Lewis acid catalyzed allylboration: Discovery, optimization, and application to the formation of stereogenic quaternary carbon centers

Lew is Acid Ca ta lyzed Allylbor a tion : Discover y, Op tim iza tion ,

a n d Ap p lica tion to th e F or m a tion of Ster eogen ic Qu a ter n a r y

Ca r bon Cen ter s

J ason W. J . Kennedy and Dennis G. Hall*

Department of Chemistry, Gunning-Lemieux Chemistry Centre, University of Alberta,

Edmonton, Alberta T6G 2G2, Canada

dennis.hall@ualberta.ca

Received February 9, 2004

A full account of the development of the first catalytic manifold for the additions of allylboronates

to aldehydes is described. The thermal additions (both diastereospecific and enantioselective) of

2-carboxyester 3,3-disubstituted allylboronates 1 to both aromatic and aliphatic aldehydes give

biologically and synthetically important exo-methylene butyrolactones 2 containing a â-quaternary

carbon center. Although the thermal reaction requires 14 d at room temperature to reach completion,

the presence of certain metal salts allows for a 12-16 h reaction while preserving the dia-

stereospecificity observed in the uncatalyzed process. Preliminary mechanistic studies on the origin

of the catalytic effect are described as well as stereoselective transformations of lactones 2 into

cyclic and acyclic stereotriads with potential usefulness as synthetic intermediates.

In tr od u ction

The addition of 3,3-disubstituted allylboronates to

aldehydes is a particularly effective route to stereodefined

quaternary carbon centers due to the high diastereo-

selectivities usually observed in these reactions. This

approach is made more attractive by the fact that the

stereochemical outcome of these additions is generally

and reliably predicted from simple Zimmerman-Traxler

models. Hoffmann and Schlapbach were among the first

to show that 3,3-disubstituted allylboronates could be

used to generate stereodefined quaternary carbon

centers.^11aAlthough these allylboronates reacted with

aldehydes in the expected manner, the reactions were

notably slower than those involving less substituted

allylboronates, requiring 5-8 days at room temperature

to reach completion. The authors also noted that the

diastereomeric purity of the homoallylic alcohol products

was lower than the purity of the initial allylboronates.

A comparable stereochemical leakage was reported by

Suzuki and co-workers during similar studies on the

reactions of 3,3-disubstituted allylboronates.^11bThis lower

level of stereocontrol, accompanied by the long prepara-

tion of the required 3,3-disubstituted allylboronates, have

hampered the wide spread use of this method for the

preparation of quaternary carbon centers.

Despite the major advances made in synthetic organic

chemistry over the past several decades, the construction

of quaternary carbon centers (i.e., carbons bearing four

carbon substituents) remains a significant challenge.^1-9

Part of this synthetic challenge derives from the con-

gested nature of these structural units, which makes

unfavorable steric interactions unavoidable along the

reaction pathway leading to these centres. The prepara-

tion of these sterically demanding centers is important

not only because of the intellectual challenge that they

present to modern synthetic chemists but also because

they are present in many interesting natural products

and synthetic drugs (e.g., asteltoxin and LY426965

below).¹⁰Since quaternary carbons in synthetic targets

often bear four nonequivalent substituents, the challenge

is further amplified by the need to prepare these centers

in a stereocontrolled fashion.

While other allylation reagents have been used to

prepare stereogenic quaternary carbon centers (e.g.,

allylstannanes^12aand allylzincs^12b), to date the bench-

Published on Web 05/21/2004

4412

J . Org. Chem. 2004, 69, 4412-4428

Lewis Acid Catalyzed Allylboration

mark method in this field belongs to Denmark and co-

workers, who can prepare quaternary carbon centers with

high levels of diastereo- and enantiocontrol using a chiral

bisphosphoramide-catalyzed allylsilation system.¹³How-

ever, a significant limitation to this chemistry is that it

is currently only applicable to aromatic aldehydes be-

cause aliphatic aldehydes fail to undergo the desired

allylation.

As reported in a previous communication,^14a2-carboxy-

ester-3,3-disubstituted allylboronates 1 undergo a di-

astereospecific allylation with aldehydes to yield syn-

thetically and biologically useful R-methylene butyro-

lactones 2 which bear a â-stereogenic quaternary carbon

center (eq 1). In retrospect, this preparation presented

us with two significant challenges. The first was the

preparation of the allylboronates 1; previously published

routes to allylboronates either offered poor prospects of

diastereocontrol and substrate generality¹⁵or were long

and cumbersome.^11,16We solved this problem by develop-

F IGURE 1. Type I and II allylation mechanisms.¹⁸

ing a general and efficient, two-step, one-pot preparation

of these important starting materials via the carbocu-

pration of alkynoate esters.

The second challenge encountered in this methodology

was the low reactivity of the 2-carboxyester allyl-

boronates 1. Conjugation of the allyl unit to the ester

lowers the nucleophilicity of these allylboronates, and

consequently, aldehyde additions with 1 were quite

sluggish. This reactivity problem prompted us to explore

the possibility of using an external Lewis acid to catalyze

the allylboration.¹⁷Allylboronates are self-activating

Type I reagents, where the allylation reaction is promoted

by internal coordination of the aldehyde to the boron

atom (Figure 1).¹⁸Because of this internal activation,

there would appear to be no need for an external

promoter. Furthermore, an external Lewis acid might

compete with the boron atom for the aldehyde, leading

to a switch from the highly diastereoselective, cyclic, Type

I mechanism to the less selective, open-chain Type II

mechanism. However, this concern turned out to be

unfounded, and in the end we were able to develop the

first successful catalytic system that allows for the

diastereospecific allylation of both aliphatic and aromatic

aldehydes with allylboronates under mild conditions in

reasonable time frames.^14b

Here we report the full account of the discovery and

optimization of the Lewis acid catalyzed allylboration,

focusing in particular on the use of this methodology for

the challenging preparation of quaternary carbon centers.

Efforts toward a novel attempt to control the absolute

stereochemistry of the allylboration using a single, car-

boxyester-based chiral auxiliary will be described. The

results of a scope and limitations study on the Lewis acid

catalyzed allylboration reaction are presented, along with

a new catalytic protocol which greatly improves the

efficiency of the addition reaction with aliphatic alde-

hydes. Finally, we describe the extension of this meth-

odology to the preparation of highly substituted butyro-

lactones with three contiguous stereogenic carbons.

(1) (a) Martin, S. F. Tetrahedron 1980, 36, 419-460. (b) Fuji, K.

Chem. Rev. 1993, 93, 2037-2066. (c) Corey, E. J .; Guzman-Perez, A.

Angew. Chem., Int. Ed. 1998, 37, 388-401. (d) Christoffers, J .; Mann,

A. Angew. Chem., Int. Ed. 2001, 40, 4591-4597. (e) Denissova, I.;

Barriault, L. Tetrahedron 2003, 59, 10105-10146.

(2) Romo, D.; Meyers, A. I. Tetrahedron 1991, 47, 9503-9569.

(3) (a) Enders, D.; Zamponi, A.; Raabe, G.; Runsink, J . Synthesis

1993, 725-728. (b) Enders, D.; Zamponi, A.; Scha¨fer, T.; Nu¨bling, C.;

Eichenauer, H.; Demir, A. S.; Raabe, G. Chem. Ber. 1994, 127, 1707-

1721.

(4) (a) Manthorpe, J . M.; Gleason, J . L. Angew. Chem., Int. Ed. 2002,

41, 2338-2341. (b) Manthorpe, J . M.; Gleason, J . L J . Am. Chem. Soc.

2001, 123, 2091-2092.

(5) Spino, C.; Beaulieu, C. Angew. Chem., Int. Ed. 2000, 39, 1930-

1932.

(6) Luchaco-Cullis, C. A.; Mizutani, H.; Murphy, K. E.; Hoveyda,

A. H. Angew. Chem., Int. Ed. 2001, 40, 1456-1460.

(7) (a) Mulzer, J .; Du¨rner, G.; Trauner, D. Angew. Chem., Int. Ed.

Engl. 1996, 35, 2830-2832. (b) Christoffers, J .; Ro¨âler, U.; Werner, T.

Eur. J . Org. Chem. 2000, 701-705. (c) Christoffers, J .; Mann, A. Chem.

Eur. J . 2001, 7, 1014-1027. (d) Enders, D.; Teschner, P.; Raabe, G.;

Runsink, J . Eur. J . Org. Chem. 2001, 4463-4477. (e) Christoffers, J .;

Baro, A. Angew. Chem., Int. Ed. 2003, 42, 1688-1690.

(8) Fukuda, Y.-i.; Sasaki, H.; Shindo, M.; Shishido, K. Tetrahedron

Lett. 2002, 43, 2047-2049.

(9) Enders, D.; Knopp, M.; Runsink, J .; Raabe, G. Liebigs Ann. 1996,

1095-1116.

(10) For some examples, see: (a) Schreiber, S. L.; Satake, K. J . Am.

Chem. Soc. 1984, 106, 4186-4188. (b) Denmark, S. E.; Fu, J . Org. Lett.

2002, 4, 1951-1953.

(11) (a) Hoffmann, R. W.; Schlapbach, A. Liebigs. Ann. Chem. 1990,

1243-1248. (b) Sato, M.; Yamamoto, Y.; Hara, S.; Suzuki, A. Tetra-

hedron Lett. 1993, 34, 7071-7074.

(12) (a) Nishigaichi, Y.; Takuwa, A. Tetrahedron Lett. 1999, 40, 109-

112. (b) Skulte, G.; Amsallem, D.; Shabli, A.; Varghese, J . P.; Marek,

I. J . Am. Chem. Soc. 2003, 125, 11776-11777.

(13) (a) Denmark, S. E.; Fu, J . P. Chem. Commun. 2003, 167-170.

(b) For a specific example on the preparation of a quaternary carbon

centre, see: Denmark, S. E.; Fu, J . Org. Lett. 2002, 4, 1951-1953.

(14) (a) Kennedy, J . W. J .; Hall, D. G. J . Am. Chem. Soc. 2002, 124,

898-899. (b) Kennedy, J . W. J .; Hall, D. G. J . Am. Chem. Soc. 2002,

124, 11586-11587.

Resu lts a n d Discu ssion

1.1. P r ep a r a tion of Tetr a su bstitu ted Allylbor -

on a tes a n d Op tim iza tion of E/Z Selectivity. Inspired

by the work of Hoffmann and Schlapbach,^11awe decided

to explore a route to tetrasubstituted allylboronates

starting from the carbocupration of alkynoate esters

(Scheme 1). Here, treatment of an alkynoate ester (3)

with a dialkylcuprate would yield alkenylcopper inter-

(15) Roush, W. R. In Stereoselective Synthesis: Methods of Organic

Chemistry (Houben-Weyl), 4th ed.; Helmchen, G., Hoffmann, R. W.,

Mulzer, J ., Schaumann, E., Eds.; Georg Thieme Verlag: Stuttgart,

1995; Vol. E21b, pp 1410-1486.

(16) For some alternate preparations of allylboronates, see: (a)

Falck, J . R.; Bondlela, M.; Ye, J . H.; Cho, S. D. Tetrahedron Lett. 1999,

40, 5647-5650. (b) Yang, F.-Y.; Wu, M.-Y.; Cheng, C.-H. J . Am. Chem.

Soc. 2000, 122, 7122-7123. (c) Yamamoto, Y.; Kurihara, K.; Yamada,

A.; Takahashi, M.; Takahashi, Y.; Miyaura, N. Tetrahedron 2003, 59,

537-542. (d) Ohmura, T.; Yamamoto, Y.; Miyaura, N. Organometallics

1999, 18, 413-416. (e) Goldberg, S. D.; Grubbs, R. H. Angew. Chem.,

Int. Ed. 2002, 41, 807-810. (f) Toure´, B.; Hoveyda, H. R.; Tailor, J .;

Ulaczyk-Lesanko, A.; Hall, D. G. Chem. Eur. J . 2003, 9, 466-474. (g)

Gao, X.; Hall, D. G. J . Am. Chem. Soc. 2003, 125, 9308-9309. (h)

Pietruszka; J . Scho¨ne, N. Angew. Chem., Int. Ed. 2003, 42, 5638-5641.

(i) Ramachandran, P. V.; Pratihar, D.; Biswas, D.; Srivastava, A.;

Reddy, M. V. R. Org. Lett. 2004, 6, 481-484.

(17) For

a recent review on the activation of allylsilanes and

allylboronates, see: Kennedy, J . W. J .; Hall, D. G. Angew. Chem., Int.

Ed. 2003, 42, 4732-4739.

(18) For a discussion of the different types of allylating reagents,

see: (a) Denmark, S. E.; Weber, E. J . Helv. Chim. Acta 1983, 66, 1655-

1660. (b) Denmark, S. E.; Almstead, N. G. In Modern Carbonyl

Chemistry; Otera, J ., Ed.; Wiley-VCH: Weinheim, 2001; pp 299-401.

J . Org. Chem, Vol. 69, No. 13, 2004 4413

Kennedy and Hall

TABLE 1. Effect of Differ en t Ad d itives on th e

Ca r bocu p r a tion -Alk yla tion Sequ en ce^a

SCHEME 1

mediate 4, which would subsequently be trapped by a

halomethaneboronate (5) to give the tetrasubstituted

allylboronate 1. This route appeared promising because

the carbocupration of alkynoate esters is known to

proceed stereospecifically to give one isomer of the

alkenylcopper species 4.¹⁹Another attractive feature of

this strategy is that a large number of alkynoate esters

and cuprates are available, which allows for a wide

variety of R¹and R²groups in the products 1. However,

the alkenylcopper intermediate 4 is notoriously unreac-

tive and is typically only trapped by very active electro-

philes.²⁰At the outset of these studies, we wondered if

electrophile 5 would be sufficiently reactive to trap this

intermediate or, if not, whether conditions could be found

which would allow for effective trapping. Also, the

1-alkoxycarbonylalkenylcopper intermediate 4 is configu-

rationally unstable at temperatures above -30 °C,²¹and

any alkylation conditions would have to contend with this

potential pitfall.

entry

additive

equiv

yield (%)

E/Z selectivity

1

2

3

4

5

6

7

HMPA

DMPU

DMEU

TMEDA

9

3

1

68

(100)

(69)

34

1:17

1:12

1:4

1:>20

1:10

1:12

40^b

9

7

54

0

a

Reactions done on 1 mmol scale in 5 mL of THF. E/Z ratio

determined by ¹H NMR of the crude reaction mixture. Yields in

b

parentheses are from unpurified products. DMPU used as a 1:1

cosolvent with THF.

used in much larger amounts in order to achieve the

same results (entry 4).

Recently, a more thorough study of the role of additives

in this carbocupration-alkylation sequence was per-

formed in our research group.²²This study showed that

HMPA plays a dual role in the alkylation step: it

stabilizes the alkenylcopper intermediate by sequester-

ing lithium cations from solution, and it enhances the

alkylation process itself. This study also showed that

excellent yields and diastereoselectivities are obtained

with as little as 2 equiv of HMPA provided that 1 equiv

of 12-crown-4 is included in the electrophilic quench.

Finally, we demonstrated that the chloromethane

boronate 5b is not as effective as the iodoelectrophile 5a

in this reaction (eq 3).²³

Initial experiments following the procedure of Deslong-

champs and co-workers yielded promising, if somewhat

modest, results.^20bMethyl propionate 3a and ethyl 2-bu-

tynoate 3b were both successfully converted into their

respective allylboronates 1a and 1b (eq 2). Unfortunately,

the yields for both compounds were low, and the E/Z

selectivity for 1a was poor.

A breakthrough came when we discovered that adding

small amounts of hexamethylphosphoramide (HMPA)

prior to the electrophilic quench allowed for the formation

of tetrasubstituted allylboronates 1 in good yield and

excellent diastereoselectivity (Table 1). In this model

study, the data show that at least 9 equiv of HMPA is

required to achieve a highly selective reaction (entry 1)

and that selectivity erodes as the amount of HMPA

decreases (entries 2 and 3). Although attempts were

made to find a replacement for the carcinogenic HMPA,

no other additive tried to date rivals HMPA in its

efficiency at providing good yields of highly stereopure

products. Although DMPU comes the closest, it must be

We were pleased to find that a large number of

tetrasubstituted allylboronates could be prepared using

this convenient, one-pot, two-step procedure (Table 2).

Initially, the allylboronates 1 were assumed to be too

sensitive to purify, and consequently, some of the yields

below are compiled from crude products. However, these

boronates eventually proved to be quite robust, and they

could be readily purified by flash chromatography on

silica gel.

As can be seen from Table 2, cuprates derived from

either alkyllithium or Grignard reagents may be used.

Branched organometallic reagents are also effective in

the reaction, although the stereoselectivity of the product

(19) (a) Corey, E. J .; Katzenellenbogen, J . A. J . Am. Chem. Soc. 1969,

91, 1851-1852. (b) Siddall, J . B.; Biskup, M.; Fried, J . H. J . Am. Chem.

Soc. 1969, 91, 1853-1854.

(20) (a) Marino, J . P.; Linderman, R. J . J . Org. Chem. 1983, 48,

4621-4628. (b) Hall, D. G.; Chapdelaine, D.; Pre´ville, P.; Deslong-

champs, P. Synlett 1994, 660-662.

(22) Zhu, N.; Hall, D. G. J . Org. Chem. 2003, 68, 6066-6069.

(23) Chloromethaneboronate 5b can be prepared according to: Wuts,

P. G. M.; Thompson, P. A. J . Organomet. Chem. 1982, 234, 137-141.

(21) Nilsson, K.; Andersson, T.; Ullenius, C.; Gerold, A.; Krause, N.

Chem. Eur. J . 1998, 4, 2051-2058.

4414 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

TABLE 2. Yield s a n d Dia ster eom er ic Ra tios of P r ep a r ed Allylbor on a tes 1^a

entry

alkynoate

boronate

R¹

R²M

yield (%)

E/Z

1^b

2

3

4

5

6

7

8

3a

3b

3c

3d

3b

3c

3d

3b

3e

3f

1a

1b

1c

1d

1e

1f

1g

1h

1i

1j

1k

1l

1m

1n

1o

1p

H

Me

Et

Bu

Me

Et

MeLi

BuLi

i-BuMgCl

s-BuLi

t-BuLi

60

(99)

68

(99)

78

45

58

72

60

24

60

(92)

28

>20:1

1:17

1:>20

>20:1

2:1

>20:1

19:1

9^b

10

11

12

13

14

15

16

BuLi

Bu

Me

octylMgCl

allylMgCl

MeLi

TMS(CH₂)₂OCH₂Li

MeLi

TBDPSOCH₂

TBDPSO(CH₂)₂

Me

Ph

1:>20

>20:1

1:>20

3b

3g

3h

38

0

TMS

MeLi

a

R ) Me, Et. Yields in parentheses are for crude products, others are of compounds purified by flash chromatography. TBDPS )

b

tert-butyldiphenylsilyl. Allylboronate prepared from 2 equiv of 5a and a room-temperature quench; see below for details.

decreases for sterically demanding reagents (cf. entries

5-8). Unsaturation may be present in the cuprate as long

as it is not directly next to the copper. Thus, allyl groups

are effectively transferred (entry 11), but vinyl- and

phenylcuprates failed to provide the desired products. In

these latter cases, the major product is the protonated

olefin, suggesting that the problem originates in the

electrophilic quench. A single, low-yielding example was

obtained with a cuprate bearing a protected alcohol (entry

14).

Similarly, there are few limitations on the kind of

alkynoate that may be used in this reaction. Alkynoates

that are not commercially available are readily prepared

by the reaction of terminal alkynes with methyl chloro-

formate.²⁴So far, alkynoates where the R¹group has been

a primary, aliphatic chain have successfully been used,

as well as alkynoates bearing protected hydroxyl groups

(entries 12 and 13). There were, however, two substrates

that did not perform well in this reaction sequence.

Methyl 3-phenylpropiolate 3g gave only a low yield of

boronate 1o along with significant amounts of unreacted

alkynoate (entry 15), and the TMS-protected 3h was

recovered unchanged after reaction (entry 16). Perform-

ing the 1,4-addition of these alkynoates at a higher

temperature (-40 °C) did not improve the outcome.

Overall, these results show that this procedure is able

to produce a relatively large scope of highly functional-

ized, tetrasubstituted allylboronates quickly and ef-

ficiently.

E-substituent, presumably due to anisotropic deshielding

from the ester. This deshielding of the Z-substituent has

1

been noted before.²⁰In addition, H NOE experiments

on boronate 1c were consistent with the proposed struc-

ture.²⁵

1.2. Op tim iza tion of Rea gen t Stoich iom etr y in th e

P r ep a r a tion of Tetr a su bstitu ted Allylbor on a tes (1).

One limitation to the sequence described above was the

use of 3 equiv of electrophile 5a in the alkylation step.

Although the preparation of 5a is not dificult,²⁶a more

effective and less wasteful use of this electrophile was

desirable. Yields and selectivities using only 2 equiv of

5a were comparable to those obtained when 3 equiv was

employed as long as the electrophilic quench was per-

formed at room temperature (Table 3, compare entry 2

with entry 4).

Interestingly, no allylboronate was formed at all if only

1 equiv of 5a was used (entry 5). This result suggests

that the first equivalent of iodomethaneboronate 5a is

consumed by preferential transfer of the methyl group

over the alkenyl group from the mixed cuprate 4a to the

electrophile 5a (Scheme 2) to give the putative ethyl-

boronate 6. Selectivity in the transfer of ligands from

unsymmetrical cuprates has long been an issue in

organocopper chemistry, and saturated ligands are gen-

erally transferred faster in substitution reactions than

unsaturated groups.²⁷Although this problem might be

circumvented by using a cuprate bearing a nontransfer-

The stereochemistry of the products was reliably

(25) See the Supporting Information for further details.

(26) Of the numerous preparations of iodomethaneboronate 5a

available, the one that we found to be the most convenient on large

scale was: Phillion, D. P.; Neubauer, R.; Andrew, S. S. J . Org. Chem.

1986, 51, 1610-1612. For alternative preparations, see refs 44 and

23.

1

determined by analysis of the H NMR spectrum. In all

1

cases, the H NMR signal of the Z-substituent (which is

cis to the ester group) was found downfield from the

(24) Piers, E.; Wong, T.; Ellis, K. A. Can. J . Chem. 1992, 70, 2058-

2064.

(27) Lipshutz, B. H.; Wilhelm, R. S.; Kozlowski, J . A.; Parker, D. A.

J . Org. Chem. 1984, 49, 3928-3938.

J . Org. Chem, Vol. 69, No. 13, 2004 4415

Kennedy and Hall

TABLE 3. Op tim iza tion of th e P r ep a r a tion of 1c w ith

Resp ect to Equ iva len ts of 5a a n d Tem p er a tu r e

The large rate differences seen between reactions with

2-carboxyester allylboronates and 3,3-disubstituted ana-

logues that lack this ester substituent suggest that steric

arguments alone cannot explain the low reactivity of 1.

This slow reaction is likely due to a reduction of the

nucleophilicity of the γ-carbon in boronates 1 by conjuga-

tion to the carboxyester. In support of this theory, an

unusually slow reaction is also observed with 3-alkoxy-

allylboronates,³¹where the alkoxy substituent can simi-

larly reduce the nucleophilicity at this position by

inductive effects. However, reaction times with allylbo-

ronates 1 are dramatically reduced under the new

catalytic manifold (see below).

Previous reports of allylations with 3,3-disubstituted

allylboronates that lack the 2-carboxyester group describe

that these reactions proceed with high but not complete

diastereoselectivity.¹¹In contrast to these reports, addi-

tionswith our allylboronates1aregenerallydiastereospecifics

the Z-group in boronate 1 (R², which derives from the

cuprate) is almost always syn in lactone 2 to the group

from the aldehyde (R³).³²Furthermore, the stereochem-

ical purity of lactone 2 mirrors that of boronate 1. Thus,

if the initial boronate is a 20:1 mixture of E/Z isomers,

then the lactone produced is correspondingly a 20:1

diastereomeric mixture. This effect was elegantly dem-

onstrated by the preparation of the epimeric lactones 2i

and 2j (Scheme 3). The only time that these reactions

were not stereospecific is when aliphatic aldehydes are

allylated at high temperatures. When reactions with

these substrates are performed in refluxing toluene, the

product lactones do show some erosion of stereoselectivity

(see below). This erosion is not observed when the

reactions are done at temperatures at or below 80 °C,

nor are they observed with aromatic substrates at any

temperature.

The diastereomer observed in these reactions is the

same as that predicted from simple Zimmerman-Traxler

models (Figure 2) where the R³group from the aldehyde

occupies the pseudoequatorial position (transition state

A). Reactions with 2-carboxyester allylboronates do not

exhibit any stereochemical leakage, in contrast to

allylations with 3,3-disubstituted allylboronates that lack

the 2-carboxyester group (see above).¹¹The reason for the

improved selectivity is likely that in the competing

transition state B, the formyl R³group and the 2-car-

boxyester group would experience an unfavorable 1,3-

diaxial interaction in addition to the usual 1,3-interaction

between the R³group and the axial boronate substituent.

This additional steric interaction further favors transition

state A over transition state B.

Finally, there remains to explain the erosion of stereo-

chemical purity observed in the reactions of aliphatic

aldehydes in refluxing toluene. One possible explanation

is that the boronate undergoes reversible borotropic

rearrangement at these temperatures,¹⁵thus leading to

“isomerized” products after the allylboration. However,

if that were the case, then aromatic aldehydes should also

equiv

of 5a

quench

conditions

yield

(%)

E/Z

selectivity

entry

1

2

3

4

5

4

3

2

1

0 °C, 3 h

0 °C, 2 h

0 °C, 4 h

rt, 2 h

72

66

27

53

0

1:>20

0 °C, 4 h

SCHEME 2

able group,²⁸electrophile 5a is too readily available to

justify an extensive search for a suitable dummy ligand.

2.1. Qu a ter n a r y Ca r bon Cen ter s fr om Dia ster eo-

selective Allylbor a tion s. We were pleased to find that

the tetrasubstituted allylboronates 1 underwent a smooth,

if somewhat slow, reaction with aldehydes (Table 4). In

contrast to previously reported allylations with 3,3-

unsubstituted 2-carboxyester allylboronates,²⁹the prod-

uct from allylations with 1 was almost always lactone 2,

and the intermediate homoallylic alcohol 7 was only

observed in the reaction of monosubstituted boronate 1a

(entry 1). This facile lactonization is likely a manifesta-

tion of the gem-dialkyl effect.³⁰

These reactions are operationally extremely simples

the aldehyde and allylboronate are stirred together in

toluene, dichloromethane, or even neat. Most aldehydes,

both aromatic and aliphatic, are effective substrates in

the reaction, and so far the only exceptions found are very

hindered aldehydes (e.g., cyclohexanecarboxaldehyde,

mesitaldehyde) and aldehydes with free amino groups (4-

dimethylaminobenzaldehyde). The only drawback is that

the reactions are quite sluggish. Similar to the results

of Villie´ras and co-workers,²⁹reaction times are on the

order of 2 weeks at room temperature but, in contrast to

these published results, only 24 h in hot toluene.

(28) Lipshutz, B. H.; Kozlowski, J . A.; Parker, D. A.; Nguyen, S. L.;

McCarthy, K. E. J . Organomet. Chem. 1985, 285, 437-447.

(29) (a) Nyzam, V.; Belaud, C.; Villie´ras, J . Tetrahedron Lett. 1993,

34, 6899-6902. (b) Nyzam, V.; Belaud, C.; Zammattio, F.; Villie´ras, J .

Bull. Soc. Chim. Fr. 1997, 134, 583-592.

(30) For a discussion of this effect, see: J ung, M. E.; Gervay, J . J .

Am. Chem. Soc. 1991, 113, 224-232.

(31) (a) Hoffmann, R. W.; Kemper, B.; Metternich, R.; Lehmeier, T.

Liebigs. Ann. Chem. 1985, 2246-2260. (b) Hoffmann, R. W.; Metter-

nich, R. Liebigs. Ann. Chem. 1985, 2390-2402.

(32) The stereochemistry of the lactones 2 was unambiguously

determined by ¹H NOE experiments on the diastereomeric lactones

2i and 2j as well as by X-ray crystallography of lactone 2n . See the

Supporting Information for details.

4416 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

TABLE 4. Dia ster eosp ecific Allylbor a tion s w ith 3,3-Disu bstitu ted Allylbor on a tes 1^a

entry

boronate

R¹

R²

R³

lactone

conditions^b

yield (%)

1

2

3

4

5

6

7

8

9

10

11

12

13

14

1a

1b

1c

1d

1e

1f

H

Me

Bu

Ph

2a

2b

2c

2d

2e

2f

2g

2h

2i

2j

2k

2l

2m

2n

B

C

E

60

78

92

40

68

60

47

81

76

67

50

48

75

26

Me

Et

TBDPSO(CH₂)₂

C₉H₁₉

B

A

C

B

D

C

B

Ph

p-MeO-C₆H₄

p-NO₂-C₆H₄

p-MeO-C₆H₄

p-NO₂-C₆H₄

Et

Bu

Me

i-Bu

allyl

Me

1k

1l

1o

TBDPSOCH₂

Ph

a

b

All butyrolactones 2 were isolated in >20:1 dr. TBDPS ) tert-butyldiphenylsilyl. Conditions: (A) toluene, rt, >12 d; (B) toluene,

60-80 °C, 16-120 h; (C) toluene, 110 °C, 16-24 h; (D) CH₂Cl₂, 40 °C, 48 h; (E) neat, rt, >12 d.

SCHEME 3

transition states become energetically attainable at these

higher temperatures. This explanation was the one

offered by Hoffmann and Schlapbach to rationalize the

erosion of stereochemistry seen with their 3,3-disubsti-

tuted allylboronates.^11aWhile transition state B in Figure

2 is the most likely candidate, we cannot exclude the

possibility that boatlike transition states which lead to

the other diastereomer of lactone 2 may also be operative.

2.2. Qu a ter n a r y Ca r bon Cen ter s fr om En a n tio-

selective Allylbor ation s: Sin gle Au xiliar y Appr oach .

The possibility of extending this methodology to enantio-

selective additions was next examined. Currently there

is no widely applicable route to the enantioselective

preparation of quaternary carbon centers using allyl-

boration chemistry.³³Inspection of the structure of bor-

onate 1 reveals that there are two positions where a

chiral auxiliary could be installedseither on the carbox-

yester or on the boronate. Since the Villie´ras group had

found that most known, boron-based, chiral diol directors

were ineffective with 2-carboxyester allylboronates,^16i,29,34

we decided to explore the effectiveness of a carboxyester-

based auxiliary.

show some levels of erosion. Furthermore, recovered

boronates from these reactions do not show evidence of

isomerization. A more likely possibility is that other

The requisite chiral 2-carboxyester allylboronates 11

were prepared by the DCC mediated esterification³⁵of

2-butynoic acid 8 with various chiral alcohols (9)³⁶

followed by carbocupration and trapping with 5a as

described above (Table 5).

Using aldehyde 12 as an aliphatic, nonvolatile, model

substrate,³⁷enantioselective allylborations with chiral

(33) For some pioneering studies on the enantioselective preparation

of quaternary carbon centres from 3,3-disubstituted allylboronates,

see: (a) Hoffmann, R. W.; Schlapbach, A. Liebigs Ann. Chem. 1991,

1203-1206. (b) Yamamoto, Y.; Hara, S.; Suzuki, A. Synlett 1996, 883-

884.

(34) Chataigner, I.; Lebreton, J .; Zammattio, F.; Villie´ras, J . Tet-

rahedron Lett. 1997, 38, 3719-3722.

(35) Fonquerna, S.; Moyano, A.; Perica`s, M. A.; Riera, A. Tetrahe-

dron 1995, 51, 4239-4254.

(36) Alcohols 9f-h were prepared according to: Yang, D.; Xu, M.;

Bian, M.-Y. Org. Lett. 2001, 3, 111-114. See also: (a) Ort, O. Org.

Synth. 1987, 65, 203-213. (b) Potin, D.; Dumas, F. Synth. Commun.

1990, 20, 2805-2813.

F IGURE 2. Proposed mechanistic pathway for diastereospe-

cific allylborations with 2-carboxyester allylboronates 1.

J . Org. Chem, Vol. 69, No. 13, 2004 4417

Kennedy and Hall

TABLE 5. P r ep a r a tion of Ch ir a l Alk yn oa tes 10 a n d

Allylbor on a tes 11

TABLE 6. En a n tioselective Allyla tion s w ith Ch ir a l

Allylbor on a tes 11a -i^a

entry

allylboronate

conditions

yield (%)

ee (%)

1

2

3

4

5

6

7

8

9

11a

11b

11c

11d

11e

11f

11g

11h

11i

neat, 8 d

89

64

70

78

51

22

(-6)^b

10

7

80

82

75

56

66

neat, 14 d

toluene, 26 d^c

neat, 8 d

toluene, 36 d^c

toluene, 14 d

CH₂Cl₂, 14 d

toluene, 14 d

(95)^d

(86)^d

74

24

6

10

62

chiral

yield of

10 (%)

crude yield

entry alcohol alkynoate

boronate

of 11 (%)^a

a

Yields calculated after silica gel chromatography. Selectivities

were determined by Chiral HPLC. See the Experimental Section

for details. The absolute configuration of 2c is inferred from other

results; see below for details. TBDPS ) tert-butyldiphenylsilyl.

^bMajor isomer is opposite to that in the other entries. ^cExcessively

long reaction times due to the performance of the reaction over a

Christmas vacation, and not due to a reactivity problem with

boronates 11c and 11e. Product contaminated with auxiliary.

While this contamination gives an artificially high yield, it did

not affect the chiral HPLC analysis.

1

2

3

4

5

6

7

8

9

9a

9b

9c

9d

9e

9f

9g

9h

9i

10a

10b

10c

10d

10e

10f

10g

10h

10i

89

80

79

78

89

82

65

35

43

11a

11b

11c

11d

11e

11f

11g

11h

11i

>99

98

94

44^b

>99

d

a

Crude yields recorded due to the presumed instability of the

b

products. Product purified by silica gel chromatography.

next section), the (S)-configuration has been assigned to

the major enantiomer of lactone 2c produced in these

reactions.

allylboronates 11 were performed at room temperature

over 10-14 days and the enantiomeric purity of the

lactone products 2c was determined by chiral HPLC

(Table 6). Lactonization of the homoallylic alcohol was

not spontaneous in reactions with the very large aryl-

menthol auxiliaries (entries 5-10), and the cyclization

had to be promoted by treatment of the crude reaction

mixture with mild acid.

Gratifyingly, we found that a carboxyester-based chiral

auxiliary could effectively direct the stereochemical

course of the reaction (Table 6). Although the simple

chiral alcohols 9a -d were not effective auxiliaries in this

reaction (entries 1-4), the leap in enantioselectivity

observed with the 8-arylmenthol auxiliaries 9e-i is

remarkable (entries 5-10). This increase is especially

impressive given the poor performance of menthol 9d (the

parent compound, entry 4). A small solvent effect was

also observed, where selectivities were higher when the

reaction was performed in toluene rather than in dichlo-

romethane (compare entries 6 and 7).

In view of the effectiveness of the 8-phenymenthol ester

11e in the reaction (entry 5), it seemed only a matter of

modifying the aromatic ring to find the auxiliary that

would provide truly effective levels of stereoinduction.

Unfortunately, enantioselectivities greater than 82%

were not achieved, despite the use of a large number of

different arylmenthols, and there currently seems to be

no advantage to using auxiliaries other than the com-

mercially available phenyl derivative 9e.

The power of 8-arylmenthol esters in stereoselective

cyclizations,^36,38Michael additions,³⁹and allylsilations⁴⁰

is well documented, and their effectiveness is generally

attributed to the ability of the aromatic ring to shield

one face of the reacting olefin.⁴¹In the case of these

allylborations, the phenyl ring presumably shields one

of the faces of the allyl group. Indeed, an X-ray crystal

structure of 11f shows that the â-naphthyl group ef-

fectively covers the Re-face of the allylboronate (as

determined from C2, Figure 3). This X-ray structure also

reveals that there is no coordination between the boron

1

and the carboxyester carbonyl. Similarly, H NMR spec-

troscopy on 11f shows that the signals for the protons

on C4, C5, and C6 are all shifted significantly upfield

compared to the analogous ethyl ester 1b, an effect that

would be expected from diamagnetic shielding from the

naphthyl ring. Similar shielding effects are observed in

other R,â-unsaturated arylmenthol esters.⁴¹

At this point in the study it would be premature to

propose a transition state to explain the observed ste-

reoselectivity, even with the crystal structure shown in

Figure 3. Indeed, a simple Zimmerman-Traxler model

with preferential Si-face approach of the aldehyde inac-

curately predicts that (R)-2c should be the preferred

(38) (a) Corey, E. J .; Ensley, H. E. J . Am. Chem. Soc. 1975, 97,

6908-6909. (b) Ohkata, K.; Miyamoto, K.; Matsumura, S.; Akiba, K.-y

Tetrahedron Lett. 1993, 34, 6575-6578. (c) Yang, D.; Xu, M. Org. Lett.

2001, 3, 1785-1788.

(39) d’Angelo, J .; Maddaluno, J . J . Am. Chem. Soc. 1986, 108, 8112-

8114.

(40) D’Oca, M. G. M.; Pilli, R. A.; Pardini, V. L.; Curi, D.; Comninos,

F. C. M. J . Braz. Chem. Soc. 2001, 12, 507-513.

(41) (a) Mezrhab, B.; Dumas, F.; d’Angelo, J .; Riche, C. J . Org. Chem.

1994, 59, 500-503. (b) Dumas, F.; Mezrhab, B.; d’Angelo, J . J . Org.

Chem. 1996, 61, 2293-2304.

In analogy to the results from the enantioselective

additions with the dual auxiliary allylboronate 13a (see

(37) Ndibwami, A.; Lamothe, S.; Guay, D.; Plante, R.; Soucy, P.;

Goldstein, S.; Deslongchamps, P. Can. J . Chem. 1993, 71, 695-713.

4418 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

F IGURE 3. ORTEP diagrams of â-naphthylboronate 11f: (a) view parallel to the plane of the â-naphthyl ring; (b) view

perpendicular to this plane.

preparation of 15 was relatively straightforward, this

chiral electrophile did not provide an effective quench in

the subsequent carbocupration reaction.

Allylboronate 13a was finally accessed by transesteri-

F IGURE 4. Design of the dual-auxiliary 2-carboxyester

allylboronate 13a .

fication of the diisopropyl allylboronate 17, which was

prepared from alkynoate 10e and electrophile 16b

(Scheme 4).⁴⁵The diisopropyl allylboronate 17 was not

isomer. Because of the uncertainty of the reactive con-

formation of the chiral allylboronate 11 at the transition

state, any models that could currently be proposed would

be purely speculative.

SCHEME 4

2.3. Qu a ter n a r y Ca r bon Cen ter s fr om En a n tio-

selective Allylbor a tion s: Du a l Au xilia r y Ap p r oa ch .

Since it was not possible to achieve sufficiently high levels

of enantioselectivity with a single chiral auxiliary, we

decided to explore a dual auxiliary approach (Figure 4).

We envisaged that allylboronate 13a , which bears chiral

auxiliaries on both the carboxyester and boronic ester,

would provide high levels of selectivity. Corey’s 8-phen-

ylmenthol 9e and Hoffmann’s bornanediol-derived 14⁴²

were chosen for this study because these two auxiliaries

were previously shown to be the most effective auxiliaries

for 2-carboxyester allylboronates.³⁴

Several preparations of allylboronate 13a were at-

tempted before an effective route was finally developed.

A simple preparation would entail replacement of the

pinacol in allylboronate 11e for diol 14. However, despite

literature precedents for this transformation,^29b,43all

attempts to transesterify the pinacol ester 11e failed in

our hands.

isolated but was rather treated with an ethereal solution

of 14 in the workup, directly yielding the desired chiral

allylboronate 13a .

We were delighted to see that this dual auxiliary

strategy did indeed afford a highly enantioselective

allylating reagent. Stirring boronate 13a in toluene with

aldehydes for 2 weeks at room temperature followed by

acid-catalyzed ring closure gave acceptable yields of

highly enantioenriched lactones 2 (Table 7). Both model

aliphatic and aromatic aldehydes were effective in this

reaction. Although the requirement for two different

auxiliaries might appear excessive, the enantioselectivi-

ties obtained are excellent. Furthermore, the methodol-

ogy is partially redeemed by the facile removal of the

auxiliaries from the product. Indeed, both auxiliaries are

simultaneously removed in the ensuing lactonization,

alleviating the need for separate cleavage steps.

We next prepared chiral electrophile 15 by trans-

esterification of the diisopropyl bromomethaneboronate

16a ⁴⁴followed by halogen exchange on the resulting

bromide (eq 4).²³We hoped to use this electrophile in

place of the pinacol electrophile 5a in the carbocupra-

tion-alkylation sequence. Unfortunately, although the

Lactone 2o was determined to have the (S)-configura-

tion by X-ray crystallography (see the Supporting Infor-

mation). The stereogenic centers in all other enantio-

enriched lactone products were assigned in analogy to

this result. Chiral HPLC data confirmed that all of the

(42) Herold, T.; Schrott, U.; Hoffmann, R. W.; Schnelle, G.; Ladner,

W.; Steinbach, K. Chem. Ber. 1981, 114, 359-374.

(43) (a) Mears, R. J .; Sailes, H. E.; Watts, J . P.; Whiting, A. J . Chem.

Soc., Perkin Trans. 1 2000, 3250-3263. (b) Singh, R. P.; Matteson, D.

S. J . Org. Chem. 2000, 65, 6650-6653.

(45) Electrophile 16b was prepared in a similar procedure as

reported in ref 44. See also: (a) Brown, H. C.; Roy, C. D.; Soundarara-

jan, R. Tetrahedron Lett. 1997, 38, 765-768. (b) Smil, D. V. Ph.D.

Thesis, University of Toronto, 2001, p 94.

(44) Michnick, T. J .; Matteson, D. S. Synlett 1991, 631-632.

J . Org. Chem, Vol. 69, No. 13, 2004 4419

Kennedy and Hall

TABLE 7. En a n tioselective Ad d ition s w ith Du a l

Au xilia r y Allylbor on a te 13a ^a

F IGURE 5. Model for stereoinduction proposed by Villie´ras

and co-workers.³⁴

entry

lactone

R

yield (%)

ee (%)

1

2

3

2b

2d

2o

Ph

C₉H₁₉

p-Br-C₆H₄-

55

80

40

>95^b

97

systems known to accelerate the addition of allyl-

boronates to carbonyl compounds. The sluggish reactivity

of the 3,3-disubstituted-2-carboxyester allylboronates 1

motivated us to look for a catalyst that could increase

the efficiency of allylations with these reagents.⁴⁶In

addition to decreasing the reaction times, establishing a

catalyzed allylboration reaction would open the door to

enantioselective catalysis and the exciting possibility that

enantiomerically pure quaternary carbon centers could

be generated with a substoichiometric chiral director

rather than with the two stoichiometric auxiliaries

currently required.

>95^b

a

Enantioselectivities determined by chiral HPLC. See the

Experimental Section for details. Selectivity uncertain because

of poor baseline resolution of the enantiomers. 95% is the lowest,

most conservative estimate.

b

TABLE 8. Ma tch ed a n d Mism a tch ed Au xilia r y Stu d ies

on Allylbor on a te 13a

The first step in our search for a catalyzed allylboration

was to identify possible metals that might act as cata-

lysts. Inspired by the study of Brown and co-workers,⁴⁷

we devised a simple, NMR spectroscopy based screen.

The reaction between benzaldehyde and allylboronate 1c

in dilute CD₂Cl₂with 0.5 equiv of potential catalyst (eq

1

5) was followed by H NMR spectroscopy. Specifically,

the disappearance of the ester CH₂signal from 1c (δ 4.18

pm, quartet) and the appearance of the signal for the

Z-olefin proton in 2f (δ 6.43 ppm, singlet) were monitored.

All of the data presented below came from ¹H NMR

experiments, despite the paramagnetic nature of some

of the metals studied which complicated the spectra.

entry

allylboronate

R¹

diol

ee (%)

1

2

3

4

13a

18

11f

13b

(-)-9e

ethyl

9f^a

(-)-14

pinacol

(+)-14

97

50

82

(-)-9e

<10

a

Reaction performed with the â-naphthyl allylboronate 11f and

not the parent 8-phenylmenthol 11e. Nevertheless, 11f is an

appropriate control substrate for this study since these two

allylboronates gave comparable results in the single auxiliary

study (see preceding section).

enantioselective allylborations in this study and the

single auxiliary study (see above) led to the same major

enantiomer of the lactone product.

A wide variety of transition and lanthanide metal salts

were then screened (see the graphs in the Supporting

Information). To our surprise, a large number of metals

were found to be capable of accelerating the allylboration

reaction. However, two salts, scandium triflate and

copper(II) triflate, displayed remarkable reactivity. In

fact, half-life times for reactions in the presence of these

salts were over 35 times shorter than those for uncata-

lyzed reactions. This large accelerating effect should be

sufficiently high to eventually allow for effective en-

antioselective catalysis without significant competition

from the racemic, background reaction.

These high selectivities are a result of a matched

combination of the (-)-8-phenylmenthol 9e and the (-)-

isomer of bornanediol 14. As seen in Table 8, the

selectivities observed with these two auxiliaries are

higher than selectivities observed with only one or the

other (entries 2 and 3) and are much higher than the

mismatched combination of (-)-8-phenylmenthol 9e and

the (+)-isomer of bornanediol 14 (entry 4).

At this stage of the project, it is not yet possible to offer

an accurate transition state model that incorporates both

the phenylmenthol auxiliary 9e as well as the bor-

nanediol 14 in the stereodifferentiating step. However,

the simple model proposed by Villie´ras and co-workers

to explain the stereochemical outcome of reactions with

boronates featuring only 14 accurately predicts that the

(S)-isomer of lactones 2 is produced in these reactions

(Figure 5).³⁴

(46) For the previously reported Lewis acid-catalyzed formation of

a reactive difluoroborane (a reaction which marks the first formal

catalysed allylboration), see: (a) Batey, R. A.; Thadani, A. N.; Smil, D.

V. Tetrahedron Lett. 1999, 40, 4289-4292. (b) Batey, R. A.; Thadani,

A. N.; Smil, D. V.; Lough, A. J . Synthesis 2000, 990-998. (c) Batey, R.

A.; Quach, T. D.; Shen, M.; Thadani, A. N.; Smil, D. V.; Li, S. W.;

MacKay, D. B. Pure Appl. Chem. 2002, 74, 43-55.

3.1. Lew is Acid Ca ta lyzed Allylbor a tion s: Ca ta -

lyst Sea r ch . Prior to our studies there were no catalytic

(47) Brown, H. C.; Racherla, U. S.; Pellechia, P. J . J . Org. Chem.

1990, 55, 1868-1874.

4420 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

TABLE 9. Lew is Acid Ca ta lyzed Allylbor a tion s w ith

Ben za ld eh yd e a n d Allylbor on a te 1c

Most of the mixtures in this initial screening were

heterogeneous, and we wondered if some of the catalysts

were performing poorly due to low solubility of the metal

salt in dichloromethane. Thus, a second screen was

carried out on a selected group of metal salts in a more

polar solvent system, 1:1 THF-d₈/CD₂Cl₂(see the graph

in the Supporting Information). Although many of the

mixtures were still heterogeneous, this second screening

identified ytterbium triflate as a viable lead catalyst.

While this salt gave only a mediocre performance in the

initial screening, it proved to be a remarkably effective

catalyst in the presence of THF.

Following the identification of lead catalysts, prelimi-

nary optimization studies determined that the triflate

salts of scandium and copper(II) were significantly more

effective than either the chloride or the acetate salts.

While ScCl₃still showed some catalytic activity, both the

acetate and chloride salts of copper were essentially

inactive in the reaction. Also, a brief survey of solvents

for the scandium triflate catalyzed reaction revealed that

the reaction rates were similar in either toluene or

dichloromethane but that coordinating solvents (e.g.,

acetonitrile, 1:1 THF/CH₂Cl₂) significantly retarded the

reaction.

In light of these studies, we were able to determine

that scandium triflate, copper(II) triflate, and ytterbium

triflate would be the best catalysts for use in a catalytic

allylboration reaction. However, before proceeding to

preparative-scale reactions we wanted to ensure that the

catalytic effect was truly due to the Lewis acidic metal.

Thus, one final series of NMR experiments was per-

formed.²⁵The reaction using tetrabutylammonium tri-

flate showed that triflate anion itself did not cause an

observable rate enhancement. We also considered the

possibility that the true catalyst might be triflic acid,

generated in situ by adventitious water and the metal

triflate. The reaction with tetrabutylammonium triflate

cast doubt on this possibility, and aqueous solutions of

rare earth metal triflates do not generate appreciable

amounts of triflic acid.⁴⁸Nevertheless, the ability of

Sc(OTf)₃to catalyze the reaction in the presence of an

acid scavenger was evaluated by performing the reaction

using a 1:1 mixture of Sc(OTf)₃/i-Pr₂NEt. This reaction

was marginally, but not significantly, slower than the

reaction without amine present, showing that the Lewis

acid was able to catalyze the reaction without generating

triflic acid. Diisopropylethylamine alone was not an

effective catalyst.⁴⁹

Following these NMR screening studies, it remained

to show that the catalytic effect would hold on a prepara-

tive scale and with truly catalytic loadings. In particular,

we hoped to achieve a reaction that was complete within

12-16 h at room temperature with a maximum of 10 mol

% catalyst loading. Equally important was the issue of

diastereoselectivity and whether the stereospecificity

seen in the thermal allylboration would be preserved or

not. Pleasingly, the Lewis acid catalyzed allylboration

satisfied all three of these conditions, and both aromatic

reaction

conditions

reaction

time (h)

yield

(%)

entry

catalyst

1

2

3

4

5

6

7

8

Sc(OTf)₃

Cu(OTf)₂

Yb(OTf)₃

none

0.5 M in toluene

1 M in toluene

1 M in CH₂Cl₂

0.5 M in toluene

1 M in toluene

0.5 M in toluene

1 M in CH₂Cl₂

toluene, 80 °C

20

24

20

16

20

16

71

93

78

67

73

91

93

60

and aliphatic aldehydes proved to be effective substrates

in this catalytic manifold. However, both kinds of alde-

hydes show minor differences in their reactivity, and for

this reason they will be discussed separately below.

3.2. Ca ta lytic Allylbor a tion w ith Ar om a tic Ald e-

h yd es. The initial catalytic reactions with benzaldehyde

showed a dramatic improvement over the thermal reac-

tions (Table 9). Reactions with all three lead catalysts

under practical conditions (10 mol % loading, 0.5-1 M

concentrations, 12-24 h) gave excellent yields of the

lactone 2f. Although the reactions are often complete

after 12 h, they were generally left overnight for practical

reasons. Most importantly, the stereospecificity of the

thermal reaction was preserved, and lactone 2f was

formed stereochemically pure in all catalyzed reactions.

It is remarkable that these catalytic reactions show the

same yields and reaction times as the uncatalyzed

processes (Table 4), except at a much lower temperature

(cf. control reaction, entry 8).

A noteworthy difference was observed in the water

tolerance of scandium triflate and copper triflate as

catalysts. Copper triflate lost all catalytic ability in the

presence of one equivalent of added water, but catalysis

could be restored to these reactions by adding molecular

sieves to the reaction mixture. However, reactions with

scandium triflate retained their catalytic effect in the

presence of up to one equivalent of added water, but

showed no catalysis at all in the presence of molecular

sieves. The reasons why sieves shut down catalysis with

scandium triflate are not clear at the moment, but this

tolerance to water led us to use scandium triflate as the

catalyst in most future allylborations.

This catalytic effect was extended to the reaction of

other allylboronates with a variety of aromatic aldehydes

(Table 10), including a sterically demanding aldehyde

(entry 6). As above, all of the lactone products were

isolated with a stereochemical purity reflecting that of

the allylboronate used.

One major limitation to this catalytic manifold is that

certain aldehydes do not give stereochemically pure

products (Table 11). To date, this problem has only been

observed with electron rich aromatic aldehydes and R,â-

unsaturated aldehydes. This stereochemical scrambling

is observed with all three lead catalysts in several

different solvents, and we have yet to find conditions that

allow for the effective, catalytic allylation of these alde-

(48) Kobayashi, S. Synlett 1994, 689-701.

(49) While the difference in reaction rates between the scandium-

catalyzed reactions in the presence and absence of diisopropylethyl-

amine might be due to a sequestering of the catalyst by the amine, it

might also be due to competitive coordination of the amine to the

boronate.

J . Org. Chem, Vol. 69, No. 13, 2004 4421

Kennedy and Hall

TABLE 10. Ca ta lytic Allylbor a tion s w ith Ar om a tic

Ald eh yd es^a

TABLE 12. Solven t a n d Ca ta lyst Su r vey for Ca ta lytic

Allylbor a tion of Alip h a tic Ald eh yd es w ith 1c^a

entry boronate lactone R¹

R²

R³

yield (%)

entry

catalyst

solvent

yield (%)

1^b

2^b

3^c

4^c

5^b

6^b

7^c

1b

1c

1e

2o

2q

2r

2s

2t

2u

2v

Me Me 4-Br-C₆H₄

62

57

48

47

58

44

75

1

2

3

4

5

6

Sc(OTf)₃

Cu(OTf)₂

Yb(OTf)₃

Sc(OTf)₃

Cu(OTf)₂

Yb(OTf)₃

toluene

CH₂Cl₂

64

63

66

62

52

38

Et Me 4-AcO-C₆H₄

Me Bu 3-I-C₆H₄

Me Bu C₆F₅

Me Bu 4-NO₂-C₆H₄

Me Bu 1-naphthyl

Me Bu 4-Me-C₆H₄

a

b

Reaction performed with 1.5 equiv of aldehyde and 0.5 M

All boronates and lactones were >20:1 dr. R ) Et or Me.

1

equiv of aldehyde used. ^c1.5 equiv of aldehyde used.

concentration.

TABLE 13. Lew is Acid Ca ta lyzed Allylbor a tion s w ith

Alip h a tic Ald eh yd es^a

TABLE 11. Lew is Acid Ca ta lyzed Allylbor a tion s w ith

P r oblem a tic Un sa tu r a ted Ald eh yd es^a

entry boronate R¹R²

R³

lactone

dr

yield (%)

entry lactone R¹R²

R³

yield (%) dr of 2

1

2

3^c

4

5

6

7

8

1c

1e

1a

Et Me BnOCH₂

Et Me i-Bu

Et Me c-Hex

Me Bu c-Hex

Me Bu i-Pr

Me Bu i-Bu

Me Bu Bu

2a d

2a e

2a f

2a g

2a h

2a i

2a j

2a k

19:1

>20:1

10:1^d

>20:1

53

46^b

54

32

61

62

33^b

1^b

2

2w

2x

Et Me 2-MeO-C₆H₄

Et Me E-2-(4-NO₂-C₆H₄)-

1-ethenyl

66

nd

2:1

2.8:1^c

3^d

4^d

5

2y

2z

2a a

2a b

Me Bu 4-MeO-C₆H₄

Me Bu E-1-propenyl

Et Bu 2-furyl

39

61

45

69^f

1.8:1

1:1

2.5:1^e

2.5:1

6

Et Bu 4-MeS-C₆H₄

H

Me BnOCH₂

a

All allylboronates 1 were >20:1 dr except as noted. nd ) yield

a

R ) Me or Et. ^b1 equiv of aldehyde used. ^cReaction performed

b

not determined. R ) Me or Et. Initial boronate was of 10:1 dr.

d

^cThermal reaction gave the product as one isomer in 80% yield.

using Cu(OTf)₂as catalyst. Allylboronate of 12:1 dr used in the

reaction.

1.5 equiv of aldehyde used. ^eThermal reaction gave the product

d

as one isomer in 74% yield. ^fReaction run for 7 days.

manifold to the allylboration of aliphatic aldehydes.

Using hydrocinnamaldehyde as a model substrate, we

performed a brief solvent and catalyst survey using

slightly different conditions than those described for the

allylation of aromatic aldehydes (Table 12). Like reactions

with aromatic aldehydes, these reactions were complete

in reasonable time frames at room temperature. All three

lead catalysts gave effective allylation in toluene, and

scandium triflate was clearly the most effective catalyst

in dichloromethane solvent. To maintain consistency with

the catalyzed reactions with aromatic aldehydes, scan-

dium triflate in toluene was chosen as the general

allylboration conditions.

A wide variety of aldehydes could be effectively al-

lylated using these established conditions (Table 13).

While the catalyzed reaction of aromatic aldehydes

showed a marked improvement over the thermal reac-

tion, the improvement seen in reactions with aliphatic

aldehydes was even more impressive. As above, the

reaction times were much shorter at lower temperatures

and the stereospecificity of the thermal reaction was also

hydes. However, all of these substrates give the expected,

stereochemically pure products in the uncatalyzed, ther-

mal reaction (see above).

We excluded the possibility that the loss of stereo-

selectivity arises from a competing Zimmerman-Traxler

transition state because this problem has not been

observed with sterically less demanding aldehydes (bu-

tanal) or with aldehydes that could coordinate to the

catalyst (benzyloxyacetaldehyde). This process appears

to be quite sensitive to the electronic nature of the

aldehyde; 4-acetoxybenzaldehyde does not suffer from

this epimerization problem (entry 2, Table 10) and may

be used as a substitute for the problematic 4-methoxy-

benzaldehyde (entry 3, Table 11).

We sought to gain evidence for this hypothesis by

attempting to induce epimerization of a stereochemically

pure sample of lactone 2x with scandium triflate. How-

ever, no epimerization of the lactone could be detected

by ¹H NMR spectroscopy, even after several days in dilute

solution and overnight in concentrated solution.

Currently the cause of this epimerization remains

unknown, and it may possibly occur at the stage of the

borate intermediate prior to the lactonization.⁵⁰

3.3. Ca ta lytic Allylbor a tion w ith Alip h a tic Ald e-

h yd es. We next focused on the extension of this catalytic

(50) There have been reports of the generation of carbocation

intermediates from allylic and benzylic alcohols by Lewis acids. See:

(a) Tsuchimoto, T.; Tobita, K.; Hiyama, T.; Fukuzawa, S.-i. J . Org.

Chem. 1997, 62, 6997-7005. (b) Crosby, S. R.; Harding, J . R.; King,

C. D.; Parker, G. D.; Willis, C. L. Org. Lett. 2002, 4, 577-580.

4422 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

SCHEME 5

Although the addition of excess aldehyde can compen-

sate for this side reaction, a more practical solution would

be to find conditions to completely suppress it. Not only

is the use of excess aldehyde wasteful (especially with

expensive or noncommercially available aldehydes), in

many cases this acetal impurity is difficult to remove

from the desired lactone product. Fortunately, the prob-

lem can be solved by the addition of a stoichiometric

amount of phenylboronic acid to the reaction mixture (eq

6). This boronic acid acts as a pinacol scavenger by

trapping the troublesome diol as the boronic ester 21.

Under these conditions, acetal formation is almost com-

pletely suppressed and reasonable yields of lactone

products are obtained from only 1 equiv of aldehyde. In

all cases studied to date, the phenylboronic pinacolate

21 is easily removed from the lactone products by flash

chromatography.

preserved. Furthermore, since these reactions are run at

temperatures well below 110 °C, the stereochemical

erosion sometimes observed with aliphatic aldehydes in

the thermal reactions is completely suppressed here.

The most significant advance of this new catalytic

manifold over the traditional thermal conditions is that

it allowed for the formation of allylation products from

hindered aldehydes (entries 3-5), substrates which

previously failed to give effective thermal reactions.

Although catalyzed reactions at room temperature with

these hindered substrates were sluggish, gentle heating

allowed for reactions to reach completion within 16 h.

Unlike reactions with aromatic aldehydes, the initial

catalytic reactions with aliphatic aldehydes were plagued

by low yields of lactones 2 and the competition of several

side reactions. Consequently, significant optimization

was required. First, the reactions with aliphatic alde-

hydes were best run at a concentration of 0.5 M in order

to avoid self-condensation of the aldehyde. While reac-

tions run at 1 M concentration only showed small

amounts of self-condensation, this side reaction was

completely suppressed under the more dilute conditions.

More importantly, however, is that the aldehyde must

be used in excess in order to obtain acceptable yields. The

use of excess aldehyde is required because of the forma-

tion of significant amounts of the pinacol acetal 19 from

the apparent reaction of unreacted aldehyde with the

borate byproduct 20 (Scheme 5). Although Lewis acids

are known to catalyze acetal formation,⁵¹the presence

of this acetal side product was quite surprising because

it was completely absent in the reactions with aromatic

aldehydes.

A comparison of the various methods for the allyl-

boration of aliphatic aldehydes is shown in Table 14.

From this table, it is clear that the catalytic reaction with

PhB(OH)₂scavenging (entries 2, 5, and 7) is superior to

both the thermal reaction (entry 3) or the use of scandium

triflate without PhB(OH)₂(entries 1, 4, and 6).

3.4. Ca ta lyzed En a n tioselective Allylbor a tion s.

Although a route to enantiopure lactones had already

been established, it was hampered by the inefficiency of

requiring two separate chiral auxiliaries for effective

stereoinduction (see above). The first attempt to increase

the efficiency of the enantioselective preparation of

lactone 2d was to apply the new catalytic manifold to

the reaction of an allylboronate bearing only one chiral

auxiliary (Table 15). However, these reactions were not

successful, and they generally gave lower yields and

selectivities than those obtained under the thermal,

TABLE 14. Com p a r ison of Meth od s for th e Allylbor a tion of Alip h a tic Ald eh yd es w ith 1^a

entry

allylboronate

lactone

equiv of R³CHO

conditions^b

Sc(OTf)₃

Sc(OTf)₃/PhB(OH)₂

thermal

Sc(OTf)₃

Sc(OTf)₃/PhB(OH)₂

Sc(OTf)₃

yield (%)

dr

1

2

3

4

5

6

7

1c

1a

2a d

2a e

2a k

1.5

1.0

1.5

1.0

53

54

33

46

66

33

66

>20:1

9.5:1^c

>20:1

Sc(OTf)₃/PhB(OH)₂

a

b

R ) Me or Et. Conditions used: Sc(OTf)₃(10 mol %) and/or PhB(OH)₂(1 equiv), rt; thermal: refluxing toluene. ^cInitial boronate

was >20:1 pure.

J . Org. Chem, Vol. 69, No. 13, 2004 4423

Kennedy and Hall

TABLE 15. Ca ta lyzed Allylbor a tion s w ith Ch ir a l

Allylbor on a tes 11e a n d 18^a

TABLE 16. Titr a tion of a Solu tion of Bor on a te 1c w ith

a

Sc(OTf)₃

entry

equiv of Sc(OTf)₃

22/1c ratio

1

2

3

4

0.5

1

1.5

2

0.5:1

1.2:1

3:1

10:1

a

Product ratio determined by ¹H NMR (400 MHz, CD₃CN).

entry allylboronate

catalyst

yield (%) ee (enantiomer)

1

2

3

4

5

6

11e

18

Sc(OTf)₃

Cu(OTf)₂

Yb(OTf)₃

Sc(OTf)₃

Cu(OTf)₂

Yb(OTf)₃

29

93

30

53

31

37

10% (R)

71% (S)

31% (S)

3% (R)

13% (R)

21% (R)

this study was to find evidence for the possible formation

of an activated boronate/scandium complex. The catalyst,

solvent, and substrate were all carefully chosen for this

study. Scandium(III) is not a paramagnetic nucleus and

its presence would not interfere with the NMR spectra

of other nuclei. The allylboronate 1c was chosen because

it is unsymmetrical about the γ-carbon and would allow

us to determine if the olefin geometry remains intact

under the catalytic conditions, and acetonitrile was used

as solvent because it gave homogeneous mixtures.

Titration of an NMR solution of allylboronate 1c with

Sc(OTf)₃led to a very clean spectrum showing only two

species, the free allylboronate 1c and the putative

complexed boronate 22 (Table 16). The amount of com-

plex formed was directly related to the amount of

scandium added, with one equivalent of complex formed

for every two equivalents of scandium added. At present,

it is not clear if this 1:2 relationship reflects the stoichi-

ometry of the complex or the position of the equilibrium

between bound and free boronate.

The amount of free and complexed scandium could not

be determined by ⁴⁵Sc NMR spectroscopy because of the

poor resolution of the spectra. The ¹⁹F NMR spectra

showed only a single, sharp peak, suggesting either the

formation of a complex where all of the triflate anions

are equivalent or that complexed and free triflate anions

are equilibrating very rapidly.

To determine the importance of the boronic ester group

in the formation of this complex, we conducted a control

experiment with tetrasubstituted acrylate 23. Treatment

of this R,â-unsaturated ester with scandium triflate

showed no evidence for the formation of a complex; the

¹H NMR spectrum of 23 was virtually unchanged even

after the addition of one full equivalent of scandium

triflate. This result shows that the boronic ester is

essential for complex formation between scandium tri-

flate and 1c, and it lends credence to the presence of a

metal-boronate interaction.

a

Noncatalyzed reaction with the â-naphthyl analogue of 11e

gave 82% ee for the (S)-enantiomer; noncatalyzed reaction with

18 gave 51% ee for the (S)-enantiomer (cf. Table 8).

noncatalyzed conditions. Interestingly, the enantiomer

produced in the catalyzed reaction was often the opposite

to that generated in the thermal reaction. These results

contrast sharply with the highly selective scandium-

catalyzed reactions of allylboronates derived from bor-

nanediol 14 but which lack the 2-carboxyester substitu-

ent.⁵²

The ultimate goal is to eventually find a chiral catalyst

that allows for the control of the absolute stereochemistry

of the lactone products using no chiral auxiliary. Al-

though Miyaura and co-workers reported modest success

in the catalyzed addition of allyl and E-crotylboronates,⁵³

there is currently no catalyst system that provides

preparatively useful levels of enantioinduction in these

reactions. A preliminary evaluation of known chiral

scandium catalysts⁵⁴failed to reveal an effective catalyst.

The general problem encountered is that the allylboration

reaction is suppressed in the presence of the large, chiral

ligands used in these catalytic systems.

4.1. P r elim in a r y Mech a n istic Stu d ies: NMR Evi-

d en ce for th e F or m a tion of a Bor on a te/Sca n d iu m

Com p lex. Aside from satisfying our scientific curiosity,

a better mechanistic understanding of this novel catalytic

process could potentially allow for a rational and more

effective approach to further optimization of the reaction,

especially in the search for enantioselective catalyzed

variants. To this end, the effect of various amounts of

scandium triflate on the NMR spectrum of allylboronate

1c in the absence of aldehyde was studied. The aim of

(51) Fukuzawa, S.; Tsuchimoto, T.; Hotaka, T.; Hiyama, T. Synlett

1995, 1077-1078.

(52) Lachance, H.; Lu, X.; Gravel, M.; Hall, D. G. J . Am. Chem. Soc.

2003, 125, 10160-10161.

(53) Ishiyama, T.; Ahiko, T.-a.; Miyaura, N. J . Am. Chem. Soc. 2002,

124, 12414-12415.

(54) (a) Kobayashi, S.; Araki, M.; Hachiya, I. J . Org. Chem. 1994,

59, 3758-3759. (b) Evans, D. A.; Masse, C. E.; Wu, J . Org. Lett. 2002,

4, 3375-3378. (c) Evans, D. A.; Sweeney, Z. K.; Rovis, T.; Tedrow, J .

S. J . Am. Chem. Soc. 2001, 123, 12095-12096.

Attempts to perform a similar study with an allyl-

boronate that lacked the 2-carboxyester group (e.g., 24)

4424 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

TABLE 17. ¹¹B NMR Sh ifts for F r ee a n d Com p lexed

Allylbor on a te 1c^a

TABLE 18. ¹H a n d ¹³C NMR Sh ifts for Allylbor on a te 1c

a

in th e P r esen ce of Sc(OTf)₃

free allylboronate 1c

complexed allylboronate 22a

33.8

12.8 (major)

2.8 (minor)

¹³C NMR (100 MHz)

¹H NMR (400 MHz)

signal

free 1c

complexed 1c

free 1c

complexed 1c

a

Spectra taken in CD₃CN at 64 MHz. The structure of the

1

2

14.6

60.8

14.2

70.5

1.23

4.09

1.43

4.62

scandium/boronate complex 22a is hypothetical.

3

4

5

6

7

8

9

10

11

170.0

124.7

147.2

30.0

13.3

20.3

14.2

84.0

25.0

181.2

123.6

167.8

29.8

13.0

23.9

19.6

84.2

24.3

failed because the reagent generally decomposed at room

temperature in the presence of scandium triflate. While

it is not yet known if a boronate/scandium complex forms

in the absence of a carboxyester group, there is ample

evidence to show that this ester group is not required

for catalysis to occur. Studies from our group⁵²and

Miyaura’s⁵³have shown that boronates that lack this

2-carboxyester group also undergo a metal-catalyzed

allylboration reaction, although the magnitude of the

catalysis observed at room temperature is much smaller

with these boronates (approximately a 3-fold rate in-

crease for 24) than with boronate 1c (35-fold increase).^14b

This result might simply reflect the large difference in

reactivity between boronates 1c and 24 (reaction with

24 is complete within a couple of hours rather than days).

Alternatively, it might suggest that the ester in 1c acts

as a second coordination site for the catalyst, thereby

increasing the binding of the boronate to the catalyst and

allowing for greater activation to occur.

NMR techniques were then employed to probe the

structure of the complex formed between boronate 1c and

scandium triflate. The most interesting result came from

the ¹¹B NMR spectrum of the complex (Table 17). While

the spectrum of 1c displayed a single peak at 33.8 ppm

in the absence of scandium triflate, the spectrum of the

complex showed two new signals, both upfield of the

original signal. The larger of these new signals resonates

at 12.8 ppm, and is consistent with the formation of a

tetrahedral complex, likely with acetonitrile as ligand

(22a ).⁵⁵This upfield signal does not form in the absence

of Sc(OTf)₃, nor is it observed if the spectrum is taken in

CD₂Cl₂. The third signal, a small, sharp peak at 2.8 ppm,

has yet to be assigned. These results may be interpreted

to mean that while the boron in 1c is normally not

sufficiently electrophilic to bind to acetonitrile, formation

of the boronate:scandium complex 22a activates the

boron enough so that it can accept the nitrile as a ligand.

The ¹H and ¹³C NMR signals for the free and com-

plexed boronate 1c are shown in Table 18. The downfield

shifts seen in the ¹³C resonances for the carbons in the

unsaturated ester (especially carbons 2, 3, and 5) are

consistent with the binding of the ester carbonyl to the

scandium metal.⁵⁶However, the resonances of the car-

2.33

1.00

1.74

1.76

2.60

1.07

1.98

1.56

1.19

1.36

6.92^b

7.96^b

a

Spectra taken in CD₃CN. Assignments confirmed by 2D NMR

techniques. Broad signals due to coordinated water not present

in the absence of scandium triflate.

b

bons and protons around the boron are not as easily

interpreted. Whereas the chemical shifts for the pinacol

carbons (10 and 11) hardly change at all, the carbon R to

the boron (carbon 9) shifts 5 ppm downfield and the

protons attached to this carbon (protons 9) shift 0.2 ppm

upfield. This difference in behavior for the carbon and

the protons next to the boron may be explained by

considering the putative acetonitrile adduct 22a . Binding

of the scandium to one or both of the boronate oxygen

should make the boron atom more electron deficient by

disrupting the overlap between the oxygen lone pair

electrons and the empty p-orbital on the boron. Inductive

effects would then shift the carbon signal downfield.

Meanwhile, the methylene protons might experience a

pronounced shielding effect from the p-orbital of the

boron, which becomes filled after coordination of the

acetonitrile, or from anisotropic effects of the proximal

ester or nitrile groups.

Many of the peaks in the NMR spectrum of the complex

were significantly broadened, especially the pinacol

methyl peaks. If scandium is capable of coordinating to

one (or both) of the pinacol oxygens, then the four methyl

groups on the pinacol ring should become nonequivalent.

While the broadened peak for the pinacol protons could

not be resolved into separate peaks, even at low temper-

atures, the fact that the broadening increases as the

temperature decreases is consistent with this theory.

One intriguing result from this NMR study was that

a long-range correlation was observed between the

tertiary carbon on the pinacol group (carbon 10) and a

coordinated water. This HMBC correlation suggests that

water might be involved in the coordination of scandium

to the boronate, and might help to explain why no

catalysis is observed with scandium triflate in the pres-

ence of molecular sieves.

(55) Lavigne, J . J .; Anslyn, E. V. Angew. Chem., Int. Ed. 1999, 38,

3666-3669.

(56) (a) Clapham, G.; Shipman, M. Tetrahedron Lett. 1999, 40,

5639-5642. (b) Clapham, G.; Shipman, M. Tetrahedron 2000, 56,

1127-1134.

J . Org. Chem, Vol. 69, No. 13, 2004 4425

Kennedy and Hall

TABLE 19. Com p a r ison of th e Rea ction s of

Allylbor on a te 1c w ith Th ose of th e An a logou s

Allylsta n n a n e 25 a n d Allylsila n e 26

4.2. Issu e of Typ e I or Typ e II Mech a n ism . One of

the first mechanistic questions to address is whether the

reaction follows the pattern of a closed, Type I mechanism

or an open, Type II mechanism (Figure 1). The strongest

evidence for one mechanism over the other comes from

the stereochemistry of the products of the allylations. The

Type I mechanism involves the transfer of stereochem-

istry of the allylboronate to the product, and thus the

diastereoisomer of the product formed in these reactions

depends on the E/Z geometry of the starting material

used. In contrast, the Type II allylation pathway tends

to be stereoconvergent, and the same isomer of product

dominates regardless of the E/ Z stereochemistry of the

starting material.^18b

The stereospecificity observed in the catalyzed allyl-

borations is consistent with that of the Type I mecha-

nism. The reaction of an allylboronate of moderate

stereochemical purity is especially telling (eq 7). This

result confirms that the high stereochemical purity

generally observed in the products of the catalyzed

allylboration reaction comes from a transfer of stereo-

chemistry from stereochemically pure allylboronates and

not from a stereoconvergent process that favors one

isomer of the product over another. Although not impos-

sible, it is unlikely that an open transition state would

allow for such a perfect transfer of stereochemistry from

the allylboronate 1c to the lactone 2a f.

complex than in the free boronate. This data is not

consistent with an increase in nucleophilicity of this

carbon, and consequently it is unlikely that ester decon-

jugation is the root of the observed catalytic effect.

Furthermore, the presence of a carboxylic ester is not

required for catalysis to occur.^14a,52,53

Another possibility is that the enhanced reaction rate

is due to the formation of a new, highly reactive al-

lylscandium species following a transmetalation process.

While some catalyzed allylstannation reactions are re-

ported to proceed via a transmetalation of the initial

allylmetal to a more reactive allylmetal species,^18bthere

are no known examples of an allylboration that proceeds

via a similar mechanism, nor have there been any reports

of a boron to scandium or ytterbium transmetalation. In

this current study, we were able to treat a solution of

allylboronate 1c with excess scandium triflate, find

evidence for the formation of a complex between the

metal and the boronate by NMR, and then afterward

recover the boronate unchanged and in good yield (Scheme

6). This recovery suggests that transmetalation of the

boronate under the catalytic conditions is unlikely.

Further evidence against a transmetalation mechanism

is the fact that catalyzed allylborations with allylbor-

onates bearing chiral diol auxiliaries showed appreciable

levels of stereoselectivity (see above). If the reaction

proceeded via transmetalation, these reactions in the

presence of an achiral Lewis acid should have shown no

selectivity at all.

Further evidence that the catalyzed allylboration is

consistent with a Type I process comes from the reaction

of benzaldehyde with allylstannane 25 and allylsilane 26

(Table 19).⁵⁷Reactions of these known Type II reagents

with benzaldehyde gave significantly different stereo-

chemical results than the analogous allylboronate 1c,

strongly suggesting that the allylboronate 1c does not

behave like a Type II reagent in the presence of an

external Lewis acid.

4.3. P ossible Or igin s of th e Activa tion Effect. All

of the evidence obtained thus far supports the hypothesis

that the catalyzed allylboration reaction proceeds via a

closed, cyclic transition state following the formation of

a scandium-allylboronate complex. However, there are

still several different ways that the formation of this

complex could cause the catalytic effect. For example,

coordination of the scandium ion both to the boronate

and the ester might cause the latter to twist out of

conjugation with the olefin, thus rendering the γ-carbon

(carbon 5 in Table 18) more nucleophilic in the complex

than in the free boronate. However, the ¹³C NMR

spectrum of the complex showed that the signal for the

γ-carbon shifts downfield by 20 ppm in the complex

compared to the free boronate, suggesting that this

carbon is significantly more electron deficient in the

Consolidating all of the evidence presented above, the

most reasonable proposal for the mode of activation

currently is the electrophilic activation of boron by

coordination of scandium to one of the boronate oxygens

(Figure 6). This coordination may be direct (transition

state A) or mediated by water (transition state B).⁵⁸

Formation of this complex would increase the Lewis

acidity of the boron atom, which then leads to enhanced

(58) Analogously, Aggarwal and co-workers reported that the rate

enhancement in the lanthanum triflate catalysed Baylis-Hillman

reaction derives from enhanced hydrogen bonds between the substrate

and coordinated hydroxyl groups on the Lewis acid rather than from

direct binding of the lanthanide to the substrate. See: Aggarwal, V.

K.; Mereu, A.; Tarver, G. J .; McCague, R. J . Org. Chem. 1998, 63,

7183-7189.

(57) Zhu, N. M.Sc. Thesis, University of Alberta, 2003, pp 36-50.

4426 J . Org. Chem., Vol. 69, No. 13, 2004

Lewis Acid Catalyzed Allylboration

SCHEME 6

TABLE 20. Dia ster eoselective Hyd r ogen a tion of

La cton es 2

allylboration with an aldehyde via the usual Zimmer-

man-Traxler transition state. Binding of the Lewis acid

to one of the boronate oxygens would disrupt the n_O

f

p_Boverlap, making the boron atom more electron defi-

cient, and would increase the strength of the boron-

carbonyl coordination and in turn lower the activation

energy of the transition state. Molecular calculations

have shown that the strength of this coordination is the

main factor in lowering the activation energy of an

allylboration reaction.⁵⁹This proposal is also in line with

the empirical study of Brown and co-workers who con-

cluded that the reactivity of a given allylboronate is

directly related to the availability of the lone pair

electrons on the boronate substituents.⁴⁷

5.1. F u r th er F u n ction a liza tion of La cton es 2. The

exo-methylene butyrolactone products of the allylboration

reactions are related to a family of biologically active

natural products, which have shown promise as potential

therapeutics in the treatment of a variety of ailments,

ranging from cancer⁶⁰to alcoholism.⁶¹The activity of

these compounds is proposed to derive primarily from

their ability to act as Michael acceptors for suitable

biological nucleophiles (e.g., the thiol of a cysteine

residue).^61,62However, most of these compounds are too

toxic for clinical use, and the need for analogues has

spurred interest in establishing synthetic routes to these

butyrolactones.^63,64

Notwithstanding the inherent value of this class of

compounds, the primary focus of our research was the

generation of stereodefined quaternary carbon centers.

We believed that our methodology would find wider

application and use if the resulting quaternary carbon

centers were not necessarily confined to a five-membered

ring. To this end, both the ester functionality and the

exocyclic olefin provide useful handles for the transfor-

mation of the lactones into other structures

We first investigated the simple hydrogenation of the

exocyclic double bond in 2. Although the product of this

reduction is still a cyclic structure, it contains three

contiguous stereogenic centers, and we expected that the

cyclic nature of the substrate might help control the

stereochemistry of the hydrogenation. In the event, this

reduction exceeded our expectations, and the olefin

underwent a highly selective reduction under mild condi-

tions to give only one diastereomer of the R-methyl

entry lactone product R¹

R²

R³

C₉H₁₉

Ph

yield (%)

1

2

3

4

2e

2f

2k

2a c

27a

27b

27c

27d

Et

Me

85

86

81

88

Me i-Bu 4-MeOC₆H₄-

Et Me PhCH₂CH₂-

lactone 27 (Table 20). No other isomer of the product

1

could be detected by H NMR spectroscopy. Pleasingly,

no benzylic C-O cleavage was observed with the aro-

matic lactones under these conditions (entries 2 and 3).

The stereochemistry of the lactone 27c was unambigu-

ously determined by X-ray crystallography, showing that

the hydrogen had added from the face opposite to the R³

group (see the Supporting Information). The stereochem-

istry of the other products was then inferred from this

result.

The reductive opening of butyrolactone 27 would lead

to an acyclic stereochemical triad, a structural motif

found in many natural products. The generation of

multiple contiguous stereogenic centers is still a chal-

lenge in synthetic chemistry, especially when one of these

centers is a quaternary carbon.⁶⁵In the present case,

after extensive investigation of several hydride reagents,

butyrolactone 27a was found to undergo reduction with

Red-Al to give diol 28 in a nonoptimized yield of 80% (eq

8). This reaction illustrates the broad synthetic potential

of the stereochemically pure lactones produced by this

methodology. The investigation of other transformations

of lactones 2 is ongoing in our laboratory.

(59) (a) Gajewski, J . J .; Bocian, W.; Brinchford, N. L.; Henderson,

J . L. J . Org. Chem. 2002, 67, 4236-4240. (b) Omoto, K.; Fujimoto, H.

J . Org. Chem. 1998, 63, 8331-8336.

(60) (a) Fardella, G.; Barbetti, P.; Grandolini, G.; Chiappini, I.;

Ambrogi, V.; Scarcia, V.; Candiani, A. F. Eur. J . Med. Chem. 1999,

34, 515-523. (b) Lee, K.-H.; Huang, B.-R.; Tzeng, C.-C. Bioorg. Med.

Chem. Lett. 1999, 9, 241-244. (c) Lawrence, N. J .; McGown, A. T.;

Nduka, J .; Hadfield, J . A.; Pritchard, R. G. Bioorg. Med. Chem. Lett.

2001, 11, 429-431.

(61) Matsuda, H.; Shimoda, H.; Uemura, T.; Yoshikawa, M. Bioorg.

Med. Chem. Lett. 1999, 9, 2647-2652.

(62) (a) Hoffmann, H. M. R.; Rabe, J . Angew. Chem., Int. Ed. Engl.

1985, 24, 94-110. (b) J anecki, T.; Blaszczyk, E.; Studzian, K.; Ro´zalski,

M.; Krajewska, U.; J anecka, A. J . Med. Chem. 2002, 45, 1142-1145.

(63) For reviews, see: Grieco, P. Synthesis 1975, 67-82. See also

ref 62a.

(64) For recent preparations, see: (a) Knochel, P.; Sidduri, A. R. J .

Am. Chem. Soc. 1992, 114, 7579-7581. (b) Sa´nchez-Sancho, F.;

Valverde, S.; Herrado´n, B. Tetrahedron: Asymmetry 1996, 7, 3209-

3246. (c) Paquette, L. A., Me´ndez-Andino, J . Tetrahedron Lett. 1999,

40, 4301-4304. (d) Xu, M.-H.; Wang, W.; Xia, L.-J ., Lin, G.-Q. J . Org.

Chem. 2001, 66, 3953-3962. (e) Dumeunier, R.; Leclercq, C.; Marko´,

I. E. Tetrahedron Lett. 2002, 43, 2307-2311. (f) Kang, S.-K.; Kim, K.-

J .; Hong, Y.-T. Angew. Chem., Int. Ed. 2002, 41, 1584-1586.

F IGURE 6. Possible transition states for the Lewis acid

catalyzed allylboration.

J . Org. Chem, Vol. 69, No. 13, 2004 4427

Kennedy and Hall

With the development of the catalytic allylboration

reaction comes the promise of an enantioselective, cata-

lyzed version. Preliminary studies with some common

chiral catalysts have yet to reveal an effective catalyst

for these reactions. While a systematic survey of other

catalysts and ligands might eventually lead to a success-

ful, substoichiometric enantioselective allylboration sys-

tem, a deeper understanding of the reaction mechanism

would allow for a more effective and rational approach

to solving this problem. Recent mechanistic studies

addressed the origin of this catalytic effect in the case of

2-unsubstituted allylboronates.⁶⁶These results give cre-

dence to the hypothesis that allylboronates 1 are subject

to electrophilic activation of the boron atom following

formation of a boronate-Lewis acid complex. It also raises

the intriguing idea that this kind of boronate activation

could be a general phenomenon that might find applica-

tions outside the realm of allylboration chemistry.

Con clu sion

This paper described the efficient generation of ste-

reogenic quaternary carbon centers via the stereospecific

reaction of 2-carboxyester-3,3-disubstituted allylbor-

onates (1) with aldehydes. This reaction generates di-

astereomerically pure exo-methylene butyrolactones 2

that embed the desired quaternary carbon center. These

products can also be formed in enantiopure fashion using

a dual-auxiliary strategy. In addition to being biologically

interesting targets in their own right, these highly

functionalized butyrolactones can undergo further useful

transformations (e.g., reduction to a 1,4-diol bearing three

contiguous stereogenic centers). The inherent utility of

these lactones coupled with the relative ease and expedi-

ence of their preparation suggest that they are well suited

to find use as synthetic intermediates in natural product

synthesis.

The most important advance to come from this study

is the development of the Lewis acid catalyzed carbonyl

allylation with allylboronates. Under this new catalytic

manifold, reactions with 1 that previously took 2 weeks

at room temperature are now complete within 12 hours

at room temperature. This catalysis not only provides a

large rate enhancement over the noncatalyzed process

but also broadens the scope of aldehydes that may be

used as substrates. Crucially, the diastereospecificity

observed in the thermal reaction of 2-carboxyester allyl-

boronates is preserved in the Lewis acid catalyzed

reaction.

Ack n ow led gm en t. This work was funded by the

Natural Sciences and Engineering Research Council of

Canada (NSERC), the University of Alberta, and by an

AstraZeneca Award in Chemistry to D.G.H. J .W.J .K.

thanks NSERC, the Alberta Heritage Foundation for

Medical Research (AHFMR), and the University of

Alberta Faculty of Graduate Studies and Research

(FGSR) for graduate scholarships. We thank Glen

Bigam, Gerdy Aarts, and Lai Kong for NMR analyses,

Dr. Robert McDonald and Dr. Michael J . Ferguson of

the University of Alberta X-ray Crystallography Labo-

ratory for structure determinations, Barry B. Toure´ for

helpful discussions during the lactone reduction, as well

as Naheed Rajabali and Melissa Chee for the prepara-

tion of diol 14 and butyrolactone 27d .

Su p p or tin g In for m a tion Ava ila ble: Full experimental

details for catalyst screening and for the preparation of all

compounds; ¹H and ¹³C NMR spectra for previously unreported

products; technical details for the collection of the X-ray

crystallography data; structural confirmation (NOE data,

X-ray structures) for allylboronate 1c and lactones 2i, 2j, 2n ,

2o, and 27c; spectra and details of ¹H, ¹³C, and ¹¹B NMR

studies of the 1c:Sc(OTf)₃complex. This material is available

free of charge via the Internet at http://pubs.acs.org.

(65) For some examples of products containing three or more

contiguous stereogenic centers, see: (a) Kocienski, P. J .; Brown, R. C.

D.; Pommier, A.; Procter, M.; Schmidt, B. Perkin Trans. 1 1998, 9-39.

(b) Yoda, H.; Katoh, H.; Takabe, K. Tetrahedron Lett. 2000, 41, 7661-

7665. (c) Lo´pez-Alvarado, P.; Garc´ıa-Granda, S.; AÄ lvarez-Ru´a, C.;

Avendan˜o, C. Eur. J . Org. Chem. 2002, 1702-1707. (c) Harrowven,

D. C.; Luca, M. C.; Howes, P. D. Tetrahedron Lett. 1999, 40, 4443-

4444. (d) Cassayre, J .; Zard, S. Z. J . Organomet. Chem. 2001, 624, 316-

326. (e) Gennari, C.; Ceccarelli, S.; Piarulli, U.; Aboutayab, K.; Donghi,

M.; Paterson, I. Tetrahedron 1998, 54, 14999-15016.

J O049773M

(66) Rauniyar, V.; Hall, D. G. J . Am. Chem. Soc. 2004, 126, 4518-

4519.

4428 J . Org. Chem., Vol. 69, No. 13, 2004

Article Doi

Lewis acid catalyzed allylboration: Discovery, optimization, and application to the formation of stereogenic quaternary carbon centers

DOI: 10.1021/jo049773m

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jo049773m

Full text of DOI:10.1021/jo049773m

Products guided by the article

R&D Labs maybe for 732302-20-4

Relevant to this article

Hot Product