions. This was rationalized by transition-state modeling, which revealed a number of cooperative weak interactions between oxidant, ligand, and counterion, together with C-H/π interactions that cumulatively account for the unusual stereoselectivity.

Full text of DOI:10.1021/jo1002906

pubs.acs.org/joc
Delineating Origins of Stereocontrol in Asymmetric Pd-Catalyzed
r-Hydroxylation of 1,3-Ketoesters
Alexander M. R. Smith, Henry S. Rzepa, Andrew J. P. White, Denis Billen,† and
King Kuok (Mimi) Hii*
Department of Chemistry, Imperial College London, Exhibition Road, South Kensington, London
SW7 2AZ, United Kingdom. †Pfizer Animal Health, 333 Portage Street, Kalamazoo, MI 49001.
mimi.hii@imperial.ac.uk
Received February 22, 2010
w This paper contains enhanced objects available on the Internet at http://pubs.acs.org/joc.
n
Systematic studies of reaction conditions and subsequent optimization led to the identification
of important parameters for stereoselectivity in the asymmetric R-hydroxylation reaction of
1,3-ketoesters. Enantioselectivities of up to 98% can be achieved for cyclic substrates and 88% for
acyclic ketoesters. Subsequently, the combination of cyclic/acyclic ketoester, catalyst, and oxidant
was found to have a profound effect on reaction rates and turnover-limiting steps. The stereochem-
istry of the reaction contradicts that observed for other similar electrophilic substitution reactions.
This was rationalized by transition-state modeling, which revealed a number of cooperative weak
interactions between oxidant, ligand, and counterion, together with C-H/π interactions that
cumulatively account for the unusual stereoselectivity.
Introduction
products that are valuable intermediates for organic synth-
esis (Scheme 1, eq 2).2
In a Rubottom oxidation reaction, a silyl enol ether
undergoes epoxidation by a peracid to furnish an unstable
intermediate that hydrolyzes to an R-hydroxyl ketone
(Scheme 1, eq 1).1 For 1,3-dicarbonyl compounds with
sufficiently acidic R-hydrogens, spontaneous deprotonation
occurs upon chelation to Lewis acid catalysts. The resultant
enolate complex reacts with electrophilic oxidants, including
m-CPBA, H2O2, and O2, to furnish densely functionalized
A significant body of work by Davis and co-workers has
identified a family of oxaziridines (Figure 1) as electrophilic
oxidants for the hydroxylation of enolates,3,4 including the
development of a camphorsulfonyl oxaziridine 3 as a chiral
reagent, which has been employed to great effect in the
synthesis of several natural products.5
(2) Christoffers, J.; Baro, A.; Werner, T. Adv. Synth. Catal. 2004, 346,
143.
(3) Davis, F. A.; Chen, B. C. Chem. Rev. 1992, 92, 919.
(4) Davis, F. A.; Liu, H.; Chen, B.-C.; Zhou, P. Tetrahedron 1998, 54,
10481.
(1) Rubottom, G. M.; Vazquez, M. A.; Pelegrin, D. R. Tetrahedron Lett.
1974, 15, 4319.
DOI: 10.1021/jo1002906
r
Published on Web 03/25/2010
J. Org. Chem. 2010, 75, 3085–3096 3085
2010 American Chemical Society

Smith et al.
JOCArticle
SCHEME 1. R-Hydroxylation reactions
FIGURE 2. Chiral catalysts for asymmetric hydroxylation
reactions.
substrates. Details of this work are described herein, including
the identification of key elements for reactivity and stereocon-
trol and the proposal of a novel transition state for R-hetero-
functionalization of 1,3-ketoesters.
FIGURE 1. Oxaziridine reagents developed by Davis et al.
In comparison, the development of catalytic asymmetric
hydroxylation reactions has emerged only in recent years.
The precedent was set by Togni and Mezzetti in 2004, with
the report of TADDOL-Ti complex 4 as a chiral Lewis acid
catalyst (Figure 2).6 Used in combination with racemic
N-sulfonyloxaziridine 1a (Ar = 4-O2NC6H4, R = Ph),
R-hydroxylation of cyclic and acyclic 1,3-ketoesters can be
realized, achieving an impressive 94% ee for the hydroxyla-
tion of tert-butyl cyclopentanone-2-carboxylate. A few years
later, Toru and Shibata reported the use of DPFOX complexes
of Zn and Ni for the hydroxylation of oxindoles and cyclic
ketoesters using saccharin-derived oxaziridine 2 as oxidant.7
The use of cinchona alkaloid derivatives as chiral organo-
catalysts for the R-hydroxylation of 1,3-ketoesters by organic
peroxides was first disclosed in the patent literature by Dupont
scientists8 and also later by Jørgensen and co-workers,9 where
moderately high ee’s can be attained (up to 80% using catalyst
5). Recently, a more successful approach was recorded by
using a combination of the chiral phosphoric acid 6 and
PhNO,10 where ee’s of up to 98% can be achieved. With the
exception of the Ti catalyst 4, most of these catalysts were only
effective for the hydroxylation of cyclic 1,3-ketoesters: acyclic
substrates were either unreported or proceeded with much
diminished yields and/or stereoselectivity.
Results and Discussion
We previously demonstrated the dicationic Pd(II) com-
plex (R)-I to be a highly effective catalyst for the enantiose-
lective addition of aromatic amines to chelating Michael
acceptors.12-14 Prior to that, seminal work by Sodeoka and
co-workers exploited the chelation of 1,3-ketoesters to cat-
alyst I as a means of stereocontrol in a series of aldol,
Mannich, Michael, and related reactions, where excellent
levels of enantioselectivity can be achieved.15 Concurrently,
highly enantioselective R-fluorination and R-amination of
1,3-ketoesters have also been reported.16-18 In light of this,
the absence of enantioselective R-hydroxylation reactions
was striking. Given that this particular class of C-O bond-
forming reactions is not well-demonstrated in asymmetric
catalysis, we decided to explore the application of dicationic
Pd(II) catalysts in these systems.
Screening of Oxidants. The cyclic 1,3-ketoester methyl 2-
carboxylate indanone (7a) was adopted as a model substrate
in the screening of oxidants (Scheme 2, Table 1). Reactions
were performed initially with four oxaziridines: racemic 1b
and 1c and both isomers of 3. While the first two oxidants
afforded similar results (entries 1 and 2), neither isomer of
the chiral camphor oxaziridine was reactive (entry 3), sug-
gesting that the size of the oxidant is important for reactivity.
Subsequently, other (smaller) electrophilic oxidants were
examined. Peroxides were less reactive and selective than
m-CPBA (entries 4-6), although opposite stereoselectivity
was observed when tert-butyl peroxide was used.19 When
Recently, we reported the use of dimethyldioxirane
(DMD) as an effective oxidant for R-hydroxylation of 1,
3-ketoesters.11 In the presence of a dicationic BINAP-
palladium catalyst, unprecedented levels of enantioselectivities
can be obtained with carbocyclic, heterocyclic, and acyclic
(5) See, for example: (a) Hauser, F. M.; Xu, Y. J. Org. Lett. 1999, 1, 335.
(b) Davies, S. G.; Epstein, S. W.; Garner, A. C.; Ichihara, O.; Smith, A. D.
Tetrahedron: Asymmetry 2002, 13, 1555. (c) Abraham, E.; Candela-Lena,
J. I.; Davies, S. G.; Georgiou, M.; Nicholson, R. L.; Roberts, P. M.; Russell,
A. J.; Sanchez-Fernandez, E. M.; Smith, A. D.; Thomson, J. E. Tetrahedron:
Asymmetry 2007, 18, 2510. (d) Tagami, K.; Nakazawa, N.; Sano, S.; Nagao,
Y. Heterocycles 2000, 53, 771.
(12) Hii, K. K. Pure Appl. Chem. 2006, 78, 341.
(13) Guino, M.; Phua, P. H.; Caille, J. C.; Hii, K. K. J. Org. Chem. 2007,
72, 6290.
(6) Toullec, P. Y.; Bonaccorsi, C.; Mezzetti, A.; Togni, A. Proc. Natl.
Acad. Sci. U.S.A. 2004, 101, 5810.
(7) Ishimaru, T.; Shibata, N.; Nagai, J.; Nakamura, S.; Toru, T.;
Kanemasa, S. J. Am. Chem. Soc. 2006, 128, 16488.
(8) Taylor, E. G.; Birch, L. US Patent WO03/040083, PCT/US02/35615,
2003.
(9) Acocella, M. R.; Mancheno, O. G.; Bella, M.; Jorgensen, K. A. J. Org.
Chem. 2004, 69, 8165.
(14) Phua, P. H.; Mathew, S. P.; White, A. J. P.; de Vries, J. G.;
Blackmond, D. G.; Hii, K. K. Chem.;Eur. J. 2007, 13, 4602.
(15) Sodeoka, M.; Hamashima, Y. Bull. Chem. Soc. Jpn. 2005, 78, 941.
(16) Hamashima, Y.; Suzuki, T.; Takano, H.; Shimura, Y.; Tsuchiya, Y.;
Moriya, K.; Goto, T.; Sodeoka, M. Tetrahedron 2006, 62, 7168.
(17) Hamashima, Y.; Sodeoka, M. Synlett 2006, 1467.
(18) Kang, Y. K.; Kim, D. Y. Tetrahedron Lett. 2006, 47, 4565.
(19) Hydrogen bonding can counteract steric effects, leading to opposite
selectivities in epoxidation reactions; see: (a) Adam, W.; Pastor, A.; Peters,
K.; Peters, E. M. Org. Lett. 2000, 2, 1019. (b) Adam, W.; Zhang, A. M. Eur. J.
Org. Chem. 2004, 147.
(10) Lu, M.; Zhu, D.; Lu, Y. P.; Zeng, X. F.; Tan, B.; Xu, Z. J.; Zhong,
G. F. J. Am. Chem. Soc. 2009, 131, 4562.
(11) Smith, A. M. R.; Billen, D.; Hii, K. K. Chem. Commun. 2009, 3925.
3086 J. Org. Chem. Vol. 75, No. 9, 2010

Smith et al.
JOCArticle
SCHEME 2. Initial Model System of Study
phosphine moiety (xylyl-Phanephos) has a detrimental effect
(13% ee). Interestingly, the introduction of a p-methoxy
group on the paracyclophane enhanced the ee (MeO-Para-
Phos vs xylyl-Phanephos), indicating that non-C2-symmetric
ligands are potentially useful. Planar-chiral diphospholanes
based on the ferrocene backbone (FerroTanes) displayed
higher levels of selectivity than DuPhos-containing stereo-
genic carbons. The study of C2-symmetric ligands concluded
with DIOP and SpirOP, which gave practically racemic
products.
Next, non-C2-symmetric ligands were examined: Dipho-
sphines such as Walphos and Josiphos offered low levels of
selectivity (11% and 8%, respectively), while Stanphos and
BoPhoz containing electronically dissimilar P-donors dis-
played a level of selectivity similar to that of BINAP (25 and
30% ee, respectively). Enantioselectivities obtained by
ligands with different donor groups (P-N) are particularly
interesting. While phosphinooxazoline (PHOX) ligands dis-
played low levels of enantioselectivity (<10%), higher
selectivity were achieved with aminophosphine ligands such
as Taniaphos (T001, 45% ee) and MandyPhos (M001, 50%
ee). The selectivity of the latter is dependent on the M/L ratio,
which was higher at 2:1 compared to 1:1, suggesting that
P,N-chelation imparts greater stereodifferentiation. Once
again, increasing the electronic/steric bulk of the P-substituent
led to a decrease in these values (T002 and M002).
The study led us to identify ligand scaffolds that can enable
higher levels of selectivity than BINAP, particularly those
that contain planar chirality, such as PhanePhos and
P,N-ligands Taniaphos and Mandyphos. Taking the avail-
ability and cost of these ligands into account, only Phane-
phos was pursued further in the present study. Subsequently,
catalyst (R)-II (Figure 4) was prepared, and its catalytic
activity was compared with (R)-I in our further work.
Reaction Conditions and Effect of Ester Substituent. The
performance of catalyst II was examined in the hydroxyla-
tion of a series of 1,3-ketoesters derived from indanone
(7a-f, Scheme 3), containing different ester substituents
(Table 2, entries 1-6). By lowering the reaction temperature,
there ensued a small but detectable increase in the ee of
product 8a, from 56 to 69% at -20 °C, at the expense of rate
of catalytic turnover (entries 1-4). However, increasing the
steric bulk of the ester did not lead to an increase in
enantioselectivity as anticipated (entries 5-10). Instead, a
decrease to 38% ee was observed with the tert-butyl ester
(entry 12). The same result was obtained using catalyst I,
where the hydroxylation of the methyl and tert-butyl esters
proceeded with very low levels of enantioselectivity (entries
13-15).
TABLE 1. Hydroxylation of 7a by Different Oxidants (Scheme 2)a
entry
oxidant
timeb/h temp/°C conv/%c
eed/%
1
2
1b (Ar = Ph, R = p-tol)
1c (Ar = 4-NO2C6H4,
R = p-tol)
18
18
rt
rt
100 (99) 30 (S)-(þ)
100 (95) 26 (S)
3
4
5
6
7
8
3e
18
18
18
18
18
rt
rt
0
0
f
H2O2
t-BuOOHg
m-CPBA
Oxone
rt
27 (24)
71 (64)
13 (R)-(-)
16 (S)
rt
rt
100 (24)h 13 (S)
100 (99) 34 (S)
DMD
0.5
-20
aTypical reaction conditions: 1,3-ketoester (0.123 mmol), (R)-I
(13.1 mg, 10 mol %), and oxidant (1.2 equivalent) in CH2Cl2
(0.5 mL). bUnoptimized. cDetermined by 1H NMR, isolated yield given
in parentheses. dDetermined by chiral HPLC. Absolute configuration of
8a was assigned by comparing its optical rotation with literature data.
f
eBoth (1R)-(-) and (1S)-(þ) isomers were tested. 30% aqueous solu-
tion. g80% aqueous solution. hPlagued with side products.
Oxone was employed, the precursor was consumed quickly
to give a capricious mixture, from which the expected
product was isolated in a low yield (entry 7). Last, but not
least, the use of dimethyldioxirane (DMD) was also assessed.
As the reagent is highly electrophilic, the reaction was
conducted at -20 °C to suppress the uncatalyzed reaction
(vide infra), where a level of enantioselectivity comparable to
that achieved by 1b was obtained. Thus, this study identified
oxaziridine 1b and DMD as the best oxidants, as they offered
the cleanest and highest conversions.
Ligand Screening. Previously, we have successfully em-
ployed Pd(OTf)2 2H2O as a catalyst precursor for the gen-
3
eration of dicationic diphosphine-palladium complexes in
situ, which facilitated high-throughput ligand screening
studies.20 Accordingly, this was employed in the current
work for the rapid assessment of ligand effects in these
hydroxylation reactions. In total, 21 phosphine-containing
ligands with different structural and electronic features
(figure 3) were tested using oxaziridine 1b as the oxidant,
as the reactions can be conducted at room temperature.21
Although individual rates were not monitored, reactions
were essentially complete within 18 h. Among the C2-sym-
metric ligands, chiral diphosphines with axial chirality
(BINAP, MeO-Cl-PHEP, and P-Phos) offered virtually
identical enantioselectivities (28-30%). In comparison, a
much higher level of selectivity was offered by the planar-
chiral (R)-Phanephos (53% ee of the opposite enantiomer
was obtained), while increasing the steric bulk of the
Attributing the lack of correlation between the enantios-
electivity and the size of the ester substituent to a poor fit
of the oxidant with the chiral pocket, the reactions were
repeated using DMD as the oxidant, which had previously
given comparable results to 1b in the hydroxylation of 7a
(Table 1, entries 2 and 8). The use of DMD can be
advantageous in that both the reagent and its side product
(acetone) are volatile, which can be evaporated to afford
the product in high purity without the need for column
chromatography. As it is a powerful oxidant, it is important
to suppress competitive uncatalyzed reactions. Thus,
the reaction of 7a and DMD was examined at different
temperatures in the absence of the catalyst, establishing the
(20) Phua, P. H.; White, A. J. P.; de Vries, J. G.; Hii, K. K. Adv. Synth.
Catal. 2006, 348, 587.
(21) Control experiments performed using Pd(OTf)2.2H2O in the absence
of ligands led to a capricious reaction, with little product formation.
J. Org. Chem. Vol. 75, No. 9, 2010 3087

Smith et al.
JOCArticle
FIGURE 3. Ligand screening. Reaction conditions: 10 mol % of Pd(OTf)2 2H2O, 11 mol % of ligand, CH2Cl2, 30 min, then 7a (1 equiv) and
3
1b (1.2 equiv) were added successively, rt. All reactions were complete in 18 h (1H NMR spectroscopy). ee values were determined by chiral
HPLC; the stereochemistry of major enantiomer is provided in parentheses. (b) 5.5 mol % of ligand.
TABLE 2. Effect of Catalyst and Ester Substituent Using 1b as
Oxidanta
entry ketoester catalyst temp/°C time/h % convb
% eec
1
2
7a
II
rt
0
4
6
100
100
56 (R)-(-)d
52e
3
-10
-20
rt
7
100
100
61e
69e
4
18
4
5
7b
7c
7d
7e
II
II
II
II
100 (62)
100 (79)
100 (63)
100 (73)
100 (67)
100 (69)
100 (14)
100 (66)
100 (99)
100
50 (-)
52
FIGURE 4. Isolated Phanephos complex (R)-II.
6
-20
rt
20
4
7
49 (-)
54
SCHEME 3. Hydroxylation of 2-Alkyl Carboxylate Indanones
8
-20
rt
20
4
9
49 (R)-(--)d
63
10
11
12
13
14
15
-20
rt
20
4
25 (R)-(-)d
38
-20
rt
20
4
7a
7e
I
I
30 (S)-(þ)d
rt
2
24
5
-30
100
10 (S)-(þ)
threshold of -20 °C for the uncatalyzed process (Table 3,
entries 1-3).
aReaction conditions: ketoester 7a-e(0.123mmol),catalyst(10mol%),
1b (1.2equiv), CH2Cl2. bConversion determined by 1H NMR spectroscopy,
isolated yields (after column chromatography) given in parentheses.
cDetermined by chiral HPLC. dAbsolute configuration assigned by com-
parison with literature data. e2.2 equiv of oxidant.
Employing catalyst I under these new conditions, a corre-
lation between enantioselectivity and size of the ester sub-
stituent can be clearly established, starting from 34% ee for
the methyl ester 7a and raising to 86% ee for the tert-butyl
ester 7e (Table 3, entries 4-9). Exceptions were found with
benzyl and adamantyl esters, which afforded lower ee’s.
Conversions were also extremely rapid; even at 5 mol %
catalyst loading, reactions were complete within 30 min at
-20 °C (entries 11-16). In contrast, the Phanephos-ligated
catalyst II furnished a much lower level of enantioselectivity,
and no clear correlation between the enantioselectivity and
the ester substituent was observed (entries 17-20).
This part of the study revealed three key factors for high
enantioselectivity: while the ligand effect is important, the
choice of oxidant can affect how the particular system
responds to changing sizes of the ester substituent.
3088 J. Org. Chem. Vol. 75, No. 9, 2010

Smith et al.
JOCArticle
TABLE 3. Effect of Catalyst and Ester Substituent Using DMD as
Oxidanta
entry ketoester catalyst temp/°C time/h % convb
% eec
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
7a
7a
none
rt
-20
-78
-20
-78
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
0.17 100
1
5
6
0
I (10)
0.5
5
100 (99) 39 (S)-(þ)d
67
47
7b
7c
7d
7e
7f
7a
7b
7c
7d
7e
7f
I (10)
I (10)
I (10)
I (10)
I (10)
I (5)
I (5)
I (5)
I (5)
I (5)
4
4
4
100 (78) 54 (þ)
100 (79) 60 (þ)
100 (67) 25 (S)-(þ)d
100 (88) 86 (S)-(þ)d
100 (99) 29 (S)-(þ)d
100 (95) 26
100 (88) 48
100 (86) 48
100 (91) 15
100 (95) 85
100 (90) 33
FIGURE 5. Cyclic carbo- and heterocyclic substrates examined in
this work.
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Hydroxylation of Acyclic Ketoesters: Unusual Match/
Mismatch Effect of Catalyst and Oxidant. Asymmetric
R-hydroxylation of acyclic 1,3-ketoesters is rare, presumably
due to the operation of competitive processes, e.g., Bayer-
Villiger rearrangement and decarboxylation.4,22 Prior to our
report, only the Ti complex 4 could achieve notable yields
and ee values (up to 84%).6 Recently, an alternate approach
to these compounds has also been achieved by the asymmetric
alkylation of O-protected R-hydroxyl-β-keto esters.23,24
Hydroxylation of a series of alkyl 2-benzylacetoacetates
21a-d and commercially available 22 and 24 (Figure 6)
proceeded slowly under Pd catalysis (Table 5, entries 1-3).
Significantly, competitive side reactions were not observed,
which allowed for increases in the amount of oxidant,
catalyst loading (20 mol %), and reaction temperature
(0 °C) to achieve acceptable yields (entries 1-3 vs 4-7).
Encouragingly, up to 88% yield was obtained with the tert-
butyl ester 21c (entry 6), which is the highest recorded for the
R-hydroxylation of acrylic substrates. On the other hand, R-
aryl-substituted ketoesters are insufficiently active (entry 9).
Conversely, oxaziridine 1b was ineffective for the hydro-
xylation of acyclic ketoesters; very little product formation
can be obtained using 10 mol % of catalyst I or II. However,
catalytic activity was restored by using dimeric μ-hydroxy-
palladium complexes III and IV (Scheme 4), prepared by the
reaction of catalyst I with 0.07 M sodium hydroxide,25 or
generated in situ by the addition of molecular sieves to II,26
respectively. Using the basic form of these catalysts, good con-
versions of acyclic ketoesters can be attained using 1b as the
oxidant (Table 6). In all cases, reactions were complete overnight
to afford products of low to moderate level of selectivity.
I (5)
7a
7c
7e
7f
II (10)
II (10)
II (10)
II (10)
100 (80)
100 (85)
4
3
100 (90) 26 (S)-(þ)d
100 (94) 16 (R)-(-)d
aReaction conditions: catalyst I or II (5 or 10 mol %), ketoester 7a-f
(0.123 mmol.), DMD (0.7 M in acetone, 1.2 equiv), CH2Cl2. bConver-
sion determined by 1H NMR spectroscopy, isolated product yield in
parentheses. cDetermined by chiral HPLC. dAbsolute configuration was
assigned by comparison to literature data.
Reaction Scope (Cyclic Substrates). Subsequently, several
cyclic carbo- and heterocyclic substrates 9-12 (Figure 5)
were subjected to the optimized reaction conditions
(Table 4). Compared to the indanone-derived substrates
7a-e, hydroxylation of the tetralone-derived substrate
9a-c proceeded with very low enantioselectivity (<20%),
even at 10 mol % catalytic loading (entries 1-3). On the
other hand, removal of the aromatic ring from the substrate
led to a better result, as 66% ee was obtained with the
cyclohexanone derivative 11 (entry 4). This applied equally
to the cyclopentanone derivatives, as the hydroxylation of
10a-c proceeded with much higher enantioselectivities than
their corresponding indanone esters 7b, 7d, and 7e (Table 4,
entries 5-7 vs Table 3, entries 12, 14, and 15). Particularly
pleasing is the result obtained for compound 17c (98% ee),
matching the best ee recorded for this substrate.6,7,10 Accord-
ingly, hydroxylation of N-heterocycles 12-14 also offered
good to excellent enantioselectivities, thus providing a route
to hydroxylated pyrrolidine rings (entries 8-13). The nature
of the protecting group used in heterocycle 12 proved to be
important; the Moc-protected product (18a) was unstable to
isolation, resulting in a low yield of the product. In compar-
ison, much better yields were obtained with Cbz- and Boc-
protected 18b and 18c. With the N-heterocycles 13a and 13b,
the reduced acidity of the R-proton led to a slower reaction.
This was overcome by raising the temperature to 0 °C to
afford the hydroxylated product in excellent yields. The best
selectivity was attained for the Boc-protected 19b (96% ee).
Last, but not least, a series of succinimide derivatives 14a-d
were also examined. Good ee’s of up to 87% can be
achieved; the planar nature of the cyclic imides may be
the cause of the slightly lower enantioselectivity. The products
were found to be highly crystalline as homochiral conglomer-
ates, e.g., the N-PMB derivative 20b can be rendered
optically pure after a single recrystallization from ethyl acetate-
hexane.
Perplexingly, DMD did not afford any product from these
substrates using either catalysts III or IV. So it appears that
catalytic activity toward achiral ketoesters is only afforded
by particular combinations of catalyst and oxidant. This was
examined further by monitoring reactions of 21a, using
catalyst I or II with oxaziridine 1b or DMD as oxidant under
comparable conditions, i.e., maintaining the same reactant
concentrations and catalyst loading (figure 7). At ambient
temperature, the evaporation of volatile DMD led to incomplete
conversion. Nevertheless, it is clear that the reaction performed
€
(22) For example, see: (a) Li, D.; Schroder, K.; Bitterlich, B.; Tse, M. K.;
Beller, M. Tetrahedron Lett. 2008, 49, 5976. (b) Watanabe, T.; Ishikawa, T.
Tetrahedron Lett. 1999, 40, 7795. (c) Christoffers, J. J. Org. Chem. 1999, 64,
7668. (d) Christoffers, J.; Werner, T.; Unger, S.; Frey, W. Eur. J. Org. Chem.
2003, 425.
(23) Hashimoto, T.; Sasaki, K.; Fukumoto, K.; Murase, Y.; Ooi, T.;
Maruoka, K. Synlett 2009, 661.
(24) Hashimoto, T.; Sasaki, K.; Fukumoto, K.; Murase, Y.; Abe, N.; Ooi,
T.; Maruoka, K. Chem. Asian J. 2010, 5, 562.
(25) Fujii, A.; Hagiwara, E.; Sodeoka, M. J. Am. Chem. Soc. 1999, 121,
5450.
(26) Attempts to isolated complex IV were unsuccessful.
J. Org. Chem. Vol. 75, No. 9, 2010 3089

Smith et al.
JOCArticle
TABLE 4. Asymmetric Hydroxylation of Carbo- and Heterocyclic 1,3-Ketoestersa
aReaction conditions: ketoester (0.123 mmol), catalyst (R)-I (5 mol %), DMD (0.7 M in acetone, 1.2 equiv), CH2Cl2. bIsolated yield. All reactions
c
were complete within the given time (1H NMR spectroscopy). Determined by chiral HPLC. dAbsolute configuration assigned by comparison to
literature data. eDetermined by chiral GC. fDetermined by X-ray crystallography (vide infra). gOptical rotation was too small to be measured.
with DMD/catalyst I is slightly faster than the reaction using
oxaziridine 1b/catalyst III (based on initial rates), while other
combinations of catalyst and oxidant were very slow.
Formation of the Pd-enolate Complex. Coordination of 1,3-dic-
arbonyl compounds to dicationic (diphosphine)palladium(II)
complexes is a well-established process. A number of such
palladium-enolate complexes have been characterized spec-
troscopically and also by X-ray crystallography.27,28 Chelation
of 1,3-dicarbonyl compounds to complex I is accompanied by
the formation of 1 equiv of triflic acid (Scheme 5, eq 1), while
(27) Hamashima, Y.; Hotta, D.; Umebayashi, N.; Tsuchiya, Y.; Suzuki,
T.; Sodeoka, M. Adv. Synth. Catal. 2005, 347, 1576.
(28) Nama, D.; Pregosin, P. S.; Albinati, A.; Rizzato, S. Organometallics
2007, 26, 2111.
3090 J. Org. Chem. Vol. 75, No. 9, 2010

Smith et al.
JOCArticle
TABLE 6. Hydroxylation of Acyclic Ketoesters by 1b Catalyzed by
Dimeric Complexes III and IV
FIGURE 6. Acyclic 1,3-ketoester substrates.
TABLE 5. Hydroxylation of Acyclic Ketoesters Catalyzed by I Using
DMD as Oxidanta
aReaction conditions: ketoester (0.123 mmol), catalyst (10 mol %), 1b
(1.2 equiv), CH2Cl2, rt. bIsolated yield after column chromatography.
c
Starting material was consumed (TLC or 1H NMR). Determined by
chiral HPLC. Absolute configuration assigned by comparison of optical
rotation with reported value.
aReaction conditions: ketoester (0.123 mmol), catalyst I, DMD
(0.7 M in acetone, 1.2 equiv), CH2Cl2, 0 °C. bIsolated yield after column
c
chromatography. Determined by chiral HPLC. dAbsolute configura-
tion assigned by comparison with reported data. e2 equiv of DMD.
SCHEME 4. Formation of Dimeric μ-Hydroxypalladium
Complexes III and IV
the chelation to III is promoted by the deprotonation of the
R-hydrogen by the basic Pd-OH moiety (Scheme 5, eq 2). The
unique Brønsted acidity or basicity associated with catalysts I
and III is often crucial for catalytic activity.29,30
FIGURE 7. Rates of hydroxylation of 21b with different combina-
tions of catalyst and oxidant (determined by 1H NMR spectrosco-
py, see the Supporting Information). Reaction conditions: 21b
(0.123 mmol), oxidant (1.2 equiv), 10 mol % of I or 5 mol % of
III, CH2Cl2 (total reaction volume = 3.09 mL), rt.
In our preliminary communication,11 we showed that the
addition of 10 equiv of cyclic ketoester 7a to complex III led
to the formation of the palladium-enolate complex V,
indicated by the appearance of an AB pattern in the 31P
NMR spectrum (δP þ29.7, 34.7, J = 27.5 Hz).31 Reactions
of acyclic ketoesters are generally slower. This is attributed
to difficulty in the formation of the activated complex V,
partly due to entropic effects (the acyclic substrate has
greater conformational freedom than the cyclic substrate)
and also unfavorable steric interactions between adjacent R1
and R2 substituents that result from the planar geometry
necessary for chelation (Scheme 5). For comparison, the
binding of the acyclic substrate 21b to catalysts I and III was
examined by 31P NMR spectroscopy (Figure 8). In contrast
(29) Sodeoka, M.; Hamashima, Y. Pure Appl. Chem. 2006, 72, 477.
(30) Sodeoka, M.; Hamashima, Y. Chem. Commun. 2009, 5787.
(31) Attempts to isolate the complex were unsuccessful.
J. Org. Chem. Vol. 75, No. 9, 2010 3091

Smith et al.
JOCArticle
SCHEME 5. Formation of Palladium-Enolate Complex V
from I and III
TABLE 7. Comparison of Catalyst and Oxidants in the Hydroxylation
of 7aa
entry
catalyst
oxidant
temp/°C
time/h
conv (ee)/%
1
2
3
4
I
III
I
1b
1b
DMD
DMD
rt
rt
-20
-20
2
2
0.5
2
100 (30)
100 (30)
100 (34)
41 (35)
III
aCatalytic conditions are as described previously in Tables 2 and 4.
SCHEME 6. Postulated Catalytic Cycle Using DMD and
Catalyst I
catalyst I (entries 3 and 4 in Table 7 and Figure 7), even
though the formation of the activated palladium-enolate
complex is less favored (figure 8); hence, step A is unlikely to
be turnover-limiting in this system. Following electrophilic
addition (step B), the epoxide intermediate VII presumably
rearranges to the putative alkoxide complex VIII (step C).
Product release/catalytic turnover will require protonolysis
of this intermediate (step D), which is expected to be more
facile in the presence of triflic acid. We propose this as the
rate-limiting step for reactions involving DMD. Conversely,
when oxaziridine 1b was used as the oxidant, reactions using
catalyst III are faster than those mediated by catalyst I. As
this corresponds to the relative binding affinities of the
ketoester to the Pd complexes, step A is likely to be turn-
over-limiting in this system.
This switch of rate-limiting step between the two systems
has significant implications for future development of this
work. We speculate that this could be due to (i) an interaction
between DMD and triflic acid, such as H-bonding, that
enhances the rate of step A, and/or (ii) product release that
is faster using 1b as the oxidant, emolliating the effect of
triflic acid on step D, which will imply some interaction
between the oxidant (or its imine byproduct) with VII or
VIII, making them unstable.
Origins of Stereochemistry. Catalyst (R)-I and related
complexes have been used previously in enantioselective R-
fluorination and R-amination of 1,3-ketoesters.16,18,32 In all
cases, electrophilic attack occurred preferentially from the
Re face of the coordinated enolate (Scheme 7). This was
rationalized by the way the enolate chelates to the dicationic
palladium center: To avoid steric interaction with an equa-
torial P-Ph substituent, the ester substituent distorts away
from a planar s-trans configuration to occupy the least
hindered quadrant of the coordination sphere, thereby
FIGURE 8. 31P NMR spectra generated from mixtures of com-
plexs I (left) and III (right) in CDCl3, with 0 (A), 10 (B), and 200
equiv (C) of 21b.
to the previous experiment, chelate formation between III
and 21b was observable only in the presence of 200 equiv of
the acyclic ketoester (δP þ29.2, þ33.4, J = 37 Hz; Mþ
=
947); i.e., the binding of acyclic ketoesters to catalyst III is at
least 20 times weaker than the binding of cyclic substrates. As
expected, the binding of 21b to complex I is further hampered
by an unfavorable equilibrium (formation of a strong acid)
and was barely detectable under the same conditions.
These binding studies highlighted important aspects of
these reactions. If the formation of the Pd-enolate complex
is the only mode of activation (and turnover-limiting), reac-
tions catalyzed by III should be faster than those catalyzed
by I. However, this applies only when the oxaziridine 1b was
employed as the oxidant, while the opposite is true when
DMD was used (Figure 7). The same trends were also
observed for the hydroxylation of cyclic ketoesters; while
there is no detectable difference between reactions catalyzed
by I or III when 1b was used as the oxidant (Table 7, entries 1
and 2), the reaction with DMD was slower when catalyst III
was employed (entries 3 and 4). This suggests that triflic acid
plays an important role in reactions involving DMD but not
1b. It is worth noting that the same ee value was obtained
using a specific oxidant, irrespective of the catalyst used; i.e.,
the presence of TfOH has no effect on the relative energies
(ΔΔGq) of the stereodefining step.
Catalytic Cycles and Possible Turnover-Limiting Steps.
Based on these observations, working models for these
reactions were constructed. For reactions employing DMD
(Scheme 6), we established that reactions are faster using
(32) Suzuki, T.; Goto, T.; Hamashima, Y.; Sodeoka, M. J. Org. Chem.
2007, 72, 246.
3092 J. Org. Chem. Vol. 75, No. 9, 2010

Smith et al.
JOCArticle
SCHEME 7. Previously Proposed Stereochemical Model for
Pd-Catalyzed R-Functionalization of 1,3-Ketoesters
FIGURE 9. Assigned stereochemistry of 15a and 29 by circular
dichroism (calculated and observed).
FIGURE 10. Three views representing key interactions present in
the favored transition state (TSSi): (a) top view depicting C-H
O
3 3 3
interactions with DMD; (b) side view depicting C-H O interac-
3 3 3
tions with triflate anion; and (c) side view depicting C-H
interactions.
π
3 3 3
hindering the approach of the electrophile from the Si face.16
The stereochemical model was also applied to related C-C
bond-forming reactions.30 In all cases, the major enantiomer
results from an electrophilic addition on the Re face.
In light of this, it came as a great surprise to find that this
does not apply to the present system. Using (R)-I as the
catalyst, hydroxylation products containing the S-config-
uration were obtained as the preferred enantiomer; i.e.
electrophilic addition occurs on the Si face, clearly defying
the stereoselectivity predicted by this model. To be certain,
the stereochemical assignment of R-hydroxyl ketoesters was
verified by comparing the optical rotations of our com-
pounds to those reported in the literature, which have been
based, invariably, on correlations to assignments of (R)-(þ)
and (S)-(-)-configurations to 15a and 29 (Figure 9) made by
Davis on the basis of circular dichroism spectra.4 These
assignments were verified independently in this work by
calculating the electronic circular dichroism (ECD) response
for 15a and 29. Using time-dependent DFT theory at the
B3LYP/aug-cc-pVTZ level with applied continuum solva-
tion field correction for methanol solvent (Web-enhanced
Table 1), the signs of the peaks match those reported,
confirming the original assignment of absolute configuration
of both these compounds.
Additionally, the absolute configuration of 25c was estab-
lished recently by a crystal structure of a closely related
derivative.24 Furthermore, we obtained a single crystal of an
optically pure sample of 20b, which was analyzed by X-ray
crystallography (Supporting Information). By use of the
Flack parameter [xþ = 0.00(12), x- = 1.04(12)],33 it was
possible to determine purely from the diffraction data that
the species present in the crystal has the S-configuration
shown.
With unequivocal validation of our stereochemical assign-
ment, an alternate explanation is needed for the observed
selectivity. A quantitative transition-state model was con-
structed for this purpose (previous models had been based on
nontransition state geometries obtained using molecular
mechanics methods27). The standard DFT procedure at the
B3LYP/cc-pVDZ level of theory (cc-pVDZ-pseudopotential
for Pd) was enhanced with a continuum solvation model
(CPCM, solvent = acetone), and the geometries of the
transition states for oxygen atom transfer from DMD were
fully located and optimized within this model. Such an
approach was necessary in order to be able to quantitatively
define model systems with a dipole moment >20 D, this
magnitude resulting from ionization of the TfO-Pd to form
an ion-pair complex. The energy of the resulting geometric
model was then corrected for thermal and entropic effects via
normal coordinate analysis to give free energies of activation
(∼11-13 kcal mol-1), which are commensurate with the
observed kinetics of these reactions. Animations of the
calculated transition normal modes for the hydroxylation
of substrate 10c (R = t-Bu) and 10d (R = Me) by DMD are
presented in the Web-enhanced Table 1, where a more
complete set of geometric features can also be explored
interactively. Several key features are also highlighted in
Figure 10.
The first striking difference from the previous model is the
preservation of the s-trans conformation of the ester moiety;
even a tert-butyl ester group can be accommodated in the
plane of the chelate ring throughout the process. The oxygen
transfer step occurs via a highly asynchronous spiro transi-
tion state, where the oxirane ring approaches C-2 practically
in a linear trajectory.34 The approach of oxirane on either
face of the enolate is strongly directed by two C-H
O
3 3 3
˚
hydrogen bonds (ca. 2.6 A), between the oxidant and the
BINAP ligand (Figure 10a). This causes significant distor-
tion in the arrangement of the P-Ph groups from the “classi-
cal” quadrant occupancy. Additionally, there are also clear
interactions between the aromatic C-H bonds with the
˚
ionized triflate anion (ca. 2.4 A), which resides in a pocket
opposite to that of the reacting DMD reagent (Figure 10b).
These interactions are present in both favored and
disfavored transition states. However, only in the favored
transition state (TSSi) are there discernible C-H/π interac-
tions between the tert-butyl substituent and the equatorial
(33) Flack, H. D. Acta Crystallogr. A 1983, 39, 876.
(34) Dufert, A.; Werz, D. B. J. Org. Chem. 2008, 73, 5514.
J. Org. Chem. Vol. 75, No. 9, 2010 3093

Smith et al.
JOCArticle
phenyl group (Figure 10c). These are distinctly absent in the
disfavored transition state, due to the intervention of one of
the methyl substituents of the oxirane ring positioned be-
tween the ester substituent and the equatorial Ph. Consid-
ered as the weakest form of hydrogen bonding, C-H/π
interactions are largely dispersive in origin (also referred to
as long-range correlation effects) and are very probably
much more common in organic and organometallic chem-
istry than would be apparent from many simple models.35
In the present system, this appears to be crucial for stereo-
selectivity.
DMD oxidant.38 A more systematic exploration of other
positions has not yet been possible because of the computa-
tional cost. For the same reason, the effect of adding a
further unit of triflic acid via hydrogen-bonding interactions
to the dimethyl dioxirane reagent has yet to be explored.
Either of these refinements might potentially affect the
predicted magnitude of the stereoselectivity. We did, how-
ever, observe that removing the triflate anion from the model
entirely (resulting in a cationic system overall) favors the
TSSi over TSRe by 3.6 kcal mol-1 in free energy (B3LYP
model, tert-butyl ester), which hints that even the nature of
the counterion may influence the outcome. We do, never-
theless, feel we have demonstrated that a fully optimized
model for such a complex reaction site is now entirely
possible for systems where an ion-pair is involved. Indeed,
it is worth reiterating that a fully quantum-mechanical model
is essential not merely to reproduce the bond-breaking at the
transition state, but the plethora of weaker interactions
between the various components.39
A more systematic analysis of the weak interactions in this
complex system can be obtained from the atoms-in-molecule
(AIM) electronic density topology of the transition state,
from which a large number of bond critical points (BCPs)
corresponding to weak interactions can be identified.
Although one must be wary about making a connection
between such a topological property of the electron density
and the clear existence of an attractive interaction between
two atomic centers,36 the presence of so many critical points
in even such a relatively simple catalytic reaction center
suggests that the overall specificity of the reaction is as likely
to be mediated by the overall sum of many such weak
interactions, as by a few particularly strong heuristic (i.e.,
memorably named) ones. No accepted methodology exists
for summing such a large number of diverse “weak” inter-
actions, each of which might need individual calibration for
the energy of interaction, and we do not attempt such here.
Nonetheless, we recognize that many, if not most, of these
interactions are indeed likely to be dispersive in nature,
originating from long-range correlation effects. The
B3LYP-DFT procedure approximates only the short-range
and not the long-range correlation effects. To approximately
estimate the correction for the latter, we also performed
single-point energy evaluations using a dispersion-corrected
DFT method (B97D)37 in which a simple empirical correc-
tion is added to mimic the long-range dispersion effects. For
the hydroxylation of the methyl ester 10d, the B3LYP þ
solvation DFT method predicts a small bias for TSSi over
the TSRe transition state for the methyl ester, but this is
inverted for the tert-butyl ester 10c, which is contrary to our
experimental observations. However, the B97D dispersion-
corrected energies enhance the TSSi preference by 0.9 kcal
mol-1 (methyl ester) and by 1.2 kcal mol-1 (tert-butyl ester),
and with the latter correction, TSSi is now favored over
TSRe for both esters. We also noted that fully reoptimizing
the transition state geometries at the B97D level (rather than
the single point energy evaluation procedure as noted above)
Conclusions
Highly enantioselective Pd-catalyzed R-hydroxylation of
1,3-ketoesters can be achieved using dicationic palladium
catalyst I and DMD as oxidant. By extensive screening of
reaction conditions, reactants, and catalysts, key parameters
important for stereocontrol have been revealed. With a
certain combination of catalyst and oxidant, the stereoselec-
tivity of the process can be controlled by varying the size of
the ester substituent. Unusual reactivity patterns were un-
covered, which was attributed to highly synergistic relation-
ship between the catalytic intermediates and the oxidant,
which has a dramatic effect on catalytic turnover.
During this work, we have uncovered a novel stereochem-
ical pathway for the R-hydroxylation of ketoesters, which
opposes that previously observed in other R-functionaliza-
tion reactions. Quantitative DFT models show that the
stereoselectivity of the process is directed by a combination
of C-H O and C-H/π interactions, with the latter con-
3 3 3
ferring facial discrimination to the hydroxylation process. As
far as we are aware, this is the first time such weak inter-
molecular forces have been implicated in an asymmetric
process.
Experimental Section
Catalytic Procedures, Method A (Catalyst I and DMD). A
Radley’s reaction tube was charged with β-ketoester (0.123
mmol), catalyst I (6.58 mg, 6.15 μmol), and CH2Cl2 (0.5 mL)
and cooled to -20 °C. A solution of DMD in acetone (0.09 M,
1.64 mL, 0.147 mmol) at -20 °C was then added dropwise. After
30 min, the mixture was concentrated and filtered through a
plug of silica. The crude product was either judged to be pure
enough by 1H NMR or purified by column chromatography.
Method B (Catalyst III and 1b). A Radley’s reaction tube was
charged with β-ketoester (0.123 mmol), catalyst III (5.51 mg,
3.08 μmol), and CH2Cl2 (0.5 mL). After 5 min, 1b (40.7 mg,
˚
led to unreasonable forming C-O bond lengths (>2.6 A
˚
compared to ∼2.2 A at the B3LYP level) and activation free
energies (∼1.7 kcal mol-1 vs 11-13 at the B3LYP level).
Although our model represents a considerable advance in
sophistication compared to previous ones, there are still
aspects that will require future investigation. Thus, the
triflate anion (which the AIM analysis shows is not entirely
passive) was optimized from an initial placement in an
obvious pocket beneath the direction of approach of the
(39) The DFT-model constructed involved up to 126 atoms, specified by
up to 1323 basis functions, and with both first- and second-energy derivatives
analytically computed within a continuum solvent field. These are close to
the current (2009) practical limits, although there is every reason to expect
that extending the models to around 150 atoms at this level will be feasible in
the near future, as will the ability for more extensive exploration of the ion-
pair and solvent spaces.
(35) Tsuzuki, S.; Fujii, A. Phys. Chem. Chem. Phys. 2008, 10, 2584.
(36) Bader, R. F. W. J. Phys. Chem. A 2009, 113, 10391.
(37) Grimme, S. J. Comput. Chem. 2006, 27, 1787.
(38) Quinonero, D.; Deya, P. M.; Carranza, M. P.; Rodriguez, A. M.;
Jalon, F. A.; Manzano, B. R. Dalton Trans. 2010, 39, 794.
3094 J. Org. Chem. Vol. 75, No. 9, 2010

Smith et al.
JOCArticle
0.148 mmol) was added and the solution stirred until judged to
be complete by consumption of the starting material (TLC). The
reaction mixture was then concentrated in vacuo and purified by
column chromatography.
Method C (Using Catalyst II). A Radley’s reaction tube was
charged with a solution of catalyst II (25.0 mg, 24.6 μmol) in
˚
CH2Cl2 (0.5 mL) and powdered molecular sieves (4 A, 125 mg).
After the solution was stirred for 30 min, a solution of
β-ketoester (0.246 mmol) in CH2Cl2 (0.5 mL) was added,
followed by 1b (81.2 mg, 0.295 mg), and the progress of
the reaction was monitored (TLC). When this was completed,
the reaction mixture was then filtered through a plug of
Celite, concentrated in vacuo, and purified by column chro-
matography.
NMR δH (CDCl3) 7.38-7.33 (2H, m), 6.88-6.83 (2H, m),
4.76-4.55 (2H, m), 3.80 (3H, m), 3.10 (1H, d, J 18.0), 2.85
(1H, d, J 18.0), 1.32 (9H, s); NMR δC (CDCl3) 174.1, 173.1,
168.55, 159.5, 130.3, 127.4, 114.0, 85.7, 76.0, 55.3, 42.25, 40.6,
27.5; MS m/z (EI) 335.1369 (Mþ, C17H21NO6 requires
335.1369), 278 (48), 121 (100), 57 (47). Chirapak OD-H column,
hexane/i-PrOH = 90/10, 1 mL/min, 254 nm, tR (minor) = 12.7
min, tR (major) = 13.8 min; [R]25 þ37.8 (c = 0.95, CHCl3,
D
>99% ee).
Crystal data for (S)-20b: C17H21NO6, M = 335.35, mono-
˚
˚
clinic, P21 (no. 4), a = 6.34741(7) A, b = 10.76431(9) A, c =
3
˚
˚
12.86992(13) A, β = 102.2709(10)°, V = 859.254(15) A , Z = 2,
Dc = 1.296 g cm-3, μ(Cu KR) = 0.824 mm-1, T = 173 K,
colorless tablets, Oxford Diffraction Xcalibur PX Ultra dif-
Method D (Ligand Screening). Under a nitrogen atmosphere,
a Radley’s reaction tube was purged with nitrogen and charged
fractometer; 3386 independent measured reflections (Rint =
0.0297), F2 refinement, R1(obs) = 0.0259, wR2(all) = 0.0688,
3221 independent observed absorption-corrected reflections
[|F0| > 4σ(|F0|), 2θmax = 145°], 223 parameters. The absolute
structure of (S)-20b was determined by a combination of
R-factor tests [R1þ = 0.0259, R1- = 0.0262] and by use of the
Flack parameter [xþ = 0.00(12), x- = 1.04(12)]. CCDC765842.
tert-Butyl 3-hydroxy-1-tert-butylsuccinimide-3-carboxylate
(20c, Table 4, entry 15): white crystalline solid; mp 55-56 °C;
IR νmax/cm-1 2980, 1744, 1707, 1370, 1348, 1265, 1172, 1148;
NMR δ H (CDCl3) 3.80 (1H, br s), 3.00 (1H, d, J 17.6), 2.78 (1H,
d, J 17.6), 1.61 (9H, s), 1.50 (9H, s); NMR δC (CDCl3) 175.7,
174.4, 168.8, 85.36, 75.69, 59.2, 40.8, 28.1, 27.7; MS m/z (NH3,
CI) 289.1765 (MNH4þ, C13H25N2O5 requires 289.1763), 233
(23), 217 (11); Chirapak OD-H column, hexane/i-PrOH = 98/2,
1 mL/min, 254 nm, tR (minor) = 11.9 min, tR (major) = 13.6
min; [R]25D þ18.1 (c = 1.10, CHCl3, 68% ee).
with Pd(OTf)2 2H2O40 (5.4 mg, 12.3 μmol), the appropriate
3
ligand (13.5 μmol), and CH2Cl2 (0.4 mL). After the solution was
stirred for 30 min, a solution of 7a (23.4 mg, 0.123 mmol) in
CH2Cl2 (0.1 mL) was added, followed after a further 5 min by a
solution of 1b (40.7 mg, 0.148 mmol) in CH2Cl2 (0.1 mL), and
the solution stirred for 18 h. The reaction mixture was then
concentrated in vacuo, filtered through a plug of silica, and
analyzed by NMR and chiral HPLC.
Benzyl 1-tetralone-2-hydroxy-2-carboxylate (15c, Table 4,
entry 3): white crystalline solid; mp 87-89 °C; IR νmax /cm-1
2936, 1739, 1686, 1602, 1455, 1227, 1189, 907, 725; NMR δH
(CDCl3) 8.07 (1H, dd, J 7.8, 1.2), 7.59-7.52 (1H, td, J 7.6, 1.4),
7.37 (1H, t, J 7.6), 7.33-7.23 (4H, m), 7.18 (2H, m), 5.25 (1H, d,
J 12.4), 5.18 (1H, d, J 12.4), 4.35 (1H, s), 3.15-3.01 (2H, m), 2.75
(1H, dt, J 13.6, 5.0), 2.27 (1H, ddd, J 13.6, 8.8, 6.6); NMR δC
(CDCl3) δ 194.5, 170.4, 143.9, 134.9, 134.3, 130.3, 128.9, 128.5,
128.3, 128.1, 127.7, 127.0, 77.8, 67.5, 32.7, 25.5; MS m/z (EI)
296.1045 (C18H16O4 requires 296.1049), 205 (22), 180 (10), 161
(30), 107 (23), 91 (100). Chirapak AD-H column, hexane/i-
PrOH = 90/10, 1 mL/min, 254 nm, tR (major) = 26.7 min, tR
(minor) = 40.2 min. The optical purity was too low (1%) for
accurate determination of [R]D.
Benzyl 2-hydroxycyclohexanone-2-carboxylate (16, Table 4,
entry 4): colorless oil; IR νmax /cm-1 2947, 1744, 1717, 1455,
1216, 1056, 737, 696; NMR δH (CDCl3) 7.43-7.31 (5H, m),
5.28-5.17 (2H, m), 4.39 (1H, s), 2.72- 2.62 (2H, m), 2.56-2.43
(1H, m), 2.05-1.95 (1H, m), 1.90-1.64 (4H, m); NMR δC
(CDCl3) 207.1, 169.8, 135.0, 128.6, 128.5, 128.1, 80.7, 67.5,
38.8, 37.6, 27.0, 21.9; MS m/z (EI) 248.1044 (C14H16O4 requires
248.1049), 232 (6), 220 (6), 157 (15), 129 (18), 111 (15), 91 (100).
Chiracel OJ-H column, hexane/i-PrOH = 90/10, 1 mL/min, 220
nm, tR (minor) = 27.7 min, tR (major) = 30.7 min; [R]25D -20.6
(c = 1.55, CHCl3, 66% ee).
1,3-Ditert-butoxycarbonyl-3-hydroxysuccinimide (20d, Ta-
ble 4, entry 16): colorless oil; IR νmax/cm-1 2981, 1720, 1370,
1141, 909, 729; NMR δH (CDCl3) 4.25 (1H, m, br, OH), 3.16 (1H,
d, J 18.0), 2.93 (1H, d, J 18.0), 1.58 (9H, s), 1.50 (9H, s); NMR δC
(CDCl3) 170.6, 168.9, 167.7, 145.6, 86.8, 86.3, 76.3, 40.8, 27.7; MS
m/z (NH3, CI) 333.1663 (MNH4 , C14H25N2O7 requires
þ
333.1662), 317 (28), 233 (35); Chirapak AD-H column, hexane/
i-PrOH = 90/10, 1 mL/min, 254 nm, tR (major) = 9.4 min, tR
(minor) = 12.1 min; [R]25D þ0.0 (c = 1.11, CHCl3, 49% ee).
Benzyl 3-phenyl-2-acetylpropionate (25d, Table 5, entry 7):
colorless oil; IR νmax /cm-1 3034, 1717, 1496, 1455, 1268, 1199,
1119, 696; NMR δH (CDCl3) 7.44-7.31 (5H, m), 7.27-7.19
(3H, m), 7.19-7.12 (2H, m), 5.28-5.15 (2H, m), 4.06
(1H, s), 3.44 (1H, d, J 14.2), 3.20 (1H, d, J 14.2), 2.24 (3H, s);
NMR δC (CDCl3) δ 203.7, 170.3, 134.5, 134.4, 130.1, 128.7,
128.7, 128.6, 128.2, 127.1, 84.3, 68.3, 40.7, 25.1; MS m/z
(ammonia, CI) 316.1554 (MNH4 , C18H22NO4 requires
þ
316.1549), 300 (4), 272 (1), 182 (1); Chiracel OJ-H column,
hexane/i-PrOH = 90/10, 1 mL/min, 254 nm, tR (minor) =
25.7 min, tR (major) = 30.5 min; [R]25D þ2.8 (c = 0.70, CHCl3,
32% ee).
N-Benzyl 3-hydroxy-3-(tert-butoxycarbonyl)succinimide (20a,
Table 4, entry 13): white crystalline solid; mp 138-143 °C; IR
ν
max/cm-1 1742, 1715, 1697, 1402, 1185, 1151, 133, 967; NMR
δ
H (CDCl3) 7.44-7.27 (5H, m), 4.77 (1H, d, J 14.0), 4.71 (1H, d,
Ethyl 2-hydroxy-2-phenylacetoacetate (27, Table 6, entry 10):
colorless oil; IR νmax /cm-1 1719, 1359, 124, 1173, 1072, 698;
NMR δH (CDCl3) 7.55 (2H, m), 7.41-7.35 (3H, m), 4.74 (1H, s),
4.37-4.24 (2H, m), 2.22 (3H, s), 1.31 (3H, t, J 7.2); NMR δC
(CDCl3) 203.8, 170.5, 133.3, 128.9, 128.6, 126.5, 84.7, 63.2, 24.3,
21.9; MS m/z (EI) 222.0893 (Mþ, C12H14O4 requires 222.0892),
180 (48), 151 (16), 105 (100), 77 (42); Chiracel OJ-H column,
hexane/i-PrOH = 90/10, 1 mL/min, 254 nm, tR (minor) =
21.2 min, tR (major) = 29.9 min; [R]25D -19.0 (c = 0.21, CHCl3,
27% ee).
J 14.0), 4.09 (1H, s), 3.12 (1H, d, J 18.0), 2.87 (1H, d, J 18.0), 1.34
(9H, s); NMR δC (CDCl3) 174.1, 173.1, 168.5, 135.1, 128.8,
128.7, 128.15, 85.7, 76.0, 42.8, 40.65, 27.5; MS m/z (EI) 305.1256
(Mþ C16H19NO5 requires 305.1263), 249 (28), 205 (34), 91 (52),
57 (100). Chirapak AD-H column, hexane/i-PrOH = 90/10, 1
mL/min, 254 nm, tR (major) = 16.0 min, tR (minor) = 17.7 min;
[R]25D þ18.5 (c = 1.29, CHCl3, 83% ee).
tert-Butyl 3-Hydroxy-1-(4-methoxybenzyl)succinimide-3-car-
boxylate (20b, Table 4, Entry 14). Compound 20b (820 mg,
2.45 mmol) was recrystallized from hot hexane (16 mL) and
EtOAc (18.9 mL) to give the optically pure sample (603 mg, 69%
yield, >99% ee) as a white crystalline solid: mp 136-138 °C; IR
Acknowledgment. We thank EPSRC and Pfizer for stu-
dentship support (A.M.R.S.) and Johnson Matthey plc for
the loan of Pd salts. Gifts of chiral ligands were received
from Solvias, Johnson Matthey, Dow Pharma, Chirotech,
ν
max/cm-1 1739, 1703, 1614, 1588, 1514, 1402, 1180, 1152, 1129;
(40) Murata, S.; Ido, Y. Bull. Chem. Soc. Jpn. 1994, 67, 1746.
J. Org. Chem. Vol. 75, No. 9, 2010 3095

Smith et al.
JOCArticle
Prof. A. Togni (ETH, Zurich), and Prof. A. Pfaltz
(University of Basel). We are grateful to Prof. K. Maruoka
(Kyoto University) for providing useful discussions and ref 24.
compounds), X-ray crystallographic data (including ORTEP
1
plot and CIF), and copies of H and 13C NMR spectra. This
material is available free of charge via the Internet at http://
pubs.acs.org. Calculated coordinates and animated transition-
state normal modes are provided via web-enhanced tables in the
HTML version of the article.
Supporting Information Available: Experimental proce-
dure, and characterization data (precursors and known
3096 J. Org. Chem. Vol. 75, No. 9, 2010