Chiral bis(oxazoline)copper(II) complexes as lewis acid catalysts for the enantioselective Diels-Alder reaction

J. Am. Chem. Soc. 1999, 121, 7559-7573

7559

Chiral Bis(oxazoline)copper(II) Complexes as Lewis Acid Catalysts

for the Enantioselective Diels-Alder Reaction

David A. Evans,* Scott J. Miller,^1aThomas Lectka,^1band Peter von Matt^1c

Contribution from the Department of Chemistry & Chemical Biology, HarVard UniVersity,

Cambridge, Massachusetts 02138

ReceiVed April 14, 1999

Abstract: Bis(oxazoline)copper(II) complexes are highly enantioselective catalysts in Diels-Alder reactions

involving bidentate dienophiles. Cationic [Cu((S,S)-t-Bu-box)]X₂complexes with different counterions have

been used as catalysts, revealing a profound influence of the counterion on the rate and stereoselectivity of the

catalyst. A square-planar catalyst-substrate complex is proposed to account for the high diastereo- and

enantioselectivities observed. Three bis(oxazoline)-Cu(II) X-ray structures have been obtained that support

this model. Double-stereodifferentiating experiments, performed employing chiral dienophiles, afforded results

that are fully consistent with the proposed square-planar transition-state assemblage. In addition to imide-

based substrates, R,â-unsaturated thiazolidine-2-thiones have been introduced as a new class of dienophiles

with enhanced reactivity. Kinetics experiments were performed to quantify the role that product inhibition

plays in the course of the reaction. Rate and equilibrium binding constants of various catalyst inhibitors were

also derived from the kinetic analysis. A comparative study was undertaken to elucidate the differences between

the bis(oxazoline)-Cu(II) catalyst and the bis(oxazoline) catalysts derived from Fe(III), Mg(II), and Zn(II).

Catalyst performance was found to be a function of a subtle relationship between bis(oxazoline) structure and

transition metal.

Introduction

(II) complexes as chiral Lewis acid catalysts in the Diels-Alder

reaction⁴with imide dienophiles (eq 3). Our preliminary

communications of this reaction indicate that (a) [Cu(box)]-

(OTf)₂and related complexes are excellent catalysts for this

reaction⁵and (b) the importance of catalyst counterions in

modulating Lewis acidity is significant.⁶For example, cationic

[Cu(box)](SbF₆)₂complexes exhibit both 20-fold greater reac-

tivity and superior enantioselection than their triflate counter-

parts.

Prior studies from this laboratory have demonstrated the utility

of chiral copper complexes in asymmetric catalysis. For

example, C₂-symmetric bis(oxazoline) (box)-derived Cu(I) and

Cu(II) complexes have proven to be excellent catalysts for

enantioselective group-transfer processes such as cyclopropa-

nation (eq 1)²and aziridination (eq 2).³

As an outgrowth of these investigations, we have undertaken

an in-depth evaluation of the potential of these and related Cu-

In this study, a detailed investigation of [Cu((S,S)-t-Bu-box)]-

(X)₂complexes 1 as chiral Lewis acid catalysts in Diels-Alder

(1) (a) National Science Foundation Predoctoral Fellow. (b) National

Institutes of Health Postdoctoral Fellow. (c) Swiss National Science

Foundation and “Ciba-Geigy-Jubila¨umsstiftung” Postdoctoral Fellow.

(2) (a) Evans, D. A.; Woerpel, K. A.; Hinman, M. M.; Faul, M. M. J.

Am. Chem. Soc. 1991, 113, 726-728. (b) Lowenthal, R. E.; Abiko, A.;

Masamune, S. Tetrahedron Lett. 1990, 31, 6005-6008. (c) Mu¨ller, D.;

Umbricht, B. W.; Pfaltz, A. HelV. Chim. Acta 1991, 74, 232-240.

(3) (a) Evans, D. A.; Faul, M. M.; Bilodeau, M. T.; Anderson, B. A.;

Barnes, D. M. J. Am. Chem. Soc. 1993, 115, 726-728. (b) Evans, D. A.;

Faul, M. M.; Bilodeau, M. T. J. Am. Chem. Soc. 1994, 116, 2742-2753.

(4) For recent reviews, see: (a) Oppolzer, W. In ComprehensiVe Organic

Synthesis; Trost, B. M., Ed.; Pergamon Press: New York, 1991; Vol. 5.

(b) Kagan, H. B.; Riant, O. Chem. ReV. 1992, 92, 1007-1019. (c) Oh, T.;

Reilly, M. Org. Prep. Proc. Int. 1994, 26, 129-158. (d) Dias, L. C. J.

Braz. Chem. Soc. 1997, 8, 289-332.

(5) (a) Evans, D. A.; Miller, S. J.; Lectka, T. J. Am. Chem. Soc. 1993,

115, 6460-6461. (b) Evans, D. A.; Lectka, T.; Miller, S. J. Tetrahedron

Lett. 1993, 34, 7027-7030.

(6) Evans, D. A.; Murry, J. A.; von Matt, P.; Norcross, R. D.; Miller, S.

J. Angew. Chem., Int. Ed. Engl. 1995, 34, 798-800.

Published on Web 08/10/1999

7560 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

reactions of chelating dienophiles is presented. These complexes

are characterized by X-ray crystallography, and a kinetics study

has been employed to evaluate the extent of product inhibition.

In addition, double-stereodifferentiating reactions have been

carried out to assess the torsional stability of the catalyst-

dienophile complex (cf. A, eq 3).

Scheme 1

Auxiliary-Based Reactions. Chiral R,â-unsaturated N-acyl-

oxazolidinones such as (R)-2, introduced by us in 1984, have

proven to be effective dienophiles in Lewis acid-promoted

Diels-Alder reactions (Scheme 1).^7,8These dienophiles offer

the advantages of high reactivity (>100-fold that of an

equivalent ester dienophile) and a well-organized dienophile-

Lewis acid complex.^7aThe stereochemical outcome of diene

cycloadditions with this dienophile is consistent with (a) the

intervention of a bidentate Lewis acid-dienophile complex,

s-cis-/s-trans-3 and (b) subsequent cycloaddition out of the more

stable s-cis conformation from the more accessible dienophile

diastereoface to afford the principal cycloadduct (S,R)-4. Cas-

tellino has investigated the solution structures of the Me₂AlCl

and SnCl₄complexes of these imide-derived dienophiles by ¹H

NMR spectroscopy and has concluded that two-point binding

is occurring with both of the indicated Lewis acids.⁹Further-

more, an analysis of nonbonding interactions in the dienophile-

Lewis acid complex suggests that the s-cis conformer of 3 would

be expected to be more stable.¹⁰In addition to the preceding

evidence, the recently reported X-ray structures of Ti(IV) imide

complexes¹¹support our original design premise that two-point

chelation between strongly Lewis acidic metal centers and this

family of dienophiles is possible and that the dienophile-Lewis

acid complex adopts the s-cis conformation in the solid state.

It is significant to the subsequent discussion that, if two-point

catalyst-dienophile binding is occurring in these reactions, then

the s-cis dienophile conformation must be inVoked to account

for the stereochemical outcome of the cycloaddition. In fact,

all published Diels-Alder reactions with this family of chiral

dienophiles conform to the stereochemical predictions that

follow from a diastereofacial analysis of the metal-complexed

s-cis-3 conformer.^7,8

reactions of cyclopentadiene with unsubstituted dienophile 5

(R ) H).^13,14Other important contributions from Corey’s

laboratory described the use of aluminum-stien complex D for

enantioselective Diels-Alder reactions of imides.¹⁵Narasaka

has reported the effective chiral Ti(IV) catalyst E (R ) Ph)

that exhibits moderate to excellent selectivities with substituted

as well as unsubstituted dienophiles 5 (R ) H, alkyl).^16,17More

recently, Kobayashi found that the chiral BINOL-Yb(III) catalyst

F (X ) OH) effectively catalyzes the reaction of 5 (R ) H,

alkyl) with cyclopentadiene at 0 °C with excellent levels of

stereocontrol.¹⁸An extension of that study using a bis-

(acylamino)binaphthalene ligand and Yb(OTf)₃(F, X ) NH-

COR) has been recently disclosed by Nakagawa.¹⁹Collins has

(13) Fe(III): (a) Corey, E. J.; Imai, N.; Zhang, H.-Y. J. Am. Chem. Soc.

1991, 113, 728-729. Mg(II): (b) Corey, E. J.; Ishihara, K. Tetrahedron

Lett. 1992, 33, 6807-6810.

(14) Other bis(oxazoline)-metal complexes have been evaluated as

catalysts. Mg (II): (a) Desimoni, G.; Faita, G.; Righetti, P. P. Tetrahedron

Lett. 1996, 37, 3027-30. (b) Carbone, P.; Desimoni, G.; Faita, G.; Filippone,

S.; Righetti, P. Tetrahedron 1998, 54, 6099-6110 and references therein.

(c) Takacs, J. M.; Lawson, E. C.; Reno, M. J.; Youngman, M. A.; Quincy,

D. A. Tetrahedron: Asymmetry 1997, 8, 3073-3078. (d) Takacs, J. M.;

Quincy, D. A.; Shay, W.; Jones, B. E.; Ross, C. R. Tetrahedron: Asymmetry

1997, 8, 3079-3087. (e) Honda, Y.; Date, T.; Hiramatsu, H.; Yamauchi,

M. Chem. Commun. 1997, 1411-1412. Zn(II): (f) Evans, D. A.; Kozlowski,

M. C.; Tedrow, J. S. Tetrahedron Lett. 1996, 37, 7481-7484. For other

Mg(II) complexes employed in Diels-Alder reactions, see: (g) Fujisawa,

T.; Ichiyanagi, T.; Shimizu, M. Tetrahedron Lett. 1995, 36, 5031-5034.

(h) Ichiyanagi, T.; Shimizu, M.; Fujisawa, T. J. Org. Chem. 1997, 62, 7937-

7941. (i) Ordon˜ez, M.; Guerrero de la Rosa, V.; Labastida, V.; Llera, J. M.

Tetrahedron: Asymmetry 1996, 7, 2675-2686.

It is on the basis of these investigations that the use of

unsaturated oxazolidinone-derived dienophiles has been subse-

quently explored by ourselves and others in Diels-Alder

reactions with chiral Lewis acid catalysts (vide infra).

Chiral Lewis Acid-Catalyzed Reactions. Complexes derived

from a range of metals [Fe(III), Mg(II), Ti(IV), Yb(III), Zr-

(IV), Al(III), Ni(II), Cu(II)] and chiral ligands have now been

reported for the Diels-Alder reactions of imide 5 (Figure 1).¹²

Corey and co-workers have reported catalysts derived from bis-

(oxazoline) complexes of Fe(III) and Mg(II) (B and C) in

(15) (a) Corey, E. J.; Sarshar, S.; Bordner, J. J. Am. Chem. Soc. 1992,

114, 7938-7939. (b) Corey, E. J.; Imwinkelried, R.; Pikul, S.; Xiang, Y.

B. J. Am. Chem. Soc. 1989, 111, 5493-5495. (c) Corey, E. J.; Imai, N.;

Pikul, S. Tetrahedron Lett. 1991, 32, 7517-7520. For applications of this

complex to cycloadditions of maleimides, see: (d) Corey, E. J.; Sarshar,

S.; Lee, D. H. J. Am. Chem. Soc. 1994, 116, 12089-12090. (e) Corey, E.

J.; Letavic, M. A. J. Am. Chem. Soc. 1995, 117, 9616-9617.

(16) (a) Narasaka, K.; Tanaka, H.; Kanai, F. Bull. Chem. Soc. Jpn. 1991,

64, 387-391. (b) Narasaka, K.; Iwasawa, N.; Inoue, M.; Yamada, T.;

Nakashima, M.; Sugimori, J. J. Am. Chem. Soc. 1989, 111, 5340-5345

and references therein. (c) Narasaka, K.; Yamamoto, I. Tetrahedron 1992,

48, 5743-5754. (d) Yamamoto, I.; Narasaka, K. Bull. Chem. Soc. Jpn. 1994,

67, 3327-3333. (e) Yamamoto, I.; Narasaka, K. Chem. Lett. 1995, 1129-

1130. (f) Iwasawa, N.; Sugimori, J.; Kawase, Y.; Narasaka, K. Chem. Lett.

1989, 1947-1950. (g) Narasaka, K.; Saitou, M.; Iwasawa, N. Tetrahe-

dron: Asymmetry 1991, 2, 1305-1318. (h) Corey, E. J.; Matsumura, Y.

Tetrahedron Lett. 1991, 32, 6289-6292.

(17) Seebach and others have provided additional contributions to this

family of chiral titanium complexes: (a) Seebach, D.; Dahinden, R.; Marti,

R. E.; Beck, A. K.; Plattner, D. A.; Kuehnle, F. N. M. J. Org. Chem. 1995,

60, 1788-1799. (b) Haase, C.; Sarko, C. R.; DiMare, M. J. Org. Chem.

1995, 60, 1777-1787. (c) Chapuis, C.; Bauer, T.; Jezewski, A.; Jurczak, J.

Pol. J. Chem. 1994, 68, 2323-31. (d) Gothelf, K. V.; Jørgensen, K. A. J.

Org. Chem. 1995, 60, 6847-6851. (e) Reference 11a.

(7) (a) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc. 1988,

110, 1238-1256. (b) Evans, D. A.; Chapman, K. T.; Hung, D. T.;

Kawaguchi, A. T. Angew. Chem., Int. Ed. Eng. 1987, 26, 1184-1187. (c)

Evans, D. A.; Chapman, K. T.; Bisaha, J. Tetrahedron Lett. 1984, 25, 4071-

4074. (d) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc.

1984, 106, 4261-4263.

(8) For applications of these dienophiles in natural products synthesis,

see: (a) Evans, D. A.; Black, W. C. J. Am. Chem. Soc. 1993, 115, 4497-

4513. (b) Morimoto, Y.; Iwahashi, M.; Nishida, K.; Hayashi, Y.; Shirahama,

H. Angew. Chem., Int. Ed. Engl. 1996, 35, 904-906.

(9) (a) Castellino, S.; Dwight, W. J. J. Am. Chem. Soc. 1993, 115, 2986-

2987. (b) Castellino, S. J. Org. Chem. 1990, 55, 5197-5200.

(10) N,N-Dialkyl acrylamides have a strong propensity for the s-cis

conformation: Montaudo, G.; Librando, V.; Caccamese, S.; Maravigna, P.

J. Am. Chem. Soc. 1973, 95, 6365-6370.

(11) (a) Gothelf, K. V.; Hazell, R. G.; Jørgensen, K. A. J. Am. Chem.

Soc. 1995, 117, 4435-4436. (b) Cozzi, P. G.; Solari, E.; Floriani, C.; Chiesi-

Villa, A.; Rizzoli, C. Chem. Ber. 1996, 129, 1361-1368.

(12) For the catalysts illustrated in Figure 1, the absolute configurations

of the ligand have been matched to the absolute configuration of the Diels-

Alder adduct (eq 5).

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7561

Figure 1. Complementary Diels-Alder catalyst systems.

demonstrated that [Zr(EBTHI)](OTf)₂(G) is also effective in

this class of reactions; NMR studies indicate that the catalyst-

bound imide adopts an s-cis conformation.²⁰The successful use

of hydrated nickel complex H as an enantioselective catalyst

for cycloadditions of substituted and unsubstituted dienophiles

5 has been recently reported by Kanemasa.²¹With catalysts B,

C, and E-H, imide dienophile 5 presumably participates in

two-point binding, although with notable exceptions, the actual

structures of these complexes have been not been determined.

One metal-ligand complex that has been characterized by X-ray

crystallography is aluminum complex D, which has been

proposed to catalyze the illustrated reaction (eq 5) by one-point

substrate binding through the s-trans conformation. Another

solid-state characterization was provided by the Kanemasa group

for the Ni(II) complex H. In addition, Jørgensen has reported

the X-ray structure of the dienophile Ti(IV) complex formed

between D (R ) Me) and 5 (R ) Ph).^11a

reactions of 5 (R ) H),²³while Ghosh has shown the utility of

the hydrated unsubstituted variant (I, R ) H, n ) 4) in reactions

with several dienophiles (5, R ) H, Me, CO₂Et).²⁴Helmchen

has also found that copper complexes derived from mixed

phosphino-oxazoline ligands (J) are efficient Diels-Alder

catalysts.²⁵Finally, a new quinoline-derived copper complex

(K) has provided good levels of enantioselectivity in one imide-

based Diels-Alder reaction.²⁶

Since our initial communications describing the utility of

chiral Cu(II) complexes as enantioselective Diels-Alder cata-

lysts, a number of chiral ligands conjugated with Cu(II) salts

have been evaluated as catalysts for the same reaction (Figure

2).²²Davies has employed a spiro bis(oxazoline) template (I,

R ) -CH₂CH₂-, n ) 0) derived from aminoindanol for

Figure 2. Cu(II)-based Diels-Alder catalyst systems.

(18) (a) Kobayashi, S.; Hachiya, I.; Ishitani, H.; Araki, M. Tetrahedron

Lett. 1993, 34, 4535-4538. (b) Kobayashi, S.; Ishitani, H. J. Am. Chem.

Soc. 1994, 116, 4083-4084. (c) Kobayashi, S.; Ishitani, H.; Hachiya, I.;

Araki, M. Tetrahedron 1994, 50, 11623-11636. (d) Kobayashi, S. Synlett

1994, 689-701. (e) Kobayashi, S.; Araki, M.; Hachiya, I. J. Org. Chem.

1994, 59, 3758-9375. A similar Sc catalyst has been reported: (f)

Kobayashi, S.; Ishitani, H.; Araki, M.; Hachiya, I. Tetrahedron Lett. 1994,

35, 6325-6328.

(19) Nishida, A.; Yamanaka, M.; Nakagawa, M. Tetrahedron Lett. 1999,

40, 1555-1558.

(20) (a) Jaquith, J. B.; Guan, J.; Wang, S.; Collins, S. Organometallics

1995, 14, 1079-1081. (b) Jaquith, J. B.; Levy, C. J.; Bondar, G. V.; Wang,

S. T.; Collins, S. Organometallics 1998, 17, 914-925.

(21) (a) Kanemasa, S.; Oderaotoshi, Y.; Yamamoto, H.; Tanaka, J.; Wada,

E.; Curran, D. P. J. Org. Chem. 1997, 62, 6454-6455. (b) Kanemasa, S.;

Oderaotoshi, Y.; Sakaguchi, S.; Yamamoto, H.; Tanaka, J.; Wada, E.;

Curran, D. P. J. Am. Chem. Soc. 1998, 120, 3074-3088. (c) Kanemasa, S.;

Oderaotoshi, Y.; Tanaka, J.; Wada, E. Tetrahedron Lett. 1998, 39, 7521-

7524.

(22) (a) Johannsen, M.; Jørgensen, K. A. J. Chem. Soc., Perkin Trans.

2 1997, 1183-1185. (b) Brimble, M. A.; McEwan, J. F. Tetrahedron:

Asymmetry 1997, 8, 4069-4078. (c) Aggarwal, V. K.; Anderson, E. S.;

Jones, D. E.; Obierey, K. B.; Giles, R. Chem. Commun. 1998, 1985-1986.

With each of the illustrated catalysts, the interpretation of

the sense of asymmetric induction has been complicated by a

number of reaction variables: catalyst architecture, one- versus

two-point catalyst-dienophile binding, s-cis versus s-trans

(23) (a) Davies, I. W.; Gerena, L.; Castonguay, L.; Senanayake, C. H.;

Larsen, R. D.; Verhoeven, T. R.; Reider, P. J. J. Chem. Soc., Chem.

Commun. 1996, 1753-1754. (b) Davies, I. W.; Senanayake, C. H.; Larsen,

R. D.; Verhoeven, T. R.; Reider, P. J. Tetrahedron Lett. 1996, 37, 813-

814. (c) Davies, I. W.; Senanayake, C. H.; Larsen, R. D.; Verhoeven, T.

R.; Reider, P. J. Tetrahedron Lett. 1996, 37, 1725-1726. (d) Davies, I.

W.; Deeth, R. J.; Larsen, R. D.; Reider, P. J. Tetrahedron Lett. 1999, 40,

1233-1236.

(24) (a) Ghosh, A. K.; Mathivanan, P.; Cappiello, J. Tetrahedron Lett.

1996, 37, 3815-3818. (b) Ghosh, A. K.; Mathivanan, P.; Cappiello, J.;

Krishnan, K. Tetrahedron: Asymmetry 1996, 7, 2165-2168. (c) Ghosh,

A. K.; Cho, H.; Cappiello, J. Tetrahedron: Asymmetry 1998, 9, 3687-

3691.

(25) Sagasser, I.; Helmchen, G. Tetrahedron Lett. 1998, 39, 261-264.

(26) Brunel, J. M.; Del Campo, B.; Buono, G. Tetrahedron Lett. 1998,

39, 9663-9666.

7562 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Table 1. Reaction of Imide 5a with Cyclopentadiene Catalyzed by

dienophile conformation, and metal ion coordination architecture

have remained largely undefined. In the following discussion,

the details of our development of a practical Cu(II) Lewis acid

catalyst for imide-based Diels-Alder reactions are presented.

[Cu(box)](OTf)₂Complexes (Eq 6)^a

ligand (R)

time, h (T, °C)

endo/exo^bendo ee (S)-7^byield (%)^c

6a (Ph)

1 (-78), 3 (-50)

95:5

94:6

96:4

98:2

30

44

58

92

93

86

6b (R-Np) 1 (-78), 3 (-50)

6c (i-Pr)

6d (t-Bu) 10 (-78)

Results and Discussion

1 (-78), 3 (-50)

>98

Catalyst Development. Our initial evaluation of Cu(II)-bis-

(oxazoline) complexes as chiral Lewis acid catalysts was based

on the reaction of acrylate imide 5a with cyclopentadiene in

the presence of complexes derived from Cu(OTf)₂and bis-

(oxazoline)²⁷ligands 6 possessing the (S,S) configuration (eq

6). The triflate counterion was chosen with the expectation that

the bidentate dienophile would successively displace these

poorly coordinating ligands to give a cationic 1:1:1-complex

between ligand, Cu(2+), and dienophile, as illustrated in

complex A (eq 7). Based on the well-established coordination

chemistry of Cu(II), we anticipated a square-planar geometry

for this ternary complex if both triflate counterions were

dissociated.²⁸On the other hand, the intervention of five-

coordinate ternary monocationic complexes was not discounted.

^aReactions were run in CH₂Cl₂with 11 mol % ligand and 10 mol

1

% Cu(OTf)₂. ^bEndo/exo ratios were determined by H NMR and/or

chiral GLC or HPLC. Enantiomer ratios were determined by chiral

GLC or chiral HPLC. ^cValues refer to isolated yields of cycloadducts.

5a with cyclopentadiene (CH₂Cl₂, -78 °C, Table 1). When the

(S)-phenylglycine-derived bis(oxazoline) 6a-Cu(OTf)₂complex

was employed, a highly diastereoselective reaction was ob-

served, affording cycloadducts with >20:1 endo/exo selectivity.

Although the enantiomeric excess of the major endo product

was low (30% ee), it was evident that the 6a-Cu(OTf)₂complex

was functioning as a catalyst for the reaction. An evaluation of

the size of the ligand substituent, R, as a function of reaction

enantioselection revealed that the tert-butyl-substituted ligand

6d was optimal. In reactions catalyzed by 6d-Cu(OTf)₂

complex 1a, the minor enantiomer of the endo product was not

detected, while high endo/exo diastereoselectivity was retained

(98:2). The absolute configuration of Diels-Alder adduct (S)-7

was determined by conversion to the derived benzyl ester, whose

rotation was compared to the literature value.^7a

Table 2. Reactions of Imide 5a with Cyclopentadiene and

Catalysts Derived from Ligand 6d and M(OTf)_n

endo/ endo yield

metal salt^aexo^bee (%)^b(%)^cmetal salt^aexo^bee (%)^b(%)^c

Cu(OTf)₂98:2

Ag(OTf) nr^d

Zn(OTf)₂95:5

Cd(OTf)₂90:10

Co(OTf)₂90:10

>98

86 Mn(OTf)₂85:15

Lu(OTf)₃75:25

85 Sm(OTf)₃80:20

50

0

14

40

80

75

78

89

75

38

10

50

80 Li(OTf)

85:15

85 Ni(OTf)₂90:10

Catalysts were prepared by stirring a solution of ligand 6

(1.1-1.2 equiv relative to Cu) and Cu(OTf)₂(typically 10 mol

%, ∼0.03 M in catalyst) in CH₂Cl₂at room temperature for

2-3 h. At this point, the resulting homogeneous green solution

was cooled to -78 °C for the execution of the cycloaddition

reaction.²⁹Ligand-metal complexation time was found to be

critical to the efficacy of the process; when the reaction was

performed after only a 30-min catalyst aging time, irreproducible

results characterized by the formation of nearly racemic cy-

cloadducts with low diastereoselectivity were obtained. In

contrast, when complexation was allowed to occur for longer

periods (2 h or longer), the cycloadducts were obtained with

uniformly high selectivities.

^aReactions were run in CH₂Cl₂with 11 mol % ligand and 10 mol

% M(OTf)_n. ^bEndo/exo ratios were determined by H NMR and/or

1

chiral GLC or HPLC. Enantiomer ratios were determined either by

chiral GLC or by chiral HPLC. ^cValues refer to isolated yields of

cycloadducts. ^dNo reaction.

In parallel with the investigation of bis(oxazoline)-Cu(OTf)₂

Lewis acid catalysts, a number of other metal triflate complexes

of 6d were evaluated (Table 2). Of the triflate salts screened,

only the Cu(II)- and Zn(II)-based catalysts produced product

mixtures with greater than 20:1 endo/exo diastereoselection;

however, the Zn system afforded products of low endo enan-

tioselectivity. Subsequent studies that optimized both ligand

architecture and catalyst counterion for the tetrahedral Zn(II)-

dienophile complex determined that the optimal Zn(II) catalyst

is the phenyl-substituted complex 8a, which affords the enan-

tiomeric cyclopentadiene adduct (R)-7 in >90% yield and 92%

ee (98:2 endo/exo, -78 °C, CH₂Cl₂, 8 h).^14e,31This observation

is consistent with prior studies that have demonstrated that

phenyl-substituted ligands exhibit optimal asymmetric induction

in this reaction when employed with the tetrahedral Mg(II)

Lewis acid (C, Figure 1).³²Quantitative observations on bis-

(oxazoline)copper and -zinc catalysts indicate that the former

are more effective with respect to both conferred enantioselec-

[Cu(box)](OTf)₂complexes derived from ligands 6a-d³⁰

were screened for their ability to catalyze the reaction of imide

(27) 6c, 6d: (a) Reference 2a. (b) Woerpel, K. A. Ph.D. Thesis, Harvard

University, 1992. (c) Evans, D. A.; Peterson, G. S.; Johnson, J. S.; Barnes,

D. M.; Campos, K. R.; Woerpel, K. A. J. Org. Chem. 1998, 63, 4541-

4544. 6a: (d) Reference 13a. 6b: Prepared by Dr. Benjamin A. Anderson,

Harvard University. Evans, D. A.; Anderson, B. A. Unpublished results.

(28) For a review of the coordination chemistry of Cu(II), see: Hathaway,

B. J. In ComprehensiVe Coordination Chemistry; Wilkinson, G., Gillard,

R. D., McClevarty, J. A., Eds.; Pergamon Press: Oxford, 1989; Vol. 5, pp

533-750.

(29) Interestingly, in some cases the color of the catalyst was found to

be thermochromic. For example, at room temperature, the catalyst solution

was bright green, while at -78 °C it had changed color to blue. Workup of

the reaction involved simple dilution with 5 mL of 1:1 ethyl acetate/hexane,

followed by filtration through a short plug of silica. This effectively removes

all Cu salts, and the product can then be analyzed. See Experimental Section

(Supporting Information) for details.

(31) Enantioselection (CH₂Cl₂, -78 °C) as a function of the Zn(II)-

box substituent R is as follows: R ) CMe₃, 56%; R ) CHMe₂, 53%; R )

Ph, 92%, ref 14e.

(32) See refs 13 and 14. The bias for the phenyl substituent with metals

that coordinate in a tetrahedral geometry could result from both steric and

electronic effects. Possible contributions from electronic effects have not

yet been addressed.

(30) The syntheses of ligands 6a, 6c, and 6d may be found in the

supplementary material of ref 2a.

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7563

tion (>98% ee vs 92% ee) and rate acceleration. The most

dramatic difference between the two catalytic centers was found

in their respective temperature/enantioselectivity profiles. While

the Cu(II) catalysts are effective over a wide range of temper-

atures (vide infra), reaction enantioselectivity falls off dramati-

cally for the Zn(II) catalysts at higher temperatures (g-50 °C).

This point has been documented in detail elsewhere.^14eIn a

related comparative study, Engberts and co-workers have

recently evaluated Cu(2+), Zn(2+), Ni(2+), and Co(2+) in

metal catalyzed Diels-Alder reactions with a chelating ketone

dienophile in water. In this comparison, the metal ion catalytic

efficiency was found to be Co(2+) < Ni(2+) < Cu(2+) .

Zn(2+).³³

Figure 3. Conversion to (S)-7 as a function of time for the reaction of

acrylimide 5a with C₅H₆catalyzed by complexes 1a-d at -78 °C (eq

10). The endo ee was >98% in all cases. Endo/exo ratios are provided

on the graph.

The four cationic Cu(II) complexes exhibited significantly

different catalytic activities in the Diels-Alder reaction (Figure

3). While the BF₄- and PF₆-derived catalysts 1c and 1d afforded

rates similar to 1a (X ) OTf), [Cu((S,S)-t-Bu-box)](SbF₆)₂(1b)

displayed greatly enhanced catalytic efficiency. After 1 h, the

percent conversion of the reaction catalyzed by 1b was greater

than that of 1a by a factor of 20. Under the indicated reaction

conditions ([5a] ) 0.1 M, -78 °C, 10 mol % 1),³⁵only SbF₆

complex 1b led to complete conversion to adduct (S)-7a.³⁶

Catalyst 1b also proved to be convenient to handle: the initial

heterogeneous catalyst mixture is filtered through Celite on the

benchtop. All catalysts afforded the endo product in >98% ee;

however, the endo/exo ratio is dependent on the catalyst

employed: the more reactive the catalyst, the less diastereose-

lective the Diels-Alder reaction (1a, endo/exo ) 98:2; 1b, endo/

exo ) 91:9). No explanation for this trend can be offered at

this time.

As a result of this set of experiments, the Cu(II) system

emerged as the most promising candidate for further develop-

ment with the bis(oxazoline) family of ligands.

Counterion Effects. In the early stages of this study, Cu-

(OTf)₂was chosen as the Cu(2+) source with the assumption

that the triflate counterions in the metal-ligand complex would

either dissociate or be displaced readily by the chelating

dienophile. On the other hand, Lewis acid catalysis through the

partially dissociated ligand-metal complex, L₂CuX(1+), could

well erode the performance of the fully dissociated Lewis acid

L₂Cu(2+) counterpart. In addition, significant differences in

Lewis acidity might be expected for the two species. Accord-

ingly, copper(II) complexes 1b-d were prepared by mixing

CuCl₂(or CuBr₂, 0.1 mmol), bis(oxazoline) ligand 6d (0.11

mmol), and the appropriate silver salt AgX (0.2 mmol) in 1

mL of CH₂Cl₂.³⁴Complexation time again proved important;

vigorous stirring at room temperature for 6-8 h was required

in order to provide fully active solutions. The resulting

heterogeneous mixtures were filtered through Celite to give clear

blue solutions of the bis(oxazoline)copper(II) complexes 1b-

d, which were used directly as catalysts (10 mol %) in the

Diels-Alder reaction of acrylate imide 5a and cyclopentadiene

(eq 10).

Table 3. Temperature Profile of Reactions of Imide 5a with

Cyclopentadiene Catalyzed by [Cu((S,S)tert-Bu-box)](X)₂(1) (Eq

10)^a

entry

catalyst

temp (°C) time^bendo/exo^cendo ee (%)^c

1

2

3

4

1a (X ) OTf)

1b (X ) SbF₆)

1a (X ) OTf)

1b (X ) SbF₆)

-78

25

10 h

4 h

15 min

10 min

98:2

96:4

88:12

86:14

>98

86

25

94

^aReactions were run in CH₂Cl₂with 10 mol % catalyst, [5a] ) 0.3

M. ^b100% conversion. ^cEndo/exo ratios were determined by ¹H NMR

and/or chiral GLC or HPLC. Enantiomer ratios were determined either

by chiral GLC or by chiral HPLC.

Catalyst Properties. [Cu((S,S)-t-Bu-box)](SbF₆)₂complex 1b

was further compared to [Cu((S,S)-t-Bu-box)](OTf)₂(1a). As

indicated in Table 3, both catalysts function with almost

complete fidelity over a range of temperatures. The reaction is

(34) Ag(I) salts have been used to generate cationic metal centers in

other Diels-Alder catalysts. See ref 13b. See also: Hayashi, Y.; Rohde, J.

J.; Corey, E. J. J. Am. Chem. Soc. 1996, 118, 5502-5503.

(35) The optimized conditions include a substrate concentration of 0.3

M. To allow an accurate % conversion determination, the reactions illustrated

graphically in Figure 2 were [5a] ) 0.1 M.

(36) While the study presented in Figure 3 indicates that the Diels-

Alder reaction of acrylimide 5a with cyclopentadiene is complete after 1 h

at -78 °C, the reactions presented below were usually run for 4-8 h to

ensure reproducibility of the results.

(33) Otto, S.; Bertoncin, F.; Engberts, J. B. F. N. J. Am. Chem. Soc.

1996, 118, 7702-7707.

7564 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Table 5. Solvent Effects on the Reaction of Imide 5a with

almost completely selective at -78 °C (entries 1 and 2, endo

ee > 98%), and the results are still good (entry 3, endo/exo )

88:12, endo ee 86%) to excellent (entry 4, endo/exo ) 86:14,

endo ee 94%) at room temperature. At this temperature, the

reaction is noticeably exothermic and is complete within 10-

15 min. This exceptional temperature profile foreshadowed the

extension of the system to less reactive dienophiles (vide infra).

[Cu((S,S)-t-Bu-box)]X₂complexes 1a and 1b also perform

well at low catalyst loadings. Although reactions were typically

screened with 10 mol % of catalyst (Table 4, entries 1 and 2),

as little as 1 mol % of SbF₆catalyst 1b led to excellent results

at -78 °C (entry 6). On the other hand, at catalyst loadings

below 5 mol %, OTf complex 1a did not promote the reaction

well, even at -50 °C (entries 3 and 5).

Cyclopentadiene Catalyzed by [Cu((S,S)-tert-Bu-box)]X₂(1) (Eq

10)^a

time (h)

endo/ endo

catalyst

solvent (temp, °C)

(conv, %) exo^b

ee^c

1a (X ) OTf) CH₂Cl₂(-50)

1b (X ) SbF₆) CH₂Cl₂(-50)

1a (X ) OTf) THF (-50)

1b (X ) SbF₆) THF (-50)

1a (X ) OTf) CH₃NO₂(-25)

1b (X ) SbF₆) CH₃NO₂(-25)

1a (X ) OTf) CH₃CN (-50)

1b (X ) SbF₆) CH₃CN (-50)

4 (100)

1.5 (100)

4 (100)

nr^e

4 (100)

2.5 (100)

4 (87)

97:3

91:9 >98

97:3

98

92:8

96:4

92:8

98:2

97:3

84

97

58

50

84

61

3.5 (97)

1a (X ) OTf) CH₂Cl₂/i-PrOH^d(-50) 4 (50)

1b (X ) SbF₆) CH₂Cl₂/i-PrOH^d(-50) 2.5 (100)

^aReactions were run with 10 mol % 1. ^bEndo/exo ratios were

determined by ¹H NMR and/or chiral GLC or HPLC. ^cEnantiomer

ratios were determined by either chiral GLC or chiral HPLC. ^dRun

with 1 equiv of i-PrOH (10 equiv relative to catalyst). ^eNo reaction.

Table 4. Effect of Catalyst Loading in Reactions of Imide 5a with

Cyclopentadiene Catalyzed by [Cu((S,S)-tert-Bu-box)](X)₂(1) (Eq

10)^a

catalyst

mol % temp (°C) time (h)^bendo/exo^cendo ee (%)^c

previous Diels-Alder studies of the chiral imides (Scheme 1),^7b

we felt that it was reasonable to assume that two-point catalyst-

dienophile binding and reaction out of the s-cis dienophile

conformation were also occurring in these catalyzed processes.

These assumptions being made, the issue of the metal coordina-

tion geometry may be addressed. As can be seen from the

drawings of the dienophile-catalyst complexes (A), the sense

of asymmetric induction is dependent on the geometry of the

metal center if other structural variables are held constant. In

fact, the stereochemical outcome of the catalyzed reactions is

consistent with a four-coordinate square-planar Cu(II) geometry

and accompanying s-cis dienophile geometry; these geometric

constraints correlate the (S,S)-bis(oxazoline) ligand chirality to

the observed product stereochemistry of cycloadduct (S)-7. The

alternate interpretation of the observed sense of asymmetric

induction would require the unlikely combination of a tetrahedral

metal center and an s-trans dienophile conformation.

1a (X ) OTf)

1b (X ) SbF₆) 10

1a (X ) OTf)

1b (X ) SbF₆)

1a (X ) OTf)

1b (X ) SbF₆)

10

-78

-50

-78

-50

-78

10

4

98:2

96:4

97:3

96:4

>98

96

5

1

12

4

96

nr^e

8

96:4

96

^aReactions were run in CH₂Cl₂; [5a] ) 0.3 M. ^b100% conversion.

^cEndo/exo ratios were determined by H NMR and/or chiral GLC or

1

HPLC. ^dEnantiomer ratios were determined by chiral GLC or chiral

HPLC. ^eNo reaction.

A study of solvent effects showed that catalysts 1a (X )

OTf) and 1b (X ) SbF₆) are tolerant of some donor solvents

(Table 5). When the reaction was run in THF at -50 °C (entry

3), high selectivities and undiminished reaction rates were

observed with [Cu((S,S)-t-Bu-box)](OTf)₂(1a) (endo/exo ) 97:

3, endo ee ) 98%, 4 h reaction time). This is particularly

significant, as it suggests that bidentate coordination of the

substrate was not compromised by the donor solvent. However,

attempts to generate 1b (X ) SbF₆) in THF failed to result in

the formation of soluble complex (entry 4). The use of

nitromethane had little effect on the course of the reaction with

either catalyst (entries 5 and 6).³⁷This solvent tolerance is

limited to non-hydroxylic oxygen-based donors, as more basic

solvents impaired the reaction. For example, when acetonitrile

was employed with either catalyst at -50 °C (entries 7 and 8),

the rate, endo/exo ratio, and enantioselectivity were all dimin-

ished. Addition of 1 equiv of 2-propanol to the reaction in CH₂-

Cl₂employing 1a (X ) OTf) retarded the rate and selectivities

significantly (50% conversion after 4h at -50 °C; endo/exo )

95:5, endo ee ) 85%). In the case of 1b (X ) SbF₆), 2-propanol

had little effect on the reaction rate, though the selectivity

decreased markedly. In reactions of piperlyene at higher

temperatures, however, 2-propanol had no effect on reaction

selectivity (vide infra). Addition of 1 equiv of a tertiary amine

base effectively terminated catalysis.

In conjunction with our efforts to understand this class of

copper catalysts, we undertook an investigation of the structural

basis for the remarkable reactivity and selectivity exhibited by

this ligand-metal system. Attempts to characterize the solution-

phase complexes by NMR experiments were unrevealing due

to broadening associated with the paramagnetic d⁹Cu center.

Efforts to gain insight into solid-state structure using X-ray

crystallography proved more fruitful, and three [Cu((S,S)-t-Bu-

box)]X₂X-ray crystal structures were obtained.³⁸Despite

extensive efforts to obtain ternary complexes of 1a and 1b

incorporating dienophile 5a, the highly crystalline bis(aquo)

complexes 9a and 9b were repeatedly obtained, even under the

nominally anhydrous conditions afforded by a drybox. Prepara-

tively, [Cu((S,S)-t-Bu-box)(H₂O)₂](OTf)₂(9a) was produced by

exposure of a green CH₂Cl₂solution of 1a to the atmosphere;

blue crystals of the hydrated complex gradually formed at room

temperature. The analogous bis(aquo) SbF₆complex 9b was

Solid-State Catalyst Characterization and Catalyst-

Substrate Complex Geometry. The interpretation of the sense

of asymmetric induction in the Diels-Alder reaction (eq 6)

depends on an assessment of the following structural variables

in the dienophile-catalyst complex: metal center geometry

(square-planar vs tetrahedral), dienophile conformation (s-cis

vs s-trans), and finally, the number of contacts between catalyst

and dienophile (one- vs two-point binding). By analogy to our

(37) Jørgensen has recently reported the use of nitromethane in [Cu-

(box)]X₂-catalyzed hetero Diels-Alder and ene reactions. Johannsen, M.;

Jørgensen, K. A. Tetrahedron 1996 52, 7321-7328.

(38) The structures of 9a and 9b have been reported. 9a: (a) Evans, D.

A.; Olhava, E. J.; Johnson, J. S.; Janey, J. M. Angew. Chem., Int. Ed. Engl.

1998, 37, 3372-3375. 9b: (b) Reference 27c.

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7565

Figure 4. Crystallographic structures of bis(aquo) complexes 9a-X-ray and 9b-X-ray, and dichloride complex 10-X-ray, and calculated structure

of A-PM3, along with selected bond lengths and dihedral angles.

similarly produced by exposure of 1b to atmospheric moisture.

Diffractable crystals of [Cu((S,S)-t-Bu-box)]Cl₂(10) were

prepared by slow evaporation of a CH₂Cl₂solution.

) 3.667 Å). The average degree of distortion of the coordinated

water molecules from the copper-ligand plane is 26°.⁴⁰The

crystallographic data for SbF₆complex 9b (Figure 4) demon-

strate that neither counterion is coordinated to the Cu center,

which is disposed in a distorted square-planar geometry, with

an average degree of distortion of the coordinated water

molecules from the copper-ligand plane of 33°. The structure

of C₂-symmetric dichloride complex 10 likewise displays a

distorted (37°) square-planar geometry.

The rate-retarding effect of the triflate-based catalyst 1a can

be explained by invoking triflate association with the catalyst

substrate-complex, as suggested by structure 9a. IR spectros-

copy affords the opportunity of determining whether triflate

counterions are dissociated in solution through observation of

the characteristic vibrational frequencies which are attributed

to bound and unbound triflate ions.⁴¹Frequencies at both 1380

(bound -OSO₂CF₃) and 1280 cm^-1(unbound -OSO₂CF₃) were

observed, presumably due to the presence of multiple species.

In speculating on how imide dienophiles associate with 1a

and 1b, one might replace the bound equatorial ligands in the

X-ray structures with the imide carbonyl oxygens. Inspection

of molecular models reveals that, when the imide dienophile is

superimposed onto the positions occupied by the H₂O (or Cl)

ligands in the X-ray structures, diene trajectory analysis leads

to the prediction of the correct cycloadduct enantiomer. It is

apparent that small distortions from square-planarity do not

result in exposure of the opposite olefin diastereoface until the

distortion more closely resembles the tetrahedral geometry.

Calculations were performed to provide representations of

the catalyst-substrate complex. Calculated [Cu((S,S)-t-Bu-box)-

The X-ray structures of these complexes are illustrated in

Figure 4 along with selected bond lengths and angles. Notable

features of these structures include monomeric, bidentate

coordination of the bis(oxazoline) ligand to the Cu center and

coordination spheres in which the bis(oxazoline) and water (or

chloride) ligands are arrayed in geometries intermediate between

planar and tetrahedral. This former feature is in marked contrast

to the helical, polymeric bis(oxazoline)-Cu(I) complex which

has been prepared in this laboratory.³⁹

The crystallographic data for triflate complex 9a reveal a

distorted square-pyramidal geometry, with one triflate counterion

weakly bound to the metal center in the apical position (Cu-

OTf ) 2.624 Å), while the other is fully dissociated (Cu-OTf

(40) Four-coordinate Cu(II) complexes exhibit a strong tendency toward

square-planar geometries. See ref 28.

(41) Lawrence, G. A. Chem. ReV. 1986, 86, 17-33.

(39) Evans, D. A.; Woerpel, K. A.; Scott, M. S. Angew. Chem., Int. Ed.

Engl. 1992, 31, 430-432.

7566 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Scheme 2

(acrylimide)](2+) complex A-PM3 is illustrated in Figure 3. It

is noteworthy that the degree of distortion from square-planarity

was approximately that found in the X-ray structure of bis(aquo)

complex 9b. Furthermore, the complex clearly predicts the

stereochemical outcome of the Diels-Alder reactions.^7a,cThese

structural data complement solution-phase EPR data, which are

consistent with square-planar catalyst-acrylimide complexes.⁴²

In summary, we believe that the X-ray and computed structural

data support a catalyst-substrate geometry which is, indeed,

square-planar, but which is distorted from the idealized dihedral

angles to some extent.

1) to afford the adduct (S,R)-4 in 87% isolated yield. In contrast,

with (S)-2, the reaction proceeded to only 20% conversion

during the 11-h period with very low diastereoselectivity (endo/

exo ) 97:3, endo(1)/endo(2) ) 68:32). It is noteworthy that, in

the mismatched case, the principle endo adduct (S,S)-4 corre-

sponds to that deriVed from the catalyst-controlled, rather than

the substrate-controlled process, a testament to the stereochem-

ical dominance of the 6d-Cu(OTf)₂catalyst and to the torsional

stability of the square-planar copper center.

The preceding experiments suggest that the Diels-Alder

reactions of (R)- and (S)-2 should revert to conventional

substrate-based stereoinduction as the stereochemical bias

imparted by the copper catalyst is attenuated. This projected

trend was evaluated with the poorly enantioselective phenyl-

substituted complex 6a-Cu(OTf)₂(Table 1), and the results

were in accord with prediction (eqs 11 and 12). As expected,

the stereochemically matched catalyst and dienophile (R)-2 (eq

11) (CH₂Cl₂, -78 °C, 11 h, 10 mol % 6a-Cu(OTf)₂) afforded

the expected cycloadduct (S,R)-4 with high diastereoselectivity

(100% conversion, endo/exo ) 99:1, endo(1)/endo(2) ) 96:4).

In the mismatched alternative, using dienophile (S)-2 (eq 12),

the reaction likewise proceeded to completion (-78 °C, 11 h),

affording (R,S)-4 as the major product (endo/exo ) 95:5,

endo(1)/endo(2) ) 9:91). This stereochemical outcome indicates

that the bias imparted by the dienophile now dominates the

process, and while the 6a-Cu(OTf)₂complex is providing the

necessary chelate organization and Lewis acid activation for

the cycloaddition, its stereochemical influence is minimal.

The preceding analysis indicates that the corresponding chiral

tetrahedral metal-bis(oxazoline) complexes should exhibit the

opposite sense of asymmetric induction. In this regard, it is

noteworthy that Zn(II) catalyst 8a and Corey’s Mg(II)-derived

catalyst C (Figure 1), which are believed to proceed through

tetrahedral catalyst-substrate complexes, produce the (R)-7

cycloadduct as the major product, in accord with expectation.

Double-Stereodifferentiating Experiments. In the absence

of X-ray structural confirmation of the proposed catalyst-imide

complex, we investigated double-stereodifferentiating experi-

ments with the chiral imides (R)-2 and (S)-2 and chiral triflate

catalyst 1a. These experiments were designed to determine

whether the principal assumptions of two-point catalyst-

dienophile binding, metal center geometry, and dienophile

conformation that were made in both the auxiliary-based reaction

and the chiral-catalyzed counterpart behaved in an internally

consistent fashion. The stereochemically matched and mis-

matched reactions are illustrated in Scheme 2. If the catalyst-

substrate complex maintains a square-planar coordination

geometry during the catalyzed process, then the reaction of (R)-2

should represent the stereochemically matched case, where the

chiral elements from both ligand and substrate cooperate to block

the Si face of the dienophile. In contrast, use of the enantiomeric

dienophile (S)-2 should present a situation where one face of

the dienophile is blocked by the substrate substituent, and the

other by the ligand substituent, creating a mismatched case. The

intermediacy of a tetrahedral coordination environment for the

catalyst-substrate complex would predict the opposite results.

Results fully consistent with the square-planar model involv-

ing the s-cis chelated dienophile were obtained. Under identical

reaction conditions (CH₂Cl₂, -78 °C, 11 h, 10% 1a (X ) OTf),

dramatically different results were observed using the enantio-

meric dienophiles (R)- and (S)-2. In the matched case employing

(R)-2, the reaction proceeded to completion with essentially

perfect stereocontrol (endo/exo >99:1, endo(1)/endo(2) ) 99:

(42) Kozlowski, M. C., unpublished results.

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7567

Anhydrous vs Aquo Complexes.⁴³The isolation of crystal-

line aquo complexes [Cu((S,S)-t-Bu-box)(H₂O)₂](OTf)₂(9a) and

[Cu((S,S)-t-Bu-box)(H₂O)₂](SbF₆)₂(9b) presented the possibility

that they could be employed as bench-stable Diels-Alder

catalysts. Therefore, these complexes were compared with their

anhydrous counterparts 1a (X ) OTf) and 1b (X ) SbF₆) in

their ability to catalyze Diels-Alder reactions. Since it has been

demonstrated that the reaction of imide 5a with piperylene is

catalyzed by either 1a or 1b with excellent selectivities at

ambient temperature,⁶this reaction was employed to determine

the relative reactivities of the various complexes (eq 13, Table

6).

Table 6. Diels-Alder Reactions of Imide 5a with Piperylene

Catalyzed by [Cu(t-Bu-box)(H₂O)_n]X₂Complexes (Eq 13)^a

catalyst

sieves time conv (%)^bselectivity^c,dee, 11 (%)^d

1a (X ) OTf)

9a (X ) OTf)

no 15 h

no 24 h

yes 15 h

94

<10

95

100

10

73:27

nd

74:26

82:18

83:17

nd

84

nd

85

95

94

nd

Triflate complex 1a (10 mol %) promoted the reaction to

94% conversion in 15 h. Cis cycloadduct 11 was formed in

84% ee, and the major enantiomer accounted for 73% of the

reaction product mixture (entry 1). When bis(aquo) complex

9a (X ) OTf) was employed under equivalent conditions, less

than 10% conversion took place in 24 h, attesting to the

detrimental effect of hydration on this catalyst (entry 2).

Molecular sieves proved efficacious in reactivating the catalyst,

however. When complex 9a was stirred first with 3-Å sieves

for 2.5 h, and then employed in the reaction, 95% conversion

was obtained after 15 h (entry 3). The product was formed with

the same selectivity as in the reaction employing the anhydrous

catalyst.

1b (X ) SbF₆) no 50 min

9b (X ) SbF₆) no 70 min

9b (X ) SbF₆) yes

4 h

The response of [Cu((S,S)-t-Bu-box)](SbF₆)₂(1b) to hydration

was surprisingly different. The anhydrous catalyst required only

50 min to promote the reaction to complete conversion; the

major isomer was formed in 95% ee and accounted for 82% of

the reaction mixture (entry 4). Bis(aquo) complex 9b was almost

as active: the reaction proceeded to completion in only 70 min,

with equivalent selectivities (entry 5). However, while sieves

fully regenerated the triflate catalyst, they suppressed the activity

of SbF₆catalyst 9b, and only 10% conversion was observed

after 4 h. The fact that hydration does not have a large effect

on [Cu((S,S)-t-Bu-box)](SbF₆)₂(1b) suggested that this catalyst

should tolerate hydroxylic additives. As described above, the

addition of 1 equiv of 2-propanol (10 equiv relative to catalyst)

to reactions of acrylimide 5a with cyclopentadiene dramatically

lowered both rates and selectivities at -50 °C (Table 5). When

2-propanol was similarly added to the reaction of piperylene

catalyzed by 1b (X ) SbF₆) at room temperature, the reaction

rate was halved, but the selectivity was unchanged. We postulate

that, at -50 °C, 2-propanol remains associated with the catalyst

throughout the reaction.

^aReactions were run in CH₂Cl₂, with 10 mol % catalyst and 10 equiv

of piperylene. In reactions with sieves, the catalyst was stirred with

3-Å molecular seives for 150 min before reaction. ^bConversion was

determined by NMR analysis of unpurified reaction mixture. ^cRatio

refers to major cis:∑(other isomers). ^dSelectivity and major enantio-

selectivity were determined by chiral GC analysis of the unpurified

reaction mixture. nd ) not determined.

â-Substituted Dienophiles. We next sought to examine the

generality of the reaction with cyclopentadiene in the context

of modifying the dienophile structure. When â-substituted

dienophiles were employed in the cycloaddition reaction, good

reactivities and selectivities were observed throughout the range

of imides surveyed (eq 14, Table 7). While [Cu((S,S)-t-Bu-box)]-

(OTf)₂(1a) provided highly enantioselective reactions, it was

in the investigation of these substrates that the potential of the

SbF₆-derived catalyst 1b became fully apparent. For example,

with 1a (X ) OTf) as catalyst, crotonate derivative 5b

underwent a highly enantioselective reaction with cyclopenta-

diene at -15 °C (30 h), affording cycloadduct 12b in 85% yield

after 30 h (entry 2, endo/exo ) 96:4, endo ee ) 97%). Using

1b (X ) SbF₆), the cycloadduct was isolated in 99% yield and

a remarkable 99% ee after only 20 h at -15 °C (entry 4, endo/

exo ) 85:15); these numbers are taken from a large-scale

reaction employing 150 mmol (23 g) of crotonate imide 5b. At

room temperature, both catalysts afforded high yields of 12b

after 8 h, with only slightly diminished enantioselectivities

(entries 1 and 3, endo ee ) 94 and 96%, respectively). In the

case of cinnamate derivative 5c reaction at 25 °C was required

with 1a (X ) OTf) for a reasonable rate of conversion, but

good selectivity was retained and product 12c was isolated in

85% yield after 24 h (entry 6, endo/exo ) 90:10, endo ee )

90%). Under the same conditions (25 °C, 24 h) with catalyst

1b (X ) SbF₆), 12c was isolated in improved yield (96%) and

Based on the preceding observations, we postulate that an

open fifth coordination site on the Cu(2+) center is required

for catalyst turnoVer through the associatiVe displacement of

the bound neutral oxygen ligands (water or Diels-Alder adduct)

by additional dienophile (Table 2). The X-ray structures of the

hydrated copper-ligand complexes 9a and 9b (Figure 4)

illustrate that one of the triflate counterions in 9a remains bound

to the metal center in the solid state. Evidence accumulated to

date implicates triflate ion association throughout the catalytic

cycle. These subtle observations on the influence of counterion

structure on catalyst turnover were critical to our development

of the glyoxylate ene reaction, which was initially observed as

a stoichiometric process with [Cu((S,S)-t-Bu-box)](OTf)₂(1a)

and later rendered catalytic with [Cu((S,S)-t-Bu-box)](SbF₆)₂

(1b) and its hydrated counterpart 9b.⁴⁴

(43) Engberts has recently reported water-enhanced enantioselectivity

in a Diels-Alder reaction: Otto, S.; Boccaletti, G.; Engberts, J. J. Am.

Chem. Soc. 1998, 120, 4238-4239. For other uses of aquo complexes in

Diels-Alder reactions, see refs 21, 24c, and 38a.

(44) Evans, D. A.; Burgey, C. S.; Paras, N. A.; Vojkovsky, T.; Tregay,

S. T. J. Am. Chem. Soc. 1998, 120, 5824-5825.

7568 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Table 7. Enantioselective Diels-Alder Reactions of â-Substituted Dienophiles 5 and 13 with Cyclopentadiene Catalyzed by 1a (X ) OTf)

and 1b (X ) SbF₆) (Eqs 14 and 15)^a

entry

dienophile

catalyst

time (h) (temp, °C)

endo/exo^b

endo ee (%)^c

yield (%)^d

product

1

2

3

4

5

6

7

8

9

10

11

12

13

5b (R ) Me)

13b (R ) Me)

5c (R ) Ph)

1a (X ) OTf)

1b (X ) SbF₆)

1a (X ) OTf)

1b (X ) SbF₆)

1a (X ) OTf)

1a (X ) OTf)^g

1b (X ) SbF₆)

1a (X ) OTf)^g

8 (25)

30 (-15)

8 (25)

20 (-15)

36 (-45)

24 (25)

48 (-10)

24 (25)

48 (-10)

72 (-35)

20 (-55)

87:13

96:4

82:18

85:15

96:4

90:10

93:7

81:19

88:12

92:8

94

97

96

99

95

85

98

99

82

85

16

96

77

86

92

88

12b (R ) Me)

14b (R ) Me)

12c (R ) Ph)

14c (R ) Ph)

12d (R ) CO₂Et)

14d (R ) CO₂Et)

94^e

90^f

94^f

96^f

98^f

97^f

95^h

87^h

96^h

5c (R ) Ph)

13c (R ) Ph)

5d (R ) CO₂Et)

13d (R ) CO₂Et)

94:6

82:18

84:16

^aReactions were run at the indicated temperature in CH₂Cl₂with 10 mol % catalyst, unless otherwise noted. ^bDiastereomer ratios were determined

1

by H NMR. ^cEnantiomer ratios were determined by chiral HPLC or chiral GLC. ^dValues refer to isolated yields of cycloadducts. ^eEnantiomer

f

ratio was determined by conversion to 12b. Enantiomer ratio was determined by conversion to the corresponding R-methyl benzyl amide. ^gReactions

were performed in the presence of 5 mol % 1a (X ) OTf). ^hEnantiomer ratio was determined by conversion to the corresponding iodolactone.

enantiomeric purity (96% ee, entry 8). Lowering the reaction

temperature to -10 °C when employing catalyst 1b (X ) SbF₆)

provided the product in 98% ee and 77% isolated yield (48 h,

entry 9). Catalyst 1a led to an impractically slow reaction at

this temperature, albeit with high selectivity (94% ee, entry 7).

The fumarate derivative 5d was very reactive and was trans-

formed to 12d in 92% isolated yield at low temperature (-55

°C), employing only 5 mol % catalyst 1a (X ) OTf) (entry

11). Selectivities were high with this substrate (endo/exo ) 94:

6, endo ee ) 95%), despite the possibility that Lewis acid

activation at the ester terminus threatened to erode the selectivity

for this substrate. This alternative binding mode may happen,

though, in the case of catalyst 1b (X ) SbF₆), which afforded

the cycloadduct in 88% yield (endo/exo ) 82:18), with a

diminished enantioselectivity of 87% ee (entry 12).

To increase the reactivity of the â-substituted dienophiles,

we investigated the use of R,â-unsaturated thiazolidine-2-thione

derivatives 13b-d (eq 15, Table 7).⁴⁵Though there was some

concern that sulfur-based dienophiles would bind irreversibly

to the copper(II) center, poisoning of catalyst 1a (X ) OTf)

was not observed. The Diels-Alder reaction of crotonate-

derived 13b (entry 5) proceeded to completion within 36 h at

-45 °C, and product 14b was formed with selectivity compa-

rable to oxygen analogue 12b (endo/exo ) 96:4, endo ee )

94%). In the case of cinnamate-derived 13b, the greater substrate

reactivity permitted lower reaction temperatures, resulting in

enhanced selectivities (entry 10). Complete reaction was

observed after 72 h at -35 °C, and the cycloadduct was obtained

in 86% isolated yield (endo/exo ) 92:8, endo ee ) 97%). Again,

fumarate dienophile 13d was exceptionally reactive; the reaction

proceeded efficiently with 5 mol % catalyst at -55 °C, and

cycloadduct 14d was obtained in 88% yield as a 84:16 mixture

of endo/exo isomers (entry 13, endo ee ) 96%).

While poisoning was not observed in reactions employing

complex 1a (X ) OTf), catalyst 1b (X ) SbF₆) was inactivated

upon addition of the sulfur-containing substrates, which were

recovered in moderate yields (60-70%). A control reaction of

13b (R ) Me) with cyclopentadiene in the presence of

stoichiometric quantities of copper(II) complex 1b (X ) SbF₆)

afforded adduct 14b in 78% yield and 94% ee (10 h, 0 °C),

highlighting turnover as the problematic step in this reaction.

Using [Cu((S,S)-t-Bu-box)](OTf)₂(1a), a direct competition

experiment between crotonate derivatives 5b and 13b showed

the latter to be 20 times more reactive; similarly, 5c (R ) Ph)

was found to be roughly 30 times more reactive than 13c (eq

16). While these observations may reflect a greater affinity of

thiazolidine-2-thione derivatives for the catalyst relative to their

oxygen counterparts more than any inherent increase in dieno-

philicity, the end result is that, for catalyst 1a (X ) OTf),

â-substituted sulfur-based dienophiles uniformly afford en-

hanced reactivity in the catalyzed Diels-Alder reaction.

(45) These heterocycles exhibit higher reactivity in acyl-transfer reactions.

See, for example: Hsiao, C. N.; Liu, L.; Miller, M. J. J. Org. Chem. 1987,

52, 2201-2206.

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7569

Scheme 3^a

reaction to only 60% conversion after a 24-h period at 25 °C,

affording product 20 in 41% ee. SbF₆-derived 1b likewise

afforded the product in 52% ee and in 65% yield. These results

are congruent with earlier studies from our laboratory involving

alkynes in auxiliary-based reactions.^7aAcetylenes present a

special challenge for the Diels-Alder process since they possess

two orthogonal sets of π-orbitals, each of which can, in principle,

participate in a cycloaddition reaction.⁴⁹In addition, the C1-

C2-C3 bond angle of 180° of the â-substituted propiolate

creates an environment wherein the π-system undergoing

cycloaddition is projected away from the chirality associated

with the Lewis acid catalyst. In contrast, the 120° C1-C2-C3

bond angle of â-substituted acrylates projects the reacting double

bond back toward the chiral substituent.

^aConditions: (a) 1.4 equiv of Et₂AlCl, CpH, CH₂Cl₂, -78 °C; (b)

LiSEt, THF, 0 °C; (c) Hg(O₂CCF₃)₂, THF/H₂O, I₂; (d) LiOH, THF/

H₂O; (e) I₂, CH₃CN.

The absolute stereochemical assignments of cycloadducts

12b,c and 14b,c were secured by conversion to the benzyl esters,

which were compared to material derived from Me₂AlCl-

promoted cycloaddition reactions employing chiral N-acyl imide

dienophiles.⁷The absolute stereochemical assignments of cy-

cloadducts 12d and 14d were also made by correlation to the

product of an auxiliary-based Diels-Alder reaction (Scheme

3). Treatment of (S)-15 with Et₂AlCl (1.4 equiv, -78 °C)

followed by cyclopentadiene (30 min, -78 °C) afforded an 87:

13 mixture of endo/exo cycloadducts (endo(1)/endo(2) ) 95:

5). The absolute stereochemistry of the major endo cycloadduct

16 was assigned by analogy to our earlier studies.⁷Cleavage of

the oxazolidinone auxiliary was accomplished by treatment of

16 with LiSEt according to the method of Damon.⁴⁶Conversion

of thioester 17 to iodolactone (-)-18 was then achieved by

treatment with aqueous Hg(O₂CCF₃)₂, followed by trapping of

the intermediate organomercurial species with I₂. The absolute

stereochemical assignment (-)-18 was confirmed by converting

the product to the known (5S,6S)-bis(hydroxymethyl)bicyclo-

[2.2.1]hept-2-ene and comparing the optical rotation (90% ee

by chiral GLC, [R]_D-18° (c 0.14, CHCl₃); lit. [R]_D-23° (c

0.80, CHCl₃).⁴⁷

Diels-Alder adducts 12d and 14d were also converted to

the iodolactones. Cycloadduct 12d underwent a procedure

analogous to that used for adduct 16. Cycloadduct 14d was

converted to the corresponding acid, and subsequent iodolac-

tonization then afforded the desired product. As anticipated, the

products obtained in reactions catalyzed by copper(II) catalysts

1a (X ) OTf) and 1b (X ) SbF₆) afforded (+)-18, confirming

that the Cu(II)-catalyzed cycloaddition reaction is stereoregular

throughout the range of â-substituted dienophiles examined.

An alternative tactic would be to employ a â-substituted imide

incorporating a group that could be eliminated after the

cycloaddition. â-Chloroacrylimide 21 was viewed as a possible

acetylene equivalent that would hold the reacting centers in

proximity to the ligand chirality (eq 18).⁵⁰With catalyst 1a (X

) OTf), cycloadduct 22 was obtained with modest selectivity

(endo/exo ) 93:7, endo ee ) 53%). Gratifyingly, the use of

SbF₆-derived catalyst 1b afforded 22 after 18 h at room

temperature in 96% yield, with an excellent endo enantiose-

lectivity of 95% (endo/exo ) 87:13, exo ee 85%). These

numbers were taken from a experiment run on 62 mmol (11 g)

of 21. Recrystallization yielded enantiomerically pure 22. Trans-

esterification to methyl ester 23 (MeOMgBr, 0 °C, 1 h), followed

by elimination (2 equiv of DBU, room temperature, 3 h), cleanly

afforded 2-(methoxycarbonyl)norbornadiene (24) (eq 19).

Acetylenes and Acetylene Equivalent Dienophiles. An

effective acetylenic dienophile in the asymmetric Diels-Alder

reaction would represent a highly desirable and novel approach

to chiral, pharmacologically active synthetic intermediates.⁴⁸

Butynoate-derived imide 19 was prepared and studied in the

catalyzed cycloaddition with cyclopentadiene (eq 17). The

reactivity of 19 was low; catalyst 1a (X ) OTf) promoted the

(46) (a) Damon, R. E.; Coppola, G. M. Tetrahedron Lett. 1990, 31,

2849-2852. (b) See also ref 8.

(47) (-)-18 was subjected to a two-step sequence [(a) t-BuLi, (b) LAH]

to afford the diol, which has been reported: Horton, D.; Machinami, T.;

Takagi, Y. Carbohydr. Res. 1983, 121, 135-161.

(48) For example, chiral norbornadienes would be useful starting

materials for the synthesis of prostaglandins. Corey, E. J.; Shibasaki, M.;

Nicolaou, K. C.; Malmsten, C. L.; Samuelson, B. Tetrahedron Lett. 1976,

17, 737-740.

(49) For successful approaches to this problem, see: (a) Ishihara, K.;

Kondo, S.; Kurihara, H.; Yamamoto, H.; Ohashi, S.; Inagaki, S. J. Org.

Chem. 1997, 62, 3026-3027. (b) Corey, E. J.; Lee, T. W. Tetrahedron

Lett. 1997, 38, 5755-5758.

(50) Norris, W. P. J. Org. Chem. 1968, 33, 4540-4541.

7570 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Figure 5. Pseudo-first-order kinetics for the early stages of reaction

of 5b with C₅H₆and catalyst 1a (X ) OTf) (eq 20).

Figure 6. Downward curvature in the log plot of the reaction of 5b

with C₅H₆and catalyst 1a (X ) OTf) (eq 20).

Reaction Kinetics. The possibility of product inhibition

affecting the catalysis of the Diels-Alder reaction, while often

implied in various catalytic processes, has rarely been quanti-

fied.⁵¹Fortunately, through some relatively straightforward

experiments, the role of product inhibition can be clearly

diagnosed.⁵²Kinetics experiments were performed at 0 °C on

the Diels-Alder reaction of the well-behaved crotonate imide

5b, where cyclopentadiene (10 M) was present in large excess

relative the concentration of dienophile (0.49 M), which was,

in turn, large relative to the concentration of catalyst 1a (X )

OTf, 0.049 M, eq 20).⁵³The background reaction was deter-

mined to be negligible at 0 °C, and cyclopentadiene dimerization

was slow relative to the overall rate of the reaction.⁵⁴Aliquots

were taken at specified time intervals and quickly quenched in

a rapidly stirred ice-cold mixture of ethyl acetate/saturated

NaHCO₃, and the residual cyclopentadiene was removed under

vacuum at 0 °C. Products were quantified by GC analysis with

4,4′-di-tert-butyldiphenyl as the internal standard.

When a competitive inhibitor (In) is present, the kinetic

scheme shown in eq 21 applies, where K_inrepresents the

equilibrium constant for dissociation of the inhibitor-catalyst

complex (eq 22):

The competitive inhibitor can be the product of the reaction

itself. For instance, when a kinetics run was performed in the

presence of added adduct 12b at 0 °C ([12b] ) 0.98 M at t )

0), an 18% reduction in rate was observed. Once again, the

reaction followed pseudo-first-order kinetics at low conversions

of substrate. For a given concentration of inhibitor, we can

quantify the degree of inhibition as

ν_o- ν

ν_o

ꢀ )

(23)

Under these conditions, the reaction was found to follow first-

order kinetics at low substrate conversions (rate ) k′[5b]), as

indicated by the linear plot of ln{[5b_init]/[5b_init] - [5b]} versus

time (Figure 5), the slope of which equals the phenomenological

rate constant k′, which, under the specified conditions, was found

to be 8.3 × 10^-3M²min^-1. As substrate conversion increased

(to greater than 30%), downward curvature became noticeable

in the plot (Figure 6), which no longer could be well-fitted to

a linear equation. This fact would seem to indicate that product

inhibition and/or catalyst decomposition was playing a role.

Consequently, we chose to perform subsequent kinetics runs in

the region of first-order behavior, where catalyst concentration

is assumed to be constant.

where ν_ois the rate of reaction in the absence of inhibitor, and

ν is the rate in the presence of inhibitor, so that an 18% reduction

in rate gives ꢀ ) 0.18. Additionally a value for K_inof 0.23 M^-1

can be estimated if we assume that the 18% reduction in rate is

attributable to an 18% effective reduction in the concentration

of catalyst through formation of a nonproductive catalyst-

inhibitor complex, such as 25.⁵⁵

(51) Kobayashi has recently reported an interesting lanthanide-based

Diels-Alder catalyst which turns over the sense of induction in the presence

of potentially two-point binding competitive inhibitors: see ref 18b.

(52) A discussion on inhibition by Laidler was found to be informative.

Laidler, K. J. Chemical Kinetics; Harper and Row: New York, 1987.

(53) A kinetic analysis has been done of a similar system: Chow, H.-

F.; Mak, C. C. J. Org. Chem. 1997, 62, 5116-5127.

(54) At the concentration of cyclopentadiene employed in the standard

kinetics run at 0 °C, dimerization was ascertained to be a maximum of 4%

over a 24 h period.

(55) If the concentration of catalyst ) x, and the degree of inhibition )

ꢀ, then the following equation relates K_into x and ꢀ:

[xꢀ]

K_in

)

[x - xꢀ][ln - xꢀ]

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7571

Even more interesting is the observation of double-stereo-

differentiating competitive inhibition in the reaction of 5b with

cyclopentadiene in the presence of the diastereomerically pure

adducts 26 and 27. Runs were performed at concentrations of

catalyst 1a (X ) OTf) (10 mol % in metal, 11 mol % ligand),

substrate (0.49 M), and inhibitor (0.98 M) identical to those

described above. In the presence of inhibitor 26, only slight

inhibition was observed (ꢀ ) 0.07, K_in) 7.7 × 10^-2M^-1),

whereas with inhibitor 27, ꢀ ) 0.16 and K_in) 0.20 M^-1. This

result indicates that the complex sterically matched for the

Diels-Alder reaction (28, eq 24), with chiral substituents on

the same side of the plane defined by the square-planar metal

geometry, is thermodynamically less stable than the correspond-

ing mismatched complex 29 (eq 25).

(Figure 1).¹³These reactions have been postulated to proceed

via bidentate substrate chelation to octahedral and tetrahedral

metal centers, respectively, and to react out of an s-cis

conformation. Our previous experience (vide supra) indicated

that the optimal ligand for a given metal is determined by the

geometry of the metal center. Thus, the best ligand for square-

planar copper has proven to be tert-butyl-bis(oxazoline) 6d,

while phenyl-bis(oxazoline) 6a is optimal for tetrahedral zinc.

We thought it illustrative to undertake a comparative study to

determine whether a similar pattern held for the Fe(III)- and

Mg(II)-based systems. In addition, it was of interest to determine

the relative tolerance of each catalyst system for â-substitution

on the dienophile.

Each permutation of metal and bis(oxazoline) ligand which

had been reported for the reaction was studied (eq 26, Table

8). Most notable is the fact that the tert-butyl ligand 6d, which

was found to be optimal for the Cu(OTf)₂-based catalyst, forms

a poor catalyst upon complexation to Fe(III), Mg(II), or Zn(II).

Similarly, phenyl-substituted ligands 6a and 6e, which were

found to be optimal for the other metal systems, form less

efficient catalysts with Cu(OTf)₂than does tert-butyl ligand 6d.

No clear explanation yet exists for the differences in behavior

induced by these changes in ligand structure. However, at least

one factor may account for the differences between the diphenyl

ligands (6a and 6e) and 6d and their efficiencies when

complexed to metals of differing geometries. It is significant

to note that the s-cis conformation in a tetrahedral catalyst-

substrate complex employing the phenyl-substituted ligands

could be stabilized by an attractive π-stacking interaction which

is absent when ligand 6d is employed. In fact, if the tert-butyl

ligand 6d replaces the phenyl ligand 6a in the Mg(II)- or Zn-

(II)-catalyzed reactions, then the resulting complexes may be

destabilized sterically. Reaction from other conformations could

then account for the erosion of enantioselection with this ligand.

Whether analogous interactions are present in the Fe(III)-

catalyzed reactions is unclear. In these systems, the larger

number of potential geometries for the octahedral system

complicates the analysis. Nevertheless, it is interesting to note

that catalyst efficiency appears to be the result of a carefully

balanced interaction between metal and ligand structure.

To gain more insight into reacting geometries, the Fe(III)-,

Mg(II)-, and Zn(II)-based systems were studied in double-

stereodifferentiating reactions of the type applied to the Cu-

(II)-based system described above. Based on the octahedral

model for the Fe(III) system, and the tetrahedral model for the

Mg(II) and Zn(II) systems, we hoped to observe clear matched

and mismatched catalyst-dienophile complexes, as we had in

the case of the Cu(II) system. However, the results preclude

simple interpretation (eqs 27 and 28, Table 9).⁵⁶In these

systems, low selectivities and slow reaction rates were observed.

Whether the results are due to unanticipated nonbonding

interactions between the substrate benzyl groups and catalysts,

or reacting geometries other than those proposed, remains an

unresolved issue.

Competitive inhibitors that more closely resemble reactants

are also of interest. When standard kinetics runs (conditions

specified above) were performed in the presence of inhibitors

30 and 31, rates of reaction are suppressed, resulting in ꢀ’s of

0.16 and 0.24, respectively. The binding constants K_inof imides

30 and 31 (0.20 and 0.33 M^-1, respectively) may roughly

represent the affinity of an analogous crotonate dienophile for

the catalyst.

Is the two-point binding nature of the products responsible

for the observed inhibition? An interesting way to address this

question is to directly compare one-point and two-point binding

cases. For example, methyl acrylate is known to be a reactive

one-point binding dienophile for the Diels-Alder reaction.

Consequently, the adduct of methyl acrylate and cyclopentadiene

(32) should be a good substrate to test. When Diels-Alder

reactions of imide 5b with cyclopentadiene were doped with

methyl ester 32 (standard conditions, 0.48 M substrate, 0.96 M

32), only slight inhibition (ꢀ ) 0.06) due to the additive was

observed, much less relative to product inhibition itself.

A Comparative Study. Independent studies by Corey and

co-workers have demonstrated that phenyl-substituted bis-

As a final point of comparison, when â-substituted dieno-

philes were tested with each of the four bis(oxazoline)-based

catalyst systems, only the Cu(II)-based system appeared to be

generally applicable (Table 10). The crotonate-derived imide

5b was examined with each of the three catalysts (eq 29) and

afforded good results when the Cu(II)-based catalysts were used

[1b (X ) SbF₆), endo/exo ) 85:15, endo ee ) 99%; 1a (X )

OTf), endo/exo ) 96:4, endo ee ) 97%]. In contrast, the Zn-

+

(oxazoline)-metal complexes [Fe(Ph-box)]I₂(B) and [Mg-

(Ph-box*)]I₂(C) (Ph-box* ) 5,5-dimethyl-4-phenyl-bis-

(oxazoline)) are also catalysts for the Diels-Alder reaction

(56) These reactions, employing Zn(II) complex 8a, were performed by

Mr. Jason S. Tedrow, Harvard University.

7572 J. Am. Chem. Soc., Vol. 121, No. 33, 1999

EVans et al.

Table 8. Diels-Alder Reaction of Imide 5a with Cyclopentadiene and Various Catalysts (Eq 26)

1

^aReactions were run in CH₂Cl₂with 11 mol % ligand and 10 mol % metal. ^bEndo/exo ratios were determined by H NMR and/or chiral GLC

or HPLC. Enantiomer ratios were determined either by chiral GLC or by HPLC. ^cValues in parentheses refer to literature experiments. ^dNot

determined. ^eReaction was run with Zn(OTf)₂as metal source.

Table 9. Double-Stereodifferentiating Reactions of (R)-2 and (S)-2

with Cyclopentadiene Catalyzed by Various Catalysts (Eqs 27 and

28)

Table 10. Diels-Alder Reactions of Imide 5b with

Cyclopentadiene and Bis(oxazoline)-Based Catalysts (Eq 29)^a

time (h)^b

(temp, °C)

endo ee

catalyst

endo/exo^c

(%)^c

[Cu(t-Bu-box)](SbF₆)₂, 1b

[Cu(t-Bu-box)](OTf)₂, 1a

[Zn(Ph-box)](SbF₆)₂, 8a

[Fe(Ph-box)]I₃‚I₂, B

20 (-15)

30 (-15)

8 (25)

40 (-15)

15 (0)

85:15

96:4

86:14

76:24

80:20

99

97

-64

-32

0

[Mg(Ph-box*)]I₂‚I₂, C

^aReactions were run in CH₂Cl₂with 11 mol % ligand and 10 mol

% metal. ^bTime for 100% conversion. ^cEndo/exo ratios were deter-

1

mined by H NMR and/or chiral GLC or HPLC. Enantiomer ratios

were determined either by chiral GLC or by chiral HPLC.

M-I bonds) creates a medium which is too highly oxidizing,

resulting in decomposition of the thione dienophiles.

^aReactions were run in CH₂Cl₂with 11 mol % ligand and 10 mol

% metal. ^bEndo/exo ratios and endo I/endo II ratios were determined

Conclusion. We have developed new chiral Lewis acid

catalysts for the Diels-Alder reaction based on bis(oxazoline)-

copper(II) complexes. The reactivity and stereoselectivity of the

Cu(II) complexes were found to be profoundly influenced by

the counterion. The advantages of the catalyst system presented

here include high diastereo- and enantioselectivities for imide

and thiazolidine-2-thione dienophiles, which have the capacity

to bind to the metal center in a bidentate fashion. The low

sensitivity to hydroxyl-containing additives and the efficacy of

an air-stable hydrated complex provide further experimental

simplicity. In addition, stereochemical probes, solid-state data,

computational studies, and kinetics experiments were performed

to document the intermediacy of a square-planar catalyst-

substrate complex of the s-cis-chelated dienophile during the

C-C bond forming event.

1

by H NMR and/or chiral GLC or HPLC.

based catalyst provided only 64% ee (endo/exo ) 84:16) at room

temperature. At -15 °C, the Fe(III)-based system was also

significantly less selective (endo/exo ) 76:24, endo ee )

-32%), and the Mg(II)-based system afforded racemic adducts

(endo/exo ) 4:1, endo ee ) 0%). In addition, when we

attempted to compare the three catalyst systems with respect to

their efficiencies by employing the thiazolidine-2-thiones 13a-

c, we were disappointed to find that both the Fe(III)- and Mg-

(II)-based systems were incompatible with these substrates.

Presumably, the generation of the active catalysts by employing

an exogenous equivalent of I₂(to facilitate labilization of the

Bis(oxazoline)Cu(II) Complexes as Lewis Acid Catalysts

J. Am. Chem. Soc., Vol. 121, No. 33, 1999 7573

Acknowledgment. We would like to thank Professor E. N.

Jacobsen for a helpful discussion regarding the kinetics study,

Dr. Benjamin Anderson for useful comments, and Dr. Jeffrey

Johnson for assistance in the preparation of this manuscript.

Support was provided by the National Science Foundation.

Supporting Information Available: Complete experimental

procedures, product characterization, and stereochemical proofs

(PDF). This material is available free of charge via the Internet

at http://pubs.acs.org.

JA991190K

Article Doi

Chiral bis(oxazoline)copper(II) complexes as lewis acid catalysts for the enantioselective Diels-Alder reaction

DOI: 10.1021/ja991190k

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja991190k

Full text of DOI:10.1021/ja991190k

Products guided by the article

R&D Labs maybe for 109299-98-1

Relevant to this article

Hot Product