Substrate Control in Addition to Carbonyls: Felkin, Ahn and Friends

“These observations are, to my knowledge, the first definitive experimental evidence that further synthesis with asymmetric systems proceeds in an asymmetric manner. Although

this statement does not at all contradict theory, it by no means follows from it” – Emil Fischer, 1884

How do we set stereocentres?1. Get them from Nature (i.e. the chiral pool!)

2. Substrate Control:Cyclic (easier) or acyclic (harder)

Also, substrate directed reactions

3. Chiral Auxiliaries:e.g. Various oxazolidones, Oppolzer’s sultams, 8-phenylmenthol, tert-butanesulfinamides, SAMP/RAMP, pseudoephedrine etc.

4. Chiral Reagents:e.g. Alpine borane, ipc-boranes, BINAL-H, chiral bases…

5. Asymmetric Catalysis:e.g. SAD/SAE, Jacobsen, CBS, Noyori, Marshall, MacMillian…

Substrate Control?“Using the inherant stereochemical preferences of the substrate, due to existing

stereocentres, to perform subsequent reactions stereoselectively”

• This is generally easier in cyclic systems as it’s much easier to say where substituents will be at any one time!

• However, it is now know that acyclic systems can react in a highly selective fashion under certain conditions.

• Addition to carbonyl groups with nearby stereocentres is one of the best understood and most important scenarios and will be the focus of this talk.

• But first, a little history!

For substrate directable reactions see: Evans, Hoveyda and Fu, Chem. Rev., 1993, 93, 1307

Cyclic Versus Acyclic Substrates

O

O

Me

OHMe

Me

O

Me

OH

MeOH

Me

Me

(–)-6-deoxyerythronolide B

OH

Me

OBnMe

Me OH

O

Me

OH

MeOH

Mealkylation with MeI

enyne metathesis

macrolactonisation

J. Am. Chem. Soc., 2013, 135, 4223

Krische (2013):

R. B. Woodward (1981):

SS

OMOMBnO

H

OO

MeMe

SS

OBnO

H

OO

MeMe

Me MeMe

OMOMO OO

MeMe

SS

OBnO

H

OO

MeMe

H

(–)-erythromycin A

O

O

Me

OHHO

MeMe

O

Me

O

MeO

MeOH

Me

O

NMe2

MeHO

OMe

OH

MeO

Me

aldol thenoxidation

J. Am. Chem. Soc., 1981, 103, 3210

Cyclic Versus Acyclic Substrates

O

O

Me

OHMe

Me

O

Me

OH

MeOH

Me

Me

(–)-6-deoxyerythronolide B

OH

Me

OBnMe

Me OH

O

Me

OH

MeOH

Mealkylation with MeI

enyne metathesis

macrolactonisation

J. Am. Chem. Soc., 2013, 135, 4223

Krische (2013):

E. J. Corey (1978):

(±)-erythronolide B

O

O

Me

OHMe

Me

O

Me

OH

MeOH

MeOH

Me

Me Me

Me

OH

J. Am. Chem. Soc., 1979, 101, 7131

Carbonyl Addition: 1,2-Induction

“These observations are, to my knowledge, the first definitive experimental evidence that further synthesis with asymmetric systems proceeds in an asymmetric manner. Although

this statement does not at all contradict theory, it by no means follows from it” – Emil Fischer, 1884

O H

HO H

HO H

H OH

OHH

OH

HO H

HO H

H OH

OHH

OH

OHH

O OH

HCNthen hydrolysis

87%

heptonic acidsingle

diastereomer

D-(+)-mannose

• The first example of diastereoselective addition to a carbonyl under substrate control was reported by Emil Fischer in 1889:

Early Models

O

RRL

RM RSNu–

M Sometimes gives the right answer.´ Bad for systems with electronegative groups in α-position.´ Theoretically flawed (Nu approaches at 90º).

• Cornforth modified it, proposing that minimisation of dipoles was an important factor when electronegative groups (Rp) were present:

• Fisher came up with a model based on his sugar experiments, but it is now defunct.• Next came the Cram model (1952):

O

RRL/RP

RM RSNu–

RP="polar group"

M

Sometimes gives the right answer. Rationalises observations for electronegative groups.´ Still quite theoretically flawed.

• However, both these models have eclipsing conformations (RL/Rp clashes with R), which appears counterintuitive on steric grounds!

A Conformational Conundrum

• It was shown by Houk in the 1980’s that for acetaldehyde (and propene/butane) that the conformer with C-H and C= X ECLIPSING is more stable in the ground state:

H

HH

HX

is stabilising in this conformer!C-H *C=X

H

H

H

HX

+2 kcal/mol

If you don’t believe this, read: Houk et al., J. Am. Chem. Soc., 1987, 109, 6591.

• The products of addition to such C=X bonds generally prefer to be staggered (like simple alkanes).

• So, what does the transition state look like?

H

HH

HX

H

H HH

X Nu

Starting material:eclipsed

Product:staggered

‡

?

A Conformational Conundrum

• It was shown by Houk in the 1980’s that for acetaldehyde (and propene/butane) that the conformer with C-H and C= X ECLIPSING is more stable in the ground state:

H

HH

HX

is stabilising in this conformer!C-H *C=X

H

H

H

HX

+2 kcal/mol

If you don’t believe this, read: Houk et al., J. Am. Chem. Soc., 1987, 109, 6591.

• However, in a landmark 1986 Science paper, Houk published further studies that showed that the transition states for addition reactions were product-like i.e. staggered!

“Alkenes or carbonyls have one allylic bond eclipsed with the double bond in the ground state, but the transition structure conformations are more product-like.

Rotational barriers involving torsional interactions with partially formed bonds are nearly as large as those involving fully formed bonds. Consequently, the assumption of staggering in transition states is just as reasonable as the assumption of staggering

in stable molecules.” –Houk, Science, 1986, 231, 1108.

Felkin, Ahn and Friends• In a seminal Tetrahedron Letter, Felkin proposed the first model for stereoselective

nucleophilic attack on carbonyls with bonds fully staggered (1968):

• For R=H (i.e. aldehydes) then B should be preferred over A on steric grounds:

O

R

RL

RM

RS

Nu–

Nu–

O

H

RL

RM

RS

Nu–

O

HRL

RM

RS

A B

Often gives the right answer! Uses (correct) staggered conformer´ Nu approach still at 90 º´ Doesn’t work for Aldehydes (R = H)

• Ahn, using studies by Bürgi, Dunitz and Eisenstein improved the model and provided computational support for it:

C=O *

C-RL *

new lower LUMO!

H

RM

RS

O

Nu–

RL

Generally gives the right answer for aldehydes and ketones

Uses (correct) staggered conformer Nu approaches at ~107 º´ Often gives wrong results with

electronegative atoms in α-position.

Felkin, Ahn and Friends

MeO > t-Bu > Ph > i-Pr > Et > Me > H• A little surprisingly, Heathcock observed that the ‘effective substituent size’ for RL was:

O

H

RL

RM

Nu–RS

O

HRL

RSRM

OH

H

RL

RM

RS

Nu

HO HRL

RSRM

Nu

C C

RL

Nu

developingbond

* C-RL

LUMO

C-NuHOMO

* C-RL

C-Nu

developingbond

• The reason: the s* C-RL orbital antiperiplanar to the developing bond has a stabilising effect, which increases with s* C-RL acceptor ability (i.e. electronegativity).

• This general rule is known as the Ahn-Eisenstein hypothesis and is often phrased thus:

“The best acceptor s * orbital is oriented anti-periplanar to the forming bond."

• Felkin-Ahn results in 1,2-syn products (with respect to RM and the new hydroxyl):

Ahn-Eisenstein: Electronic Considerations

Me

H

O

Me

H

O

Me

Me

OMe

OLi Me

OH

Me MeOMe

O

Me

OH

Me MeOMe

O

Felkin predominatesdr = 9:1

Felkin predominatesdr = 200:1

Flippin et. al., Tetrahedron Lett., 1985, 26, 973

* C-Ph

H

Me

H

O

Nu–

Ph

C-Ph

E* Csp2-Csp3 is lower in energy than * Csp3-Csp3, making it a better acceptor

* C-Cy

H

Me

H

O

Nu–

Cy

C-Cy

“The best acceptor s * orbital is oriented anti-periplanar to the forming bond."

• The slightly lower energy of the s* Csp2–Csp3 orbital exerts a significant effect!

• Phenyl is smaller than cyclohexyl, so why is the reaction more diastereoselective?

• Nucleophilic attack can be very sensitive to small electronic effects:

Other Factors

PhO

H

Me

Me

H

O

H Nu–

Me

OH

Me

Me

OH O

Ot-Bu

MeLi or MeMgBr

Nucleophile

OLi

Ot-Bu

OTBS

Ot-Bu

4:1

24:1

Reetz et. al., Angew. Chem. Int. Ed., 1982, 21, 135

9:1

2:1

dr

Felkin products

MeTi(Oi-Pr)3

• Unsurprisingly, bigger nucleophiles give better selectivity (branched enolates are especially good).

• However, counter ions and the nature of the nucleophile can be important for high selectivity:

H

CyMe

H

Cl

Me

H

Cl

Me

MeMe2CuLi•ZnCl2 Felkin product

dr > 100:1

Nakamura et al., J. Org. Chem., 1993, 58, 5121

• The Felkin-Ahn model can also be used to predict the outcome of certain SN2’ reactions.

A Notable Exception: Borane Reagents

Me

Me

O

Me

Me

OH

Me

Me

OH

Anti-Felkin productFelkin product

MH

Reagent drMe

Me

Me

Sia =LiBH(s-Bu)3

HB(Sia)2 1:4

22:1

Midland et. al., J. Am. Chem. Soc., 1983, 105, 3725

• The Felkin-Ahn model gives good results in many cases, but the borane reduction of ketones usually gives the anti-Felkin diastereomer:

RL

RM

R

O

H

RL

RMOR

H B RR

H

RL

RM

RO

HBR

RH

RL

RM

RHO

H

RL R

OH

RM

H

RL

RM

OHR

H

RL R

OH

RM

R2BH

R2BH

Felkin addition

Anti-Felkin addition Favoured

Disfavoured

stericclash!

• This is due to the “non-spherical” nucleophile and the nature of its approach:

1,2-Chelation Control – Cram Model

R1

OR2H

O

H

R1

H

O

Nu–

OR2R1

OR2Nu

OH

H

OR2

H

O

Nu–

R1

R1

OR2Nu

OH

Anti-Felkin product (Minor)

Felkin product (Major)

H

R1

H

OR2O

Nu–

M

R1

OR2Nu

OH

Major productunder chelation control

Remember OR2

always counts as RL

in Felkin-Ahn model

O

HRL

RSRM

HO HRL

RSRM

Nu

• Normal diastereoselectivity can be reversed by carrying out reactions under chelation control

• This normally entails the use of Zn-, Al- or Ti-based reagents/Lewis acids, however hydrogen bonds can also be used.

• These reactions are highly solvent dependent (more polar usually means a better dr).

An example of 1,2-chelation control

Me

Me

O

OPg

MeMe

Me

OH

OPg

MeMe

Me

OH

OPg

Me

Felkin-Ahn Chelation

LiAlH4,

Pg

BOM

BOM

TBDPS

SolventRatioF-A:C

THF

THF

30:70

2:98

95:5

Et2O

No chelation; use F-A with RL = OTBDPS

L. Overman, Tetrahedron Lett., 1982, 23, 2355

Solvent,–10 °C

• Silicon protecting groups usually negate chelation, EXCEPT for reactions with certain Zinc (e.g. ZnR2) or Aluminium (e.g. Me2AlCl or MeAlCl2) nucleophiles/Lewis acids

1,3-Induction: Reetz Cyclic Model

O

H

OBn

R

OH

R

OBn

R

OH

R

OBn

R

SiMe3

n-Bu2Zn

NucleophileRatio

Anti:SynR

n-Bu

95:5

90:1

R

Me

Me

99:1SiMe3

Men-Bu

Me

Nu, TiCl4CH2Cl2, –78 °C,

>90% yields1,3-anti 1,3-syn

M. Reetz, J. Am. Chem. Soc., 1983 105, 4883.

TiLn

O

O

H

R

H

H

Bn

Nu

‡

For bidentate Lewis acidsreaction proceeds via asimple cyclic transition state (according to Fürst-Plattner)

The Reetz Modelfor cyclic 1,3-induction

+

Major

“The reactions mediated by BF3•OEt2… represent the real surprise” –Reetz

• Bidentate Lewis acids can catalyse the addition of nucleophiles to b-alkoxycarbonyls with good diastereoselectivity

• These reactions are thought to proceed through cyclic transition states:

1,3-Induction: Evans Polar Model

81:19 87

73:27 90

43:57 79

83:17 84

OSiMe3

i-Pr

O

H

X

R

OH X

R

O

i-Pr

OH X

R

O

i-Pr

BF3•OEt2+

1,3-anti 1,3-syn

RRatio

Anti:SynYield(%)

H HOH LA

X HR

Nu

The Evans Polar Modelfor acyclic 1,3-induction

‡OPMB

OTBS

OAc

Cl

Me

(CH2)2Ph

(CH2)2Ph

(CH2)2Ph

(CH2)2Ph

(CH2)2Ph 58:42 88

X

CH2Cl2, –78 °C

D. A. Evans, Tetrahedron Lett., 1994, 35, 8537.

Major

• Surprisingly, stereoinduction remains good even for monodentate Lewis Acids!• Reetz proposed an acyclic model to explain this stereoinduction but it has now been

superseded by the newer Evans Polar Model:

• Note that both chelate and polar models favour 1,3-anti products. • Non-coordinating (i.e. alkyl) substituents poorly induce diastereoselectivity.

Multiple Stereocentres• As stated earlier: 1,2-stereoinduction favours syn products (Felkin-Ahn Model); 1,3-

induction (whether cyclic or not) favours anti products.

• Thus, for molecules with α- and β-stereocentres, these effects can potentially reinforce or interfere with each other.

O

H R

OPg

Me

Non-Reinforcing

O

H R

OPg

Me

H MeOH LA

PgO HR

Nu‡

Reinforcing

OH

Nu R

OPg

Me

OH

Nu R

OPg

Me

Felkin productgood dr

product dependson Nu (and conditions)

Nu–

• For the Non-Reinforcing α,β-syn case: large nucleophiles react under Felkin-Ahn control; small nucleophiles tend to react under 1,3-control (to give anti-Felkin products).

• For the Reinforcing α,β-anti case excellent diastereomeric induction (often >99:1) is usually observed, almost irrespective of the nucleophile or subsituents.

HR

O

Me

NuR

OH

Me

H MeH O

Nu–

R

O

H

OPg

R

OH

Nu

OPg

R

H HOH LA

PgO HR

Nu‡

‡

1,2-stereoinduction favours syn 1,2-stereoinduction favours anti

syn anti

Key reference: Evans et al., J. Am. Chem. Soc., 2001, 123, 10840.

Felkin-Ahn meets Zimmerman-Traxler

HR1

OMLn

Me

O

LnM O

R2

Me

H

R1

O

R1

Me

R2

OH

(E)-enolate Z-T chair-like TS

‡ anti

(±)-anti aldol product

HR1

OMLn

MeO

LnM O

R2Me

H

R1

O

R1

Me

R2

OH

(Z)-enolate Z-T chair-like TS

‡ syn

(±)-syn aldol product

R2CHO R2CHO

• In a simple aldol reaction, the major diastereomer formed depends on the geometry of the enolate, which is transferred to the product via a simple cyclic transition state.

• However, additional stereocentres in the two components can also influence the reaction outcome, and their effects must be considered if present.

Combining Models

H

OMLn

Me

O

M O

RH

MeO

Me

R

OH

‡

RCHORL

Me

OMLn

Me

Me

H

RL

C-Cbond

rotation H RL

MeMeH

RL anti-anti aldol product

H

O

MLnO

RH

Me

‡

MeHRL

O

Me

R

OH

RL

Me

RCHO

syn-anti aldol product

(E)-enolatesform SYN-ANTI

products

(E)-enolate(to minimiseA1,3 strain)

L

L

H

OMLn

Me O

LnM O

RMe

HO

Me

R

OH

‡

RCHORL

Me

OMLn

MeRL

Me

H

C-Cbond

rotation

H

RL

MeH

MeRL syn-syn aldol product

H

O

MO

RMe

H

‡

HMe

RL

O

Me

R

OH

RL

Me

RCHO

anti-syn aldol product

(Z)-enolatesform ALL SYN

products

L

L

(Z)-enolate

(to minimiseA1,3 strain)

• First, let’s consider an adjacent stereocentre on the enolate component.• (Z)-enolates with α-stereocentres generally form all syn products:

• Conversely, (E)-enolates with α-stereocentres generally form syn-anti products:

Combining Models

H

O

LnM O

H

Me

R

‡

H

O

MLnO

H

Me

R

‡

R

OMLn

Me

O

HRL

Me

H

Me

RL

MeRL

destabilised bysyn-pentane interaction

OH

RL

Me

R

O

Me

OH

RL

Me

R

O

Me

Remember:Felkin-Ahn Modelpredicts 1,2-syn

O

HRL

RM

Nu

OH

RL

RM

Nu–

1,2-syn

1,2-anti

Felkin product(E)-enolate

+

Roush (only!), J. Org. Chem., 1991, 56, 4151.

H

O

LnM O H

Me

R

‡

H

O

MLnO

R

‡

R

OMLn

MeO

HRL

Me

H

Me

RL

MeRL

destabilised bysyn-pentane interaction

OH

RL

Me

R

O

Me

OH

RL

Me

R

O

Me

1,2-syn

1,2-anti

Felkin product

H

Me

Anti-Felkin product

(Z)-Enolate

• Aldehydes with a-stereocentres give anti-syn (Felkin) products with (E)-enolates…

• …and syn-anti (Anti-Felkin) products with (Z)-enolates:

Combining Models

Evans et al., J. Am. Chem. Soc. 1995, 117, 9073.

Conclusions“The on-going development of chemical reactions of ever increasing selectivity coupled

with an evolving sophistication in the tenets of synthesis design now provide one with the basic tools to design and execute rationally the laboratory synthesis of an impressive array of organic structure. Without question, the prime activating function for the contruction of

carbon-carbon bonds is the carbonyl function” – Evans, 1984

• Even in acyclic systems, there are many tricks that can be used to perform highly diastereoselective reactions using simple non-chiral reagents in conjuction with the existing stereocentres present.

• This talk has only covered a few simple cases for 1,2- and 1,3-asymmetric induction in addition of nucleophiles to carbonyl compounds but sometimes up to 1,6-induction is routinely used (see for example Paterson, J. Org. Chem., 2005, 70, 150)!

Possible Reading

Books:• Classics in Stereoselective Synthesis by Carreira and Kvaerno• Modern Physical Chemistry by Anslyn and Dougherty• Strategy and Control by Wyatt and Warren

Reviews/Lectures• “Around and Beyond Cram's Rule” A. Mengel and O. Reiser, Chem. Rev., 1999, 99, 1191.• “Structural, mechanistic, and theoretical aspects of chelation controlled carbonyl

addition reactions.” M. Reetz, Acc. Chem. Res., 1993, 26, 462.• D. A. Evans’ Harvard Chem 206 Lectures, particularly 1-7, 17, 21 and 22.