The exceptional chelating ability of dimethylaluminum chloride and methylaluminum dichloride. The merged stereochemical impact of α- and β-stereocenters in chelate-controlled carbonyl addition reactions with enolsilane and hydride nucleophiles

Article abstract of DOI:10.1021/ja011337j

A systematic investigation of the stereoselectivity in Lewis acid-promoted (Mukaiyama) aldol reactions of achiral unsubstituted enolsilanes and chiral β-hydroxy aldehydes proceeding under conditions favoring chelation control is presented. Good stereocontrol can be realized for enolsilane aldol reactions of β-alkoxy and β-silyloxy aldehydes bearing only an α- or a β-stereogenic center. Examination of the chelated intermediates for α,β-disubstituted aldehydes concludes that the syn aldehyde diastereomer possesses the arrangement of stereocenters wherein the α- and β-substituents impart a reinforcing facial bias upon the aldehyde carbonyl. Aldol reactions of syn aldehydes were thus observed to proceed with uniformly excellent diastereofacial selectivity. Aldol reactions of the corresponding anti aldehydes containing opposing stereocontrol elements at the α- and β-positions exhibit variable and unpredictable selectivity.

Full text of DOI:10.1021/ja011337j

10840
J. Am. Chem. Soc. 2001, 123, 10840-10852
The Exceptional Chelating Ability of Dimethylaluminum Chloride
and Methylaluminum Dichloride. The Merged Stereochemical Impact
of R- and â-Stereocenters in Chelate-Controlled Carbonyl Addition
Reactions with Enolsilane and Hydride Nucleophiles
David A. Evans,* Brett D. Allison, Michael G. Yang, and Craig E. Masse
Contribution from the Department of Chemistry & Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138
ReceiVed May 31, 2001
Abstract: A systematic investigation of the stereoselectivity in Lewis acid-promoted (Mukaiyama) aldol reactions
of achiral unsubstituted enolsilanes and chiral â-hydroxy aldehydes proceeding under conditions favoring
chelation control is presented. Good stereocontrol can be realized for enolsilane aldol reactions of â-alkoxy
and â-silyloxy aldehydes bearing only an R- or a â-stereogenic center. Examination of the chelated intermediates
for R,â-disubstituted aldehydes concludes that the syn aldehyde diastereomer possesses the arrangement of
stereocenters wherein the R- and â-substituents impart a reinforcing facial bias upon the aldehyde carbonyl.
Aldol reactions of syn aldehydes were thus observed to proceed with uniformly excellent diastereofacial
selectivity. Aldol reactions of the corresponding anti aldehydes containing opposing stereocontrol elements at
the R- and â-positions exhibit variable and unpredictable selectivity.
Objectives
The integration of chelate organization into the design of
stereoselective processes is widespread. Numerous examples
incorporate this stereochemical control element into diastereo-
selective and enantioselective carbonyl addition,1,2 chiral eno-
late-electrophile reactions,3 and cycloadditions.4 During the
development of the oxazolidinone-based Diels-Alder reactions
some years ago,4a a number of Lewis acids, including both SnCl4
and TiCl4, were surveyed by us for their ability to activate the
dienophilic component through chelate organization; however,
none of these Lewis acids delivered either the reactivity or the
diastereoselectivity displayed by dimethylaluminum chloride
(Me2AlCl), which was proposed to chelate the substrate through
the illustrated cationic complex (eq 1).4a,5 The present study
has extended the exceptional chelating potential of Me2AlCl,
and its companion Lewis acid MeAlCl2, to chelate-organized
carbonyl addition reactions where the chelating heteroatom may
include alkyl ethers as well as hindered silyloxy substituents
(eq 2).5
The objectives of this investigation are two-fold: (A) to
document the scope and limitations of Me2AlCl and MeAlCl2
as chelating Lewis acids in enolsilane addition reactions with
â-alkoxycarbonyl substrates and (B) to document that the
chelate-promoted enolsilane addition reactions of diastereomeric
aldehydes A and B reveal either enhanced carbonyl face
selectivities for the syn aldehyde diastereomer (eq 3) or eroded
face selectivities for the corresponding anti diastereomer (eq
4). Such predictions follow from an analysis of the metal
(1) (a) For an excellent review of Cram’s rule, including a review of
chelation controlled reactions, see: Mengel, A.; Reiser, O. Chem. ReV. 1999,
99, 1191-1223. (b) Reetz, M. T. Acc. Chem. Res. 1993, 26, 462-468 and
references therein.
(2) Evans, D. A.; Kozlowski, M. C.; Murry, J. A.; Burgey, C. S.; Connell,
B. J. Am. Chem. Soc. 1999, 121, 669-685. Evans, D. A.; Burgey, C. S.;
Kozlowski, M. C. J. Am. Chem. Soc. 1999, 121, 686-699 and references
therein.
(3) Evans, D. A. Aldrichim. Acta 1982, 15, 23.
(4) (a) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc. 1988,
110, 1238-1256. (b) Evans, D. A.; Miller, S. J.; Lectka, T. J. Am. Chem.
Soc. 1993, 115, 6460-6461. (c) Evans, D. A.; Olhava, E. J.; Johnson, J.
S.; Janey, J. M. Angew. Chem., Int. Engl. 1998, 24, 3372-3375.
(5) For a preliminary communication of aspects of this study see: (a)
Evans, D. A.; Allison, B. A.; Yang, M. G. Tetrahedron Lett. 1999, 40,
4457-4460. (b) Evans, D. A.; Halstead, D. P.; Allison, B. A. Tetrahedron
Lett. 1999, 40, 4461-4462.
10.1021/ja011337j CCC: $20.00 © 2001 American Chemical Society
Published on Web 10/13/2001

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10841
chelates derived from syn and anti aldehyde diastereomers,
respectively. This stereochemical issue has not been systemati-
cally explored particularly in comparison with the corresponding
reactions under nonchelating conditions.6 This investigation is
intended as a companion to our earlier study in this area where
it was documented that the two stereocenters in syn aldehyde
A are nonreinforcing in carbonyl additions promoted by
nonchelating Lewis acids while the two stereocenters in anti
aldehyde B are reinforcing.7
Background
The analysis of carbonyl π-facial selectivity has attracted
immense interest since Cram’s pioneering studies on the
stereoselective addition of organometallic reagents to chiral
acyclic carbonyl substrates bearing vicinal alkyl and heteroatom
substituents.8 In his series of investigations, “open-chain” and
“chelation” transition state models were proposed to account
for stereoselective nucleophilic carbonyl addition reactions to
these families of substrates, respectively. While Cram’s models
for stereoinduction followed from the results of organometallic
addition reactions, these transition state models through their
modern refinements (Felkin-Anh)9 have been applied to the
broader field of Lewis acid-mediated carbonyl addition pro-
cesses.10 Carbonyl substrates such as A and B (eqs 3and 4) that
exhibit the potential for chelation-controlled addition1,11 are of
particular interest in this investigation. However, such substrates,
while exhibiting the potential for chelate control, may react
through either “open-chain” (Felkin) or “chelated” transition
states. In substrates such as R-alkoxy carbonyl derivatives, the
consequence of either Felkin monodentate (eq 5) or chelate
carbonyl activation (eq 6) has a direct bearing on the stereo-
chemical outcome of the reaction. In fact, the stereochemical
outcome of this reaction provides strong circumstantial evidence
of the mode of carbonyl activation.12
allylstannanes to R-alkoxy aldehydes (eqs 7 and 8).14 In the cited
examples, the impact of the oxygen protecting group on the
mode of substrate activation is illustrated. This and related cases
provide additional evidence that hindered silyl ethers do not
generally participate in chelate organization.
â-Chelation: 1,2-Induction. The currently embraced Felkin-
Anh model8d,e and the chelate-controlled addition model are
illustrated in Scheme 1 for R-methyl â-alkoxy aldehydes 1 and
2 (eqs 9 and 10). As with R-alkoxy aldehydes, the two control
elements lead to different product diastereomers. It has been
well precedented that metal ion chelation between the carbonyl
and â-oxygen substituent provides a conformationally con-
strained six-membered ring having sterically differentiated
diastereofaces (eq 10). Observation of Lewis acid-substrate
complexation by NMR spectroscopy suggests that the favored
chelate conformation positions the R-alkyl substituent in the
pseudoequatorial position of the chair conformer.16 Addition of
the nucleophile to the anti-Felkin17 diastereoface opposite the
R-alkyl group affords the 1,2-anti OH-Me relationship in the
adduct.18 The NMR study not withstanding, both half-chair and
boat transition state chelate geometries rationalize the sense of
asymmetric induction (eq 10).
A number of factors are responsible for determining which
mode of Lewis acid-substrate activation might be anticipated.
Such factors include the nature of the coordinating Lewis acid
(BF3‚OEt2 vs TiCl4) and the nature of the oxygen protecting
group, P (Bn vs t-BuMe2Si),13,14 and the reaction solvent (CH2-
Cl2 vs THF).15 The impact of many of these variables has been
highlighted by Keck in his study of the catalyzed addition of
(6) For a brief report on organocuprate additions, see: Still, W. C.;
Schneider, J. A. Tetrahedron Lett. 1980, 21, 1035-1038.
(7) Evans, D. A.; Dart, M. J.; Duffy, J. L.; Yang, M. J. J. Am. Chem.
Soc. 1996, 118, 4322-4343.
(8) (a) Cram D. J.; Abd Elhafez, F. A J. Am. Chem. Soc. 1952, 74, 5828-
5835. (b) Cram, D. J.; Kopecky, K. R. J. Am. Chem. Soc. 1959, 81, 2748-
2755. (c) Cram, D. J.; Leitereg, T. H. J. Am. Chem. Soc. 1968, 90, 4019-
4026.
(9) Several “open chain” models have been presented since Cram: (a)
Cornforth, J. W.; Cornforth, R. H.; Mathew, K. K. J. Chem. Soc. 1959,
112-127. (b) Karabatsos, G. J. J. Am. Chem. Soc. 1967, 89, 1367-1371.
(c) Cherest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1968, 2199-
2204. (d) Anh, N. T.; Eisenstein, O. NouV. J. Chem. 1977, 1, 61-70. (e)
Anh, N. T. Top. Curr. Chem. 1980, 88, 145-162. (f) For an excellent review
of Cram’s rule see ref 1a.
(10) For reviews of Lewis acid-promoted reactions, see: (a) Enolsi-
lanes: Gennari, C. In ComprehensiVe Organic Synthesis: Additions to C-X
π-Bonds Part 2; Trost, B. M., Fleming, I., Heathcock, C. H., Eds.; Pergamon
Press: New York 1991; Chapter 2.4. (b) Allylsilanes and allylstannanes:
Fleming, I. In ComprehensiVe Organic Synthesis: Additions to C-X
π-Bonds Part 2; Trost, B. M., Fleming, I., Heathcock, C. H., Eds.; Pergamon
Press: New York 1991; Chapter 2.2.
(13) (a) Overman, L. E.; McCready, R. J. Tetrahedron Lett. 1982, 23,
2355-2358. (b) Keck, G. E.; Castellino, S.; Wiley, M. R. J. Org. Chem.
1986, 51, 5478-5480. (c) Keck, G. E.; Andrus, M. B.; Castellino, S. J.
Am. Chem. Soc. 1989, 111, 8136-8141. (d) Reetz, M. T.; Hu¨llmann, M. J.
Chem. Soc., Chem. Commun. 1986, 1600-1602. (e) Bloch, R.; Gilbert, L.;
Girard, C. Tetrahedron Lett. 1988, 29, 1021-1024. (f) Keck, G. E.; Palani,
A.; McHardy, S. F. J. Org. Chem. 1994, 59, 3113-3122. (g) Crimmins,
M. T.; Rafferty, S. W. Tetrahedron Lett. 1996, 37, 5649-5652. (h) Frye,
S. V.; Eliel, E. L. Tetrahedron Lett. 1986, 28, 3223-3226. (i) Frye, S. V.;
Eliel, E. L. J. Am. Chem. Soc. 1988, 110, 484-489. (j) Ukaji, Y.; Kanda,
H.; Yamamoto, K.; Fujisawa, T. Chem. Lett. 1990, 597-600. For evidence
supporting chelation of an OTBS group, see: (k) Chen, X.; Hortelano, R.
R.; Eliel, E. L.; Frye, S. V. J. Am. Chem. Soc. 1992, 114, 1778-1784. (l)
Williard, M. J.; Hintze, M. J. J. Am. Chem. Soc. 1987, 109, 5539-5541.
(14) (a) Sujishi, S.; Witz, S. J. Am. Chem. Soc. 1954, 76, 4631-4636.
(b) Shea, K. J.; Gobeille, R.; Bramblett, J.; Thompson, E. J. Am. Chem.
Soc. 1978, 100, 1611-1613. (c) West, R.; Wilson, L. S.; Powell, D. L. J.
Organomet. Chem. 1979, 178, 5-9. (d) Kahn, S. D.; Keck, G. E.; Hehre,
W. J. Tetrahedron Lett. 1987, 28, 279-280. (e) Shambayati, S.; Blake, J.
F.; Wierschke, S. G.; Jorgensen, W. J.; Schreiber, S. L. J. Am. Chem. Soc.
1990, 112, 697-703.
(11) (a) Eliel E. L. In Asymmetric Synthesis; Morrison, J. D., Ed.;
Academic Press: New York, 1983; Vol. 2, Chapter 5, pp 125-155. (b)
Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1984, 23, 556-569. (c) Reference
1. (d) Heathcock, C. H.; Kiyooka, S.; Blumenkopf, T. A. J. Org. Chem.
1984, 49, 4214-4223.
(12) For the first direct evidence for chelate control see: Reetz, M.;
Hu¨llmann, M.; Seitz, T. Angew. Chem., Int. Ed. Engl. 1987, 26, 477-479.
(15) Keck, G. E.; Boden, E. P. Tetrahedron Lett. 1984, 25, 265-268.
(16) Keck, G. E.; Castellino, S. J. Am. Chem. Soc. 1986, 108, 3847-
3849.

10842 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Scheme 1
Scheme 2
Under traditionally favorable chelating reaction conditions
(TiCl4, benzyl protecting group), good levels of chelation control
can be obtained from the R-stereocenter of a â-alkoxy aldehyde
(eq 12). If chelate organization is to be suppressed, tert-
butyldimethylsilyl (TBS) or related protecting groups are
employed (cf. eq 7). With such substrates, a return to Felkin
control occurs despite use of potentially chelating Lewis acids
(eq 11).19 Accordingly, a stereochemical analysis of these
reactions (eqs 11 and 12) provides circumstantial evidence for
the operational stereochemical control element for the addition
process. From these data, one may reasonably conclude that
the TBS-protected aldehyde 2 is not chelate-activated by TiCl4.
Spectroscopic support for the lack of chelation for this substrate
with SnCl4 and MgBr2 has also been provided by Keck,
substantiating that hindered silicon protecting groups thwart
chelate control in Lewis acid-mediated reactions.20
â-Chelation: 1,3-Induction. Our open-chain 1,3-induction
model E21 and the corresponding chelate-controlled model F
are illustrated in Scheme 2 for â-alkoxy aldehydes substituted
in the â-position (eqs 13 and 14).22 In contrast to the previous
case (Scheme 1), the open-chain and chelation addition modes
cannot be distinguished by the stereochemical outcome of the
reaction as both control elements lead to the 1,3-anti-diol product
diastereomer. It seems reasonable that nucleophile addition
through either the boat conformation F1 or either of the half-
chair conformations F2 or F3 might be considered for the
chelated transition state (eq 14). Excellent anti diastereoselec-
tivity in these systems can be achieved under standard chelating
conditions (eqs 1521 and 1613b). Keck has provided spectroscopic
evidence that F3 is the preferred conformation for the TiCl4-
chelate when protecting group P is sterically more demanding
than a methyl group due to the destabilizing gauche P T R
interaction.13b The authors conclude that high reaction diaste-
reoselection requires reaction via this conformation. The critical
evidence upon which this conclusion rests is the direct correla-
tion of the size of the ether substituent with chelate-controlled
addition diastereoselection. In this study, stereoelectronic issues
were not raised (vide infra). Finally, it is evident that good 1,3-
anti induction is also possible where chelation is precluded by
the choice of Lewis acid (eq 17).20 Accordingly, a stereochem-
(17) The term “anti-Felkin” refers to the carbonyl diastereoface that is
disfavored according to the Felkin-Anh model for carbonyl π-facial
selectivity.8c-e The anti-Felkin adduct resulting from addition to the anti-
Felkin diastereoface can be recognized for all R-substituted aldehydes in
this study as that adduct diastereomer in which the aldehyde R-methyl group
and the new hydroxyl group are anti to one another when the product is
drawn with the carbon backbone extended. The anti-Felkin product is the
product of chelation control for all substrates in this study. The “Felkin”
adduct is the product with the 1,2-syn Me T OH relationship.
(18) (a) Kiyooka, S.; Heathcock, C. H. Tetrahedron Lett. 1983, 24, 4765.
(b) Reetz, M. T.; Kessler, K.; Jung, A. Tetrahedron Lett. 1984, 25, 729. (c)
Nakata, T.; Tani, Y.; Hatozaki, M.; Oishi, T. Chem. Pharm. Bull. 1984,
32, 1411-1415. (d) Burke, S. D.; Piscopio, A. D.; Marron, B. E.; Matulenko,
M. A.; Pan, G. Tetrahedron Lett. 1991, 32, 857-858.
(21) Evans, D. A.; Duffy, J. L.; Dart, M. J. Tetrahedron Lett. 1994, 35,
8537-8540.
(22) For an early study on the addition of silcon nucleophiles to
â-alkoxyaldehydes see: Reetz, M. T.; Jung, A. J. Am. Chem. Soc. 1983,
105, 4833-4835.
(19) This study, Table 1.
(20) Keck, G. E.; Castellino, S. Tetrahedron Lett. 1987, 28, 281-284.

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10843
Scheme 3
Scheme 4
Chart 1
ical analysis of these reactions provides no evidence of the
operational stereochemical control element.
Stereoelectronic Considerations. The high diastereoselection
observed for the chelate-controlled addition might be attributed
solely to steric factors; however, stereoelectronic control ele-
ments cannot be ignored. While stereoelectronic factors for these
addition reactions have not been systematically addressed, these
electronic effects may be extrapolated from the addition of
nucleophiles to six-membered cyclic oxo-carbenium ions (Scheme
3), which exhibit a conformational bias for nucleophilic attack
from the pseudoaxial carbonyl diastereoface (eqs 18 and 19).23
Recent examples provided by Woerpel document the strong bias
for axial attack by allyltrimethylsilane on 4-alkyl-substituted
oxocarbenium ions (eq 18). The analysis of the stereochemical
outcome for the allylsilane addition to the 3-methyl analogue
is more complicated (eqs 19a,b). Woerpel suggests that axial
nucleophile addition syn to the methyl substituent, while favored
stereoelectronically, is disfavored sterically (eq 19a, Path a).
Nucleophilic addition to the oxo-carbenium diastereoface anti
to the methyl substituent, while favored sterically, is forced to
proceed via a twist-boat transition state (eq 19b, Path b). In the
present instance, the poorly diastereoselective outcome suggests
that steric and electronic factors are closely balanced. While
these cases implicate a stereoelectronic component in chelate-
controlled additions, the longer metal-oxygen bond lengths in
the Lewis acid-chelated transition structures afford greater
conformational flexibility, including the possible intervention
of lower energy boat geometries (vide infra). Such geometry
changes will therefore necessarily modify the stereochemical
trends documented by the oxo-carbenium ion analogies.
Table 1.a Lewis Acid-Promoted Aldol Reactions of R-Substituted
Aldehydes 1 and 2 (Eq 20)
a Reactions were carried out in CH2Cl2 at -78 °C for 20 min. Ratios
were determined by GLC analysis after silylation (TMS-imidazole) or
acylation (Ac2O) of the unpurified reaction mixtures. Yields are reported
for the mixture of diastereomers. b Reactions were run with 1.0 equiv
of BF3‚OEt2, SnCl4, and TiCl4 and 2.5 equiv of Me2AlCl and MeAlCl2.
Use of 2.5 equiv of BF3‚OEt2, SnCl4, and TiCl4 had no effect on
diastereoselectivity. c This reaction was run at -90 °C. At -78 °C,
96:04 (32).
Model Reactions. The reliability of the data reflecting
chelation control in this study hinges upon establishing ap-
propriate model aldol reactions. We selected the set of simple
â-alkoxy aldehydes 1-8 (Chart 1) as substrates for study. The
benzyl (Bn) and tert-butyldimethylsilyl (TBS) protected â-ox-
ygen substituents were chosen as representative alkyl and silyl
protecting groups, and the isopropyl â-carbon substituent was
selected to model the steric environment of polypropionate
aldehydes. Addition reactions to substrates 1-4 were defined
to assay the facial bias afforded by the aldehyde R and â
stereocenters independently under chelating conditions. The
merged impact of both stereocenters on the addition process
will then be evaluated with aldehydes 5-8. Chelation-controlled
reactions were carried out with TiCl4, SnCl4, Me2AlCl, and
MeAlCl2 as chelating Lewis acids. In addition, each reaction
was run with BF3‚OEt2 to allow for direct comparison with an
unambiguously Felkin-controlled reaction.1a
Results and Discussion
The objectives in this study are to develop a consistent set
of data that might reveal diastereoselectivity trends in the
chelate-controlled addition reactions of nucleophiles to syn and
anti aldehyde chelates I and J (Scheme 4). In this analysis, the
individual contributions from the R and â stereocenters will be
documented from the diastereoselectivity trends observed for
the chelate-controlled additions of the monosubstituted chelates
G and H. In support of model predictions, the syn relationship
in chelate I is predicted to be reinforcing while the anti chelate
diastereomer J is predicted to be opposing. A parallel theme in
this study has been the documentation of the “super-chelating”
capabilities of Me2AlCl and MeAlCl2, Lewis acids that will
chelate with virtually any alkoxy substituent. Our results are
detailed in the following discussion.
1,2-Induction. Studies began with an examination of 1,2-
induction in aldol addition reactions of R-methyl-substituted
aldehydes 1 and 2 (Table 1, eq 20).24 The analysis of a Lewis
acid-catalyzed nucleophilic addition to aldehydes 1 and 2 may
be readily achieved since Felkin control leads to the syn product
diastereomer while chelation control affords the analogous anti
(23) (a) Romero, J. A. C.; Tabacco, S. A.; Woerpel, K. A. J. Am. Chem.
Soc. 2000, 122, 168-169. (b) Lewis, M. D.; Cha, J. K.; Kishi, Y. J. Am.
Chem. Soc. 1982, 104, 4976-4978.
(24) Unambiguous stereochemical proofs for all product diastereomers
are detailed in the Supporting Information.

10844 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Scheme 5
a Reactions were carried out in CH2Cl2 at -78 °C for 20 min. Ratios were determined by GLC analysis after silylation (TMS-imidazole) of the
unpurified reaction mixtures.
Table 2.a Lewis Acid-Promoted Aldol Reactions of â-Alkoxy
Aldehydes 3 and 4 (Eq 23)
product diastereomer (cf. Scheme 1). Felkin-controlled addition
to either aldehyde (BF3‚OEt2, entry A) afforded the 1,2-syn
adduct with good to high stereoselectivity. For reactions of 1,
the chelating Lewis acids all showed the expected increased
proportion of chelation-product 9 with the TiCl4-promoted
addition exhibiting the highest level of chelation control (97:3
entry C) followed by Me2AlCl (90:10 entry D). Surprisingly,
SnCl4 showed the lowest propensity for chelate control with
aldehyde 1 (50:50 entry B). The unique chelating ability of Me2-
AlCl and MeAlCl2 becomes apparent when aldol additions to
the TBS-protected aldehyde 2 were carried out. While TiCl4
and SnCl4 exhibited good Felkin control (93:7), both of the
aluminum halide based Lewis acids retained the capacity for
chelation even with the OTBS moiety. The high chelate
a See Table 1, footnote a. b See Table 1, footnote b.
selectivity afforded by Me2AlCl (97:3, entry D) identifies the
unique role that this Lewis acid can play in this and related
addition reactions.
this type of ligand metathesis is precedented for aluminum halide
complexes,25 it has not been a widely recognized strategy for
generating highly Lewis acidic metal complexes.26 We first
encountered the highly chelating nature of Me2AlCl in our
imide-based Diels-Alder investigations some years ago (eq 1).4b
In this study a dramatic change in dienophilic reactivity and
selectivity accompanied an increase in the Lewis acid:dienophile
stoichiometry. Castellino has reported spectroscopic studies
supporting the proposed dienophile-Lewis acid complex il-
lustrated in eq 1.27 The trends associated with Me2AlCl are also
observed with MeAlCl2 in the enolsilane aldol reactions
investigated here.
1,3-Induction. The stereochemical outcome of the Lewis
acid-catalyzed nucleophilic addition to aldehydes 3 and 4 may
not be readily interpreted since both open-chain and chelation
control lead to the same product diastereomer (cf. Scheme 2,
eqs 13 and 14). In accord with expectation, reactions of both
benzyl- and TBS-protected aldehydes 3 and 4 selectively
afforded the anti product diastereomer for all the chelating Lewis
acids, with the lone exception of SnCl4, which afforded minimal
selectivity (Table 2). We and others have amply documented
that the 1,3-anti product stereochemistry results from nonchelate
controlled addition to â-alkoxy aldehydes (entry A).28 While
the origin of the stereochemical control element cannot be
Figure 1. PM3 minimized cationic aluminum chelates of aldehyde 1
in boat and half-chair conformations.
The chelating ability of Me2AlCl is dependent on the Lewis
acid:substrate stoichiometry (Scheme 5). At low ratios of Lewis
acid, the addition process exhibits dominant Felkin selectivity
(eq 22a). As the relative amount of Me2AlCl is increased, the
carbonyl face selectivity reverses and the process becomes
highly chelate selective (eq 22b). The reversal in aldehyde face
selectivity is consistent with the Me2AlCl induced conversion
of complex K to the chelated cationic boat complex L or its
less stable half-chair conformer (vide infra, cf. Figure 1). While
(25) (a) Lehmkuhl, H.; Kobs, H.-D. Liebigs Ann. Chem. 1968, 719, 11-
19. (b) Reference 4a.
(26) (a) Renslo, A. R.; Danheiser, R. L. J. Org. Chem. 1998, 63, 7840-
7850. (b) Midland, M. M.; Koops, R. W. J. Org. Chem. 1992, 57, 1158-
1161.
(27) (a) Castellino, S.; Dwight, W. J. J. Am. Chem. Soc. 1993, 115,
2986-2987. (b) Castellino, S. J. Org. Chem. 1990, 55, 5197-5200.
(28) (a) Reference 20. (b) Reetz, M. T.; Kesseler, K.; Jung, A.
Tetrahedron Lett. 1984, 25, 729-732. (c) Yamamoto, Y.; Komatsu, T.;
Maruyama, K. J. Organomet. Chem. 1985, 285, 31-42.

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10845
Table 3. Dependence of the Selectivity of Felkin-Controlled
Reactions on Nucleophile Size (Eq 26)
Scheme 6
a Reactions were run in toluene at -78 °C.
assigned for the chelating Lewis acids on the basis of the product
stereochemical analysis, it is reasonable to conclude that the
BF3‚OEt2-mediated addition is representative of the stereo-
chemical control afforded by a nonchelate-controlled addition.
It is noteworthy that the diastereoselectivity for nearly all of
the ostensibly chelation-controlled reactions is surprisingly
similar to the nonchelate-controlled additions. The most dia-
stereoselective addition was that observed for the MeAlCl2-
promoted addition to the TBS-protected aldehyde 4.
Merged 1,2- and 1,3-Asymmetric Induction. In our previ-
ous work on this topic, we carried out a detailed study of the
BF3‚OEt2-promoted additions to diastereomeric aldehyde pairs
5,6 and 7,8 to determine whether any trends might be established
with regard to the relative contributions of the individual R and
â stereocenters on the stereochemical outcome of the Felkin-
controlled reactions.29 From the independent analysis of R and
â aldehyde stereocenters in Lewis acid-induced enolsilane
addition under chelating and nonchelating conditions, the
following trends are noted: For the aldehyde R-stereocenter, it
is evident from our data that the 1,2-syn (OH T Me) relationship
is favored under nonchelating conditions while the 1,2-anti (OH
T Me) relationship is favored under chelating conditions (Table
1). For the â-stereocenter, nucleophilic addition favors the
formation of the 1,3-anti (OH T OR) relationship under both
chelating and nonchelating conditions (Table 2).
resident stereocenters in syn aldehydes 5 and 6 should be
reinforcing (eq 25b) under chelating conditions. As eqs 27 and
28 imply, the R and â stereocenters in the syn aldehyde
diastereomer family mutually reinforce the chelate-mediated
addition process thus favoring the anti-Felkin/1,3-anti adduct
(eq 27). Conversely, the resident stereocenters in anti aldehydes
7 and 8 should be nonreinforcing (eq 25a) under chelating
conditions. The chelate geometries depicted below are illustrated
for a generic metal in one of the half-chair conformations.
Semiempirical calculations (PM3) carried out on the cationic
dimethylaluminum chelates suggest that the boat chelates are
lower in energy than their chair counterparts (vide infra, cf.
Figure 3).
Nonchelating Lewis Acids. In our integrated model for
Felkin-controlled additions with this family of substrates, the
data lead to the conclusion that the resident stereocenters in
anti aldehydes 7 and 8 both support addition to the same
activated aldehyde diastereoface to afford the 1,2-syn/1,3-anti
adduct diastereomer (Scheme 6, eq 24a). Accordingly, the anti
aldehyde diastereomeric relationship was identified as stereo-
reinforcing under nonchelating conditions. This addition process
is stereoregular for all enolsilane structures. Conversely, the
resident stereocenters in syn aldehyde diastereomers 5 and 6
were identified as nonreinforcing under nonchelating conditions
(Scheme 6, eq 24b). With sterically demanding enolsilanes, the
R stereocenter and its associated steric effects are the dominant
stereocontrol element where the preference for the 1,2-syn (OH
T Me) relationship overrides the 1,3-anti (OH T OR) electronic
bias imposed by the â-OR substituent. As the enolsilane steric
requirements diminish, t-Bu f Me (Table 3), the electrostatic
contributions of the â-OR substituent in 5 and 6 become
dominant and a reversal in face selectivity is noted.29
To properly assay for the mode of activation in the Lewis
acid-activated additions to the syn aldehydes 5 and 6, we have
carefully chosen pinacolone enolsilane as the participating
nucleophile. With this enolsilane, Felkin- and chelation-
controlled additions lead to opposite product diastereomers
(compare Table 3, entry A with eq 27). In contrast, less sterically
demanding enolsilanes afford the same product diastereomer
independent of the mode of activation (compare Table 3, entry
C with eq 27). The reaction of syn aldehyde 5 and the
pinacolone-derived enolsilane with BF3‚OEt2 (nonchelation)
afforded the expected Felkin/1,3-syn aldol adduct 23 with high
(95:5) selectivity (Table 4, entry A) while the same reaction
promoted by Me2AlCl (chelation) afforded chelate-mediated
adduct 22 in 99:1 selectivity (Table 4, entry D). These data
demonstrate the dramatic reversal of stereochemistry in compar-
ing the two modes of activation. The cationic aluminum Lewis
acid affords exceptional chelation control as contributed by the
reinforcing stereocontrol elements in 5. The same trend is also
observed for the Me2AlCl- and MeAlCl2-catalyzed additions
with the silyl-protected syn aldehyde 6 (entries D and E, Table
Chelating Lewis acids. In the present investigation, we have
evaluated the complementary chelate-controlled addition reac-
tions of syn aldehydes 5 and 6 and the anti diastereomers 7
and 8 to formally establish the trends in face selectivity for
enolsilane nucleophiles. By inspection, it was predicted that the
(29) (a) Evans, D. A.; Dart, M. J.; Duffy, J. L.; Yang, M. J.; Livingston,
A. B. J. Am. Chem. Soc. 1995, 117, 6619-6620. (b) Evans, D. A.; Dart,
M. J.; Duffy, J. L.; Yang, M. J. J. Am. Chem. Soc. 1996, 118, 4322-4343.

10846 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Figure 2. PM3 minimized cationic aluminum chelates of aldehyde 3 and the corresponding â-OMe aldehyde analogue.
Figure 3. PM3 minimized cationic aluminum chelates of syn-substituted aldehydes 5 and 6 and anti-substituted aldehydes 7 and 8. The tert-
butyldimethylsilyl (TBS) protecting group in 6 and 8 has been substituted by a trimethylsilyl (TMS) group to simplify the calculations.
Table 4.a Lewis Acid-Promoted Aldol Reactions of Syn-Substituted
R-Methyl-â-alkoxy Aldehydes 5 and 6 (Eq 29)
4). In each case a dramatic turnover in stereochemistry is seen,
∼5:95 Felkin f >96:4 chelation, upon going from single-point
activation (BF3‚OEt2) to chelate activation. It is noteworthy that
the degree of selectivity in chelation-controlled reactions of 5
and 6 is enhanced relative to that in reactions of aldehydes 1-4,
which bear only one stereocenter (cf. Tables 1 and 2). The trends
established in Table 4 for the Me2AlCl-mediated addition
reactions also hold for less hindered enolsilanes (Table 5, eq
30).30
5 P ) Bn
22:23 (%)
6 P ) TBS
24:25 (%)
When the catalyzed pinacolone enolsilane additions with
aldehydes 5 and 6 (eq 29) were employed to assay the chelating
ability of other common chelating Lewis acids, it was found
that both TiCl4 and SnCl4 exhibit little to no chelating capability.
While TiCl4 does effect excellent levels of chelate organization
entry
Lewis acidb
A
B
C
D
E
BF3‚OEt2
SnCl4
TiCl4
Me2AlCl
MeAlCl2
05:95 (78)
05:95 (32)
38:62 (22)
99:01 (73)
99:01 (81)
04:96 (91)
01:99 (41)
02:98 (71)
97:03 (51)
96:04 (71)
(30) The reaction of acetone enolsilane and 6 (Table 5, entry A) appears
to proceed by addition to the nonchelated 1:1 complex of 6 with Me2AlCl.
If only 1 equiv of Me2AlCl is used, nearly identical results are observed.
Additionally, these results are consistent with the same reaction mediated
by BF3‚OEt2, which afforded a 42:58 anti-Felkin/Felkin ratio of aldol
adducts.
a See Table 1, footnote a. b See Table 1, footnote b.
in additions to less hindered substrates (no â-substituent, Table
1), we suggest that the more pronounced steric congestion in

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10847
Table 5.a Lewis Acid-Promoted Aldol Reactions of Syn-Substituted
R-Methyl-â-alkoxy Aldehydes 19 and 6 with Various Enolsilanes
(Eq 30)
a See Table 1, footnote a.
It is instructive to revisit some of the catalyzed addition
reactions of 1,2-syn disubstituted aldehydes reported in the
literature. For example, diastereoselective allylstannane and
allylsilane additions (eqs 33 and 34) have been reported during
the course of studies by Keck (rhizoxin)31 and Panek (myco-
trienin I).32 Both authors have suggested that these addition
reactions are chelate-controlled processes based on the Lewis
acid employed (TiCl4) and the stereochemical outcome of the
addition; however, the analysis, based on stereochemical
outcome alone, is deceptive for syn aldehydes. We have just
established that Felkin and chelate-controlled additions with
sterically nondemanding nucleophiles afford the same product
diastereomer. For example, the analogous BF3‚OEt2 catalyzed
allylstannane addition also affords the “chelate” product (eq
35).20 Accordingly, our studies call into question the intervention
of chelation control in eqs 33 and 34. Furthermore, from the
data presented in the preceding discussion, it has been demon-
strated (Table 4) that TiCl4 is not a good chelating Lewis acid
for this family of aldehyde substrates. Hence we conclude that
the additions illustrated in eqs 33 and 34, in contrast to the
authors’ suggestions, are likely not to be chelate-controlled
processes.
Table 6. Addition of Allylsilanes to Aldehyde 6 (Eq 31)a
a See Table 1, footnote a. b See Table 1, footnote b.
Table 7. Evaluation of Silicon Protecting Groups with Aldehydes
32-34 and 6 (Eq 32)a
Inspection of the chelate of the anti-substituted aldehyde
reveals the nonreinforcing nature of this stereochemical array.
The chelated intermediate disposes the R and â substituents on
opposite sides of the coordinated carbonyl. Nucleophile ap-
proach from the anti-Felkin face of the carbonyl encounters
steric encumbrance from the â-alkyl substituent (eq 36) while
the R-methyl group hinders nucleophilic addition to the Felkin
carbonyl diastereoface (eq 37). Ultimately, addition to the anti-
substituted aldehyde should result in diminished stereoselectivity
under chelate control. Semiempirical calculations (PM3) carried
out on the cationic dimethylaluminum chelates suggest that the
boat chelates are lower in energy than their chair counterparts
(vide infra, cf. Figure 3).
Lewis 32 P ) TMS 33 P ) TES 6 P ) TBS 34 P ) TIPS
entry acidb
35:36
35:36
35:36
35:36
A
D
E
BF3‚OEt2
Me2AlCl
MeAlCl2
95:05
02:98
02:98
98:02
02:98
10:90
96:04
03:97
04:96
97:03
35:65
38:62
a See Table 1, footnote a. b See Table 1, footnote b.
the chelates of aldehydes 5 and 6 thwart chelate organization
in these more hindered aldehyde substrates.
Other nucleophiles such as allylsilanes and allylstannanes are
also accommodated by the aluminum Lewis acids (Table 6, eq
31). Good to excellent chelation control is observed for all
reactions of silyl-protected aldehyde 6. While these studies have
emphasized the stereochemical elements of the addition process,
these reactions perform successfully at preparative scale with
no degradation of yield or stereoselectivity demonstrating the
synthetic utility of these transformations. The generality of
silyloxy group chelation with Me2AlCl and MeAlCl2 was also
evaluated with aldehydes 32-34 (Table 7, eq 32). Excellent
levels of chelation control are maintained with silyl groups
sterically smaller than TBS; however, the chelating ability of
the aluminum Lewis acids is partially curtailed in the limit by
the sterically demanding triisopropylsilyl (TIPS) group in 34.
(31) Keck, G. E.; Savin, K. A.; Weglartz, M. A.; Cressman, E. N. K.
Tetrahedron Lett. 1996, 37, 3291-3294.
(32) Masse, C. E.; Yang, M.; Solomon, J.; Panek, J. S. Am. Chem. Soc.
1998, 120, 4123-4134.

10848 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Table 8.a Aldol Reactions of Anti-Substituted Aldehydes 7 and 8
(Eq 38)
may be the preferred conformation of dimethylaluminum
chelates of â-alkoxy aldehydes.
Calculations on the dimethylaluminum chelates of â-substi-
tuted â-alkoxy aldehyde 3 also signal that boat conformations
are preferred over their chair counterparts.35 Within the boat
conformation manifold, the disposition of the â-isopropyl
substituent (pseudoaxial vs pseudoequatorial) is examined
through structures 48-51 (Figure 2). There is a modest
preference, 0.9 kcal/mol, for the pseudoaxial â-substituent in
the chelate of 3 (48 vs 49). Presumably the vicinal gauche
interaction between the oxygen protecting group and the
isopropyl substituent (Bn T CHMe2) in chelate 49 is destabiliz-
ing thus forcing the isopropyl substituent into a pseudoaxial
orientation. The size of the oxygen protecting group influences
the relative energies of the boat conformers. For example, the
pseudoaxial isopropyl boat conformer 50, in which the â-ben-
zyloxy substituent has been replaced with a â-methoxy sub-
stituent, is now only 0.5 kcal/mol more stable than the
pseudoequatorial isopropyl boat conformer 51.36 This trend
supports the premise that the gauche interaction between the
oxygen protecting group and the isopropyl substituent (Bn T
CHMe2) in chelate 49 is more strongly destabilizing than the
analogous interaction (Me T CHMe2) in chelate 51. Diaste-
reoselection trends in the chelate-mediated allylstannane addition
to â-alkoxy aldehydes document that the steric requirements
of the â-alkoxy substituent directly correlate with reaction
diastereoselection, with the larger alkoxy residues being more
diastereoselective (eq 40).13b The suggestion, supported by the
calculations of Figure 3, is that the pseudoaxial isopropyl group
is responsible for good chelation control with these type of
substrates.
a See Table 1, footnote a. b See Table 1, footnote b.
Table 9.a Aldol Reactions of Anti-Substituted Aldehydes 41 and 8
with Representative Enolsilanes (Eq 39)
a See Table 1, footnote a.
This prediction is borne out experimentally with Me2AlCl
and MeAlCl2 as chelating Lewis acids (Table 8, eq 38). Reaction
of the anti aldehydes 7 and 8 under chelating conditions provides
low to moderate stereoselection (entries B and C). These data
substantiate that modest reaction diastereoselection is to be
anticipated for the nonreinforcing anti aldehyde diastereomer
under chelate-controlled substrate activation. In contrast, the anti
diastereomeric relationship is reinforcing under single point
Lewis acid activation. Again, this point is confirmed by the
exceptional Felkin selectivity observed with BF3‚OEt2 (99:1).
Modest stereoselectivity is also observed across the range of
substituted enolsilanes with Me2AlCl (Table 9, eq 39). Based
on the weight of evidence, it is presumed that chelate control
is operating in all of these addition processes.
Chelate Models. A series of semiempirical calculations
(PM3) were carried out to probe the conformations of the
putative cationic aluminum chelates involved in the above
reactions.33 While one cannot make any definitive statements
about transition state geometries from these ground-state
calculations, some inferences may be made on probable reacting
geometries. Geometry optimization of the Me2Al(+) chelate of
R-methyl substituted aldehyde 1 from different starting geom-
etries located the two low-energy boat and half-chair conforma-
tions (Figure 1). These calculations signal that the boat
conformer 46 is more stable than the half-chair conformer 47
by approximately 3 kcal/mol.34 By inspection, the observed
sense of asymmetric induction may be rationalized from either
chelate conformer, but the implication is that boat geometries
Computationally generated structures for the Me2Al(+)
chelates of syn and anti aldehydes 5-8 are shown in Figure 3.
The syn-substituted chelates 52 and 53 incorporate the best
features of the above models for 1,2 and 1,3 induction, and
mutually reinforcing stereocenters are evident. The lowest
energy conformations are the illustrated boat geometries with
a pseudoaxial â-isopropyl group and a pseudoequatorial R-
methyl group, which both direct nucleophilic approach to the
anti-Felkin aldehyde diastereoface. While our original stereo-
chemical predictions were based upon consideration of chair-
like models, these boat structures do not alter those predictions.
For the anti chelates the more energetically favorable boat
conformations are the illustrated ones in which both the R and
â substituents reside equatorially.37 These structures closely
(35) PM3 optimizations lead to the following relative energies: boat-
[Al(Me)2(3)]+, Erel ) 0 kcal/mol; half-chair-[Al(Me)2(3)]+, Erel ) +1.9
kcal/mol.
(36) Calculations on the TMS-protected aldehyde chelates corresponding
to 48 and 49 showed only a small (0.3 kcal/mol) energy difference. The
very bulky silyl group equalizes the energy differences between the two
boat conformers, which are due to steric interactions between the protecting
group and the pseudoaxial or pseudoequatorial isopropyl group.
(37) The trans-diaxial boat conformations were also examined. These
conformation were higher in energy than the illustrated diequatorial ones
apparently as a result of the severe 1,4-diaxial repulsion between the
R-methyl group and the axial methyl group on aluminum. Erel(diaxial anti-
Bn chelate) ) +7.7 kcal/mol. Erel(diaxial anti-TMS chelate) ) +9.6 kcal/
mol.
(33) Stewart, J. J. P. J. Comput. Chem. 1989, 10, 209-220. Calculations
were performed within the SPARTAN computational platform on SGI
Indigo workstations: SPARTAN Version 5.0, Wavefunction, Inc.; Irvine,
CA.
(34) PM3 calculations on the corresponding TMS-protected aldehyde
show the same trend. The boat conformer is 2.7 kcal/mol lower in energy
than the half-chair.

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10849
aldehyde is converted to the middle spectrum, which is assigned
to the single-point activated 6‚Me2AlCl complex 56. There is a
clear, significant downfield shift of the aldehyde proton indicat-
ing carbonyl complexation. There are downfield shifts of the
aliphatic protons in the 1.5-4 ppm region as well, with the
degree of shift commensurate with the distance of the protons
from the coordinated aldehyde. A new upfield 6-proton singlet
at -0.9 ppm due to the methyl groups on aluminum has
appeared. At any number of equivalents of Me2AlCl above 1.0,
1
however, no change in the H spectrum is observed with the
exception of the growth of the resonance due to uncomplexed
excess Me2AlCl at -0.5 ppm (top spectrum, Figure 4). Although
57 was not observed, the existence of the cationic chelate is
not entirely ruled out as one may still postulate that the chelate
is an equilibrium intermediate present in a low concentration
below the sensitivity threshold of the NMR spectrometer.
It was thought that a surrogate for the aldehyde carbonyl
possessing slightly greater Lewis basicity may allow generation
of the cationic chelate in observable quantities. Toward this end,
methyl ketone 58 was prepared. Salient portions of the 1H and
13C NMR spectra of methyl ketone 58 with Me2AlCl are
reproduced in Figure 5. Again, the bottom spectra are those of
the uncomplexed substrate prior to addition of Lewis acid at
-70 °C. At 1.0 equiv of Me2AlCl, complete conversion to a
single new species is again observed (middle spectra). This is
assigned as the 1:1 complex 59. The most significant downfield
Figure 4. 1H NMR spectra of aldehyde 6 with 0.0, 1.0, and 4.0 equiv
of Me2AlCl recorded at -70 °C.
resemble those found for the R-methyl chelates (Figure 1)
suggesting that at best, chelate-controlled additions to these
aldehydes may behave similarly to reactions of the R-methyl
aldehydes. However, the stereochemical results indicate that the
â-alkyl group is not a passive substituent as these boat chelates
may suggest. In fact, the â-substituent is the dominant control
element in several reactions of the anti-substituted aldehydes
in which the Felkin/1,3-anti product diastereomer is seen to
predominate (cf. Tables 8 and 9). From this analysis no clear
trends are present for prediction of stereocontrol for chelate-
controlled reactions in the nonreinforcing scenario. For the most
part, diminished stereoselectivity is to be expected in these
reactions.
NMR Studies and Mechanism of OTBS Chelation. Al-
though the evidence in support of chelation of OTBS groups in
aldehydes 2, 4, and 6 with Me2AlCl or MeAlCl2 as Lewis acids
is strong (Tables 1, 2, and 4), this stereochemical evidence is
indirect at best. For lack of further evidence, the results may
be attributed to anomalous cases of anti-Felkin selective
nonchelation-mediated addition reactions. Since Lewis acid
coordination of silyl-bearing oxygen groups has been a conten-
tious issue for some time,12,13 it was appropriate to seek more
direct evidence for the proposed chelation. Earlier NMR studies
have shown the lack of chelation of â-OTBS-substituted
aldehydes with SnCl4 and MgBr2 as Lewis acids.20 Our
stereochemical results with TiCl4 and SnCl4 continue to support
this trend, while the behavior of Me2AlCl and MeAlCl2 was
markedly different for identical reactions of OTBS-substituted
aldehydes. The following results from low-temperature NMR
complexation studies indicate that the proposed cationic alu-
minum chelates can be observed for certain â-alkoxy carbonyl
substrates and implicate true chelates in the highly selective
aluminum-mediated reactions of â-OTBS aldehydes presented
above.
1
shifts in the H spectrum of 59 are those of the methyl group
and the methine proton which flank the ketone. The proton
neighboring the OTBS group (3.6 ppm) is nearly unaffected.
The 13C spectrum also clearly indicates single-point binding to
the ketone. A 23 ppm downfield shift of the ketone carbon is
observed with no change in the chemical shift of the carbon
bearing the OTBS substituent (77 ppm). Unlike for the aldehyde
6, a new species is produced as greater than 1 equiv of Me2-
AlCl is titrated into the NMR sample. Intermediate spectra
between 1.0 and 4.0 equiv show increasing ratios of the new
complex, indicating an equilibrium that is driven toward the
new complex by mass action with excess Lewis acid. The final
spectra recorded at 4.0 equiv of Me2AlCl show near complete
conversion to the new species, which is assigned as the cationic
1
complex 60 based upon downfield shifts in both the H and
13C spectra. Particularly striking are the shifts of the carbon
bearing the OTBS group, which now lies 11 ppm downfield
from the 1:1 complex, and the proton on this carbon, which
1
has moved 0.5 ppm farther downfield. In the H spectrum the
methyl ketone singlet undergoes an additional downfield shift,
and the diastereotopic methyl groups bound to silicon in the
TBS group undergo a 0.5 ppm downfield shift. The two new
methyl singlets located upfield at -0.5 ppm are assigned to the
methyl groups on the cationic aluminum center of 60. These
methyl groups have become diastereotopic from the 1:1 complex
as a result of the generation of the rigid six-membered chelate
ring. In identical complexation studies with TiCl4 and SnCl4,
unambiguous coordination to the carbonyl was observed, but
no significant chemical shift changes were seen for any protons
or carbons associated with the â-OTBS group.
Chelate-Controlled Reductions. In an effort to expand the
scope of the Al-mediated chelate-controlled addition reactions,
the reductions of â-alkoxy and â-silyloxy ketones were explored.
These reductions were expected to follow the same trends in
facial selectivity as has been observed with the enolsilane
nucleophiles in the preceding parts of this paper. Stereoselective
reductions of this type would provide access to valuable 1,3-
polyol synthons.38 The use of Me2AlCl as the chelating Lewis
Attempts were first made to observe the chelate of syn
aldehyde 6; however, even in the presence of up to 4.0 equiv
of Me2AlCl, the cationic 2:1 (Me2AlCl‚aldehyde) complex 57
could not be unambiguously discerned in the 1H NMR spectrum
(Figure 4). The bottom spectrum represents the free aldehyde
at -70 °C with no added Lewis acid. When 0.5 equiv of Me2-
AlCl is added, two species are observed in a 1:1 ratio, the
uncomplexed aldehyde and a new species whose NMR is the
middle spectrum. At exactly 1 equiv of Me2AlCl, all of the

10850 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Figure 5. 1H and 13C NMR spectra of methyl ketone 58 with 0.0, 1.0, and 4.0 equiv of Me2AlCl recorded at -70 °C. Significant downfield
1
chemical shift changes in the H NMR spectra are indicated by arrows.
Table 10.a Lewis Acid-Promoted Reductions of â-Alkoxy Ketones
61 and 62 (Eq 41)
Table 12.a Lewis Acid-Promoted Reductions of Anti-Substituted
R-Methyl-â-alkoxy Ketones 73 and 74 (Eq 43)
a Reactions were carried out in CH2Cl2 at -78 °C for 1 h. Ratios
1
a See Table 10, footnote a. b See Table 10, footnote b.
were determined by H NMR analysis (500 MHz) of the unpurified
reaction mixtures. Yields are reported for the mixture of diastereomers.
b Reactions were run with 1.0 equiv of BF3‚OEt2 and 2.5 equiv of
Me2AlCl and MeAlCl2. Use of 2.5 equiv of BF3‚OEt2 had no effect on
diastereoselectivity.
explored the use of nBu3SnH as a more nucleophilic hydride
source. The use of nBu3SnH at -78 °C in methylene chloride
proved to be optimal for the reduction of the selected ketone
substrates.40
Table 11.a Lewis Acid-Promoted Reductions of Syn-Substituted
R-Methyl-â-alkoxy Ketones 67 and 68 (Eq 42)
â-Chelation: 1,3-Asymmetric Induction. The stereochem-
ical outcome of the Lewis acid mediated reductions of ketones
61 and 62 was consistent with the Felkin and chelate transition
state models (cf. Scheme 2, eqs 13 and 14). In these cases, the
Felkin and chelate models lead to the same major product
diastereomer; however, the chelating Lewis acids proved far
more selective for the 1,3-syn product diastereomer than BF3‚
OEt2 (Table 10). It is presumed that the selectivity of the BF3‚
OEt2-mediated reduction is representative of the stereochemical
control that can be achieved by a purely Felkin-controlled
addition. The most diastereoselective reaction was that observed
for the Me2AlCl- or MeAlCl2-promoted reduction of the benzyl-
protected ketone 61.
a See Table 10, footnote a. b See Table 10, footnote b.
Merged 1,2- and 1,3-Asymmetric Induction. We have
evaluated the chelate-controlled reductions of syn ketones 67
and 68 (Table 11) and the anti diastereomers 73 and 74 (Table
12) to verify trends in facial selectivity in the case of a hydride
nucleophile. The transition state models for the syn ketones 67
and 68 (eqs 27 and 28) predict that the resident stereocenters
should be mutually reinforcing under chelate-controlled condi-
tions. As a result, reductions of the syn-disubstituted ketones
should strongly favor the 1,3-syn adduct (Table 11, eq 42).
Indeed, the reductions of syn ketones 67 and 68 promoted
by Me2AlCl (chelation) afforded the anti-Felkin/1,3-syn products
69 and 71 with high levels of selectivity (Table 11, entries A
and B). The data in Table 11 clearly illustrate the reversal of
acid in these reductions should allow for the incorporation of
both â-alkoxy and â-silyloxy substituents into the stereoselecive
reduction. We have selected a set of simple â-alkoxy ketones
as substrates for this study (61/62, Table 10; 67/68, Table 11;
73/74, Table 12).
Our initial investigation into these chelate-controlled reduc-
tions centered around the choice of a mild hydride source. A
number of trialkylsilanes were surveyed39 as hydride donors in
the reductions of ketones 61 and 62 (Table 10). However, all
of the trialkylsilanes proved to be ineffective at temperatures
ranging from -78 °C to room temperature. As such, we
(38) Oishi, T.; Nakata, T. Synthesis 1990, 635-645.
(39) Silanes surveyed included Et3SiH, Me2PhSiH, Ph3SiH, and
(TMS)3SiH.
(40) For further information on the optimal reaction conditions see the
Supporting Information.

Exceptional Chelating Ability of Me2AlCl and MeAlCl2
J. Am. Chem. Soc., Vol. 123, No. 44, 2001 10851
a
reported by Still and Schneider.6 This study revealed that lithium
dimethylcuprate adds to aldehyde 79 in a highly diastereose-
lective fashion (eq 45). The anti stereochemical outcome (cf.
Scheme 1, Table 1) provides compelling circumstantial evidence
that LiCuMe2 (in diethyl ether) is participating in a chelate-
controlled addition. In contrast, MeMgBr exhibits no diaste-
reoselectivity. While the stereochemical outcome of this addition
is not general for all organocuprates,6,41 this case is relevant to
the following two examples (eqs 46 and 47). We now know
that in syn aldehyde 80 the two stereocenters are mutually
reinforcing for a chelate-controlled addition while in the anti
aldehyde diastereomer 81 the two stereocenters are nonrein-
forcing (cf. eqs 3 and 4, Scheme 6). The observed trends in
diastereoselection support the premise that chelation control is
operating in all three cases.
Table 13. Survey of Hydride Reductions with â-Benzyloxy
Ketone 61 (Eq 44)
a Reactions were carried out in CH2Cl2 at -78 °C for 1 h. Ratios
were determined by H NMR analysis (500 MHz) of the unpurified
1
reaction mixtures. Yields are reported for the mixture of diastereomers.
b Reactions were carried out in THF at -78 °C for 1 h.
The four titanium tetrachloride-catalyzed additions illustrated
below (eqs 48-51) deserve comment. The highly diastereose-
lective allylsilane addition to aldehyde 82 has been reported by
Roush (eq 48)42 while the related addition has been carried out
by Panek (eq 49).43 On the surface, both of these reactions
appear to be chelate controlled on the basis of the stereochemical
outcome (eq 3, Table 6); however, syn aldehydes such as 82
and 83 afford the indicated stereochemical outcome even with
a nonchelating Lewis acid such as BF3‚OEt2 in allylmetal
additions (eq 35). As we have previously demonstrated, syn
aldehyde diastereomers such as 82 and 83 also deliver the same
observed stereochemical outcome from open-chain Felkin-like
additions with sterically “small” nucleophiles. In these additions,
the two stereocenters are nonreinforcing and the dominant
control element is the â-alkoxy substituent.29b If chelate control
were not operating in these additions, sterically more demanding
nucleophiles would exhibit a reVersal in aldehyde face selectiv-
ity as documented in Table 3 (eq 26). On the other hand, if
chelate organization was involved, sterically demanding nu-
cleophiles would maintain the same carbonyl face selectivity.
Such a case has been recently reported by Panek (eq 50).44 If
chelate control were not operational, the indicated hindered
enolsilane should afford the Felkin alcohol diastereomer. Since
the anti-Felkin (chelation) product is observed in this case, we
conclude that all three reactions appear to be chelate controlled.
Our own studies provide some indication that the maintenance
of chelation control in titanium tetrachloride-catalyzed additions
is not universal and is subject to subtle steric effects (see Table
4, eq 29). For example, the aldol addition to aldehyde 5 affords
principally the Felkin adduct (62:38, eq 51). By inspection,
aldehyde 5 carries the branched isopropyl substituent at the â
oxygen-bearing carbon while aldehydes 82-84 carry un-
branched substituents at this position. Further studies with
homogeneous families of nucleophiles might be useful in
pinning down the origin of these rather subtle steric effects.
Chelate Control in Synthetic Planning. There are several
conclusions that may be drawn with regard to the prediction of
reliably stereoselective addition reactions to R-alkyl-â-alkoxy
aldehydes from the data contained in this and our previous
paper29 on this topic. (A) The illustrated syn aldehyde diaste-
reomer, activated by a chelating Lewis acid, will undergo
predictable, stereoregular additions to afford the anti-Felkin/
1,3-anti product diastereomer (eq 52). (B) The illustrated anti
stereochemistry for the monodentate (BF3‚OEt2) and bidentate
(Me2AlCl, MeAlCl2) modes of activation. The cationic alumi-
num Lewis acids provide exceptional chelation control for the
silyl-protected syn ketone 68 due to the reinforcing stereochem-
ical elements. The diastereoselectivity in the chelate-controlled
reductions of 68 is enhanced relative to silyloxy ketone 62 which
bears only one stereocenter (cf. Table 10).
The transition state models outlined in eqs 36 and 37 are
relevant to the chelate-controlled reductions of the anti-
disubstituted ketones (73 and 74). In these cases, the R- and
â-substituents are nonreinforcing as the nucleophile will en-
counter steric encumbrance upon approach to either carbonyl
diastereoface. Therefore, the reductions of the anti-disubstituted
ketones are predicted to show diminished stereoselectivity under
chelation control. This prediction is confirmed experimentally
with both Me2AlCl and MeAlCl2 as the chelating Lewis acids
(Table 12, eq 43). The reduction of ketones 73 and 74 under
chelate conditions affords low levels of diastereoselection (Table
12, entries A and B). These data are consistent with the
diastereoselectivities observed for the addition of enolsilanes
to anti-substituted aldehydes (cf. Table 8). Reductions under
nonchelating conditions also afford equally poor levels of
stereoinduction (Table 12, entry C).
Synthetic Utility of the Al-Mediated Reductions. A com-
parison of the utility of these chelate-controlled reductions versus
other commonly employed reducing agents is provided in Table
13. The Me2AlCl-mediated reduction afforded the highest levels
of diastereoselectivity with the â-benzyloxy ketone to furnish
the syn adduct 63. Comparable yields were obtained for each
of the reducing agents. The reducing agent L-Selectride also
proved to be moderately selective (Table 13, entry D) and
provided a comparable yield of the chelate product.
Other Literature Examples. There are a significant number
of literature examples of diastereoselective addition to R-alkyl,
â-alkoxy aldehydes. Several early cases (eqs 45-47) were
(41) For additional cases see: Burke, S. D.; Piscopio, A. D.; Marron, B.
E.; Matulenko, M. A.; Pan, G. Tetrehedron Lett. 1991, 32, 857-858 and
references therein.
(42) Roush, W. R.; Marron, T. G.; Pfeifer, L. A. J. Org. Chem. 1997,
62, 474-478.
(43) Panek, J. S.; Berseis, R. T.; Celatka, C. A. J. Org. Chem. 1996, 61,
6494-6495.
(44) Zhu, B.; Panek, J. S. Org. Lett. 2000, 2, 2575-5578.

10852 J. Am. Chem. Soc., Vol. 123, No. 44, 2001
EVans et al.
Scheme 7
conditions necessary to favor the desired stereochemical out-
come. The 1,2-anti Me T OH relationship signals a transform
for chelation control while the analogous syn relationship calls
for nonchelation control.
aldehyde diastereomer, activated by a nonchelating Lewis acid,
will undergo stereoselective additions to afford the Felkin/1,3-
anti product diastereomer (eq 53). In both instances, the
stereochemical relationships are reinforcing under the stated type
of Lewis acid activation. In contrast, the anti aldehyde diaste-
reomer, reacting under the influence of chelate control, will
exhibit lower diastereoselection as will the syn aldehyde
diastereomer, reacting with nonchelating Lewis acids.
This analysis is, of course, dependent upon dominant stereo-
control emanating from the chirality resident in the aldehyde
fragment and considers only the prochirality of the trigonal
aldehyde carbon. An additional control element is introduced
if the enolate derivative is also prochiral, and furthermore, if
the enolate bears chirality, then a deeper level of analysis will
be necessary for the double stereodifferentiating aldol coupling
process.45 In the higher order analysis, the above inherent
stereochemical preferences of the aldehyde chirality remain the
same, but stereocontrol elements in the aldehyde may be
subjugated by stereochemical influences from the enolate
component.
Acknowledgment. Support for this research was provided
by the NSF, NIH (GM 33327-18), and the Merck Research
Laboratories. The authors wish to thank Dr. Sherry R. Chemler,
Department of Chemistry, Sloan-Kettering Institute for Cancer
Research, for her insightful comments supplied during the
critique of this paper.
Taken together, these stereochemical relationships outline a
basis for first-order analysis of stereocontrol in polypropionate
synthesis (Scheme 7). In the retrosynthetic analysis of a 1,3-
polyol chain, one key stereochemical element is the anti 1,3-
diol array because the reinforcing stereochemical conditions for
both chelation and nonchelation control favor generation of the
adduct bearing the anti 1,3-diol (eqs 52 and 53). Carbon-carbon
bond disconnections targeted around this stereochemical feature
should be stereoregular. In principle, the carbon-carbon bond
at either terminus of the anti 1,3-diol could be disconnected
separating the target molecule into an R-alkyl-â-alkoxy aldehyde
and an enolate derivative. Once an appropriate bond discon-
nection is chosen, the relative stereochemistry of the central
methyl-bearing carbon will determine the type of reaction
Supporting Information Available: Experimental details
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA011337J
(45) (a) Masamune, S.; Choy, W.; Peterson, J. S.; Sita, L. R. Angew.
Chem., Int. Ed. Engl. 1985, 24, 1-30. (b) Evans, D. A.; Dart, M. J.; Duffy,
J. L.; Rieger, D. L. J. Am. Chem. Soc. 1995, 117, 9073-9074. (c) Evans,
D. A.; Yang, M. G.; Dart, M. J.; Duffy, J. L.; Kim, A. S. J. Am. Chem.
Soc. 1995, 117, 9598-9599. (d) Evans, D. A.; Yang, M. G.; Dart, M. J.;
Duffy, J. L. Tetrehedron Lett. 1996, 37, 1957-1960. (e) Dart, M. J., Ph.D.
Thesis, Harvard University, May, 1995. (f) Duffy, J. L., Ph.D. Thesis,
Harvard University, December, 1995.