Tải bản đầy đủ - 0 (trang)
2 ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS

2 ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS

Tải bản đầy đủ - 0trang

392



ORGANIC REACTIONS OF ANIONS



Cram’s model

O



O



M



S



M

Nuc−



S



Nuc−



R L



R L

Karabastos’ model

OL



OM

Nuc−

L



Nuc−

S



S



M



R



R

Felkin’s model



O



O

M



S

Nuc−



L

R



L



Nuc−

M



S



R



Scheme 6.1



30∘ increments for the two diastereomers, they generated 24 geometries, for which

HF/STO-3G energies were computed.

30°



O



H



Me



H

1.5Å



H

1.5 Å



Cl

H



O



H



Me



Cl

H



30°



20



Figure 6.11 displays the relative energies of these 24 geometries against the rotation angle about the C1 –C2 bond. These energies used in this plot were obtained in

a method slightly modified from the Anh–Eisenstein procedure. First, the geometry

of the lowest energy TS leading to the major and minor products were fully optimized at B3LYP/6-31++G(d). Next, 12 conformers of each structure were obtained

by rotating by 30∘ about the C1 –C2 bond, holding all other geometric parameters

fixed except the O–C–Hincoming angle that was allowed to relax. This process produces more rigorously defined rotation energy TS surface than that performed by

Anh and Eisenstein but the results are qualitatively very similar.

The lowest energy TS, which leads to the major product, has just the conformation predicted by Felkin. The lowest energy TS leading to the minor product is also

that predicted by Felkin. In other words, Felkin’s assumption of placing the largest



ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS



393



Anh and Eisenstein analysis



45

40

35



Erel



30

25

20

15



H O

H



10



Cl

Me



H



Cl



5

0



O Me

H H



0



90



180



270



H



360



Dihedral angle



Figure 6.11 Model transition state rotational energy surface for the reaction of H− with

2-chloropropanal. The full line represents transition states leading to the major product and

the dashed line represents those leading to the minor product. Modified from the original

model proposed by Anh and Eisenstein.54



group perpendicular to the carbonyl and antiperiplanar to the incoming nucleophile

is correct.

Anh and Eisenstein also supplied a reason for the preference of the M group

being in the “inside” position. The incoming hydride does not enter along a path

perpendicular to the carbonyl, but rather on a path making an angle of about 103∘

(21). Therefore, the path past the smallest group (S) in the “outside” position

will be sterically less congested than that if the M group were in the “outside”

position.

O



S



Me



O



L



L

H



S



Nuc

21



Nuc

Me



H



Felkin’s model relied on reducing steric interactions and strain energy about

the carbonyl group. Anh and Eisenstein, however, supplied an alternative reason

for the preferred transition-state conformation. The dominant orbital interaction in

this reaction is between the highest occupied molecular orbital (HOMO) of the

nucleophile and the lowest unoccupied molecular orbital (LUMO) of the carbonyl



394



ORGANIC REACTIONS OF ANIONS



σ *C–L

π *CO



HOMONuc



Smaller

HOMO–LUMO

gap



C



O



C



L



Scheme 6.2



(π∗CO ). When the L group is located antiperiplanar to the incoming nucleophile,

the antibonding orbital between carbon and the L group (σ∗C−L ) can effectively

lower the π∗CO energy, allowing for greater interaction with the incoming nucleophile (Scheme 6.2). The definition of the L group, therefore, becomes the group

that has the lowest σ∗C−L orbital. So while the Felkin model relies on reducing steric

interactions, the Anh–Eisenstein variation relies on maximizing hyperconjugative

stabilization in the TS. Heathcock55 has modified this viewpoint, arguing that in

fact, both electronic (Anh’s hyperconjugation argument) and steric (Felkin’s argument) effects determine which group is the “largest.”

Early ab initio computations fully supported the arguments of Anh and Eisenstein. Schleyer and Houk56 optimized the transition structures for the reaction of

formaldehyde with LiH (22), methyllithium (23), and methyllithium dimer (24)

at HF/3-21G. These TSs, reoptimized at HF/6-31G*, are shown in Figure 6.12.

The nucleophile adds from the face perpendicular to the carbonyl group. Wong

and Paddon-Row57 examined the reaction of cyanide with 2-fluoropropanal. They

located six TSs at HF/3-21G, differing by which group is anti to the incoming

cyanide nucleophile. The two lowest energy TSs (25 and 26, Figure 6.13) correspond to the two Felkin–Anh conformations, where the fluorine atom is anti to the

incoming cyanide. TS 25, the one predicted by the Felkin–Anh model to lead to



1.680



2.135



2.057



2.058



2.078



2.669

89.3



1.206



22



1.915



2.588



105.9

1.220



23



1.857



3.021



113.7



1.874



1.203



24



Figure 6.12 HF/6-31G* optimizations of the transition states for the reaction of formaldehyde with HLi, CH3 Li, and the CH3 Li dimer.



ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS



O Me



Me



F

H H



CN



NC



395



O

F

HH



1.926

1.922



25



26



Figure 6.13 HF/6-31G* optimized transition states for the reaction of cyanide with

2-chloropropanal.



the major product, is 6.9 kcal mol−1 lower in energy at MP2/6-31+G*//HF/3-21G

than the alternative TS 26. While the CNuc –C1 –O angle is nearly identical in the

two TSs (about 113∘ ), the CN –C1 –C2 –(outside group) dihedral angle is 12∘ larger

in 26 than in 25, reflecting the poorer steric interactions when the nucleophile must

pass by the medium group rather than by the smallest group.

Unquestionably, the work of Anh and Eisenstein is seminal. Even though their

computations were performed at what is today considered an extremely rudimentary level, and as we will see, it is not the last word on the subject, the predictive

power of the Felkin–Anh model has become a standard tool in organic chemistry.58

As argued by Houk,59 the Ahn–Eisenstein “was important not only because it

solved an important general problem in stereoselectivity but also because it demonstrated the power of quantum mechanical calculations to solve important problems

on real organic systems.”

Cieplak60 countered the Anh explanation with an alternative orbital model.

He noted that reductions of cyclohexanones and other additions at carbonyls

occasionally resulted in the major product coming from the Felkin–Anh minor

TS. Arguing that since the incipient bond was electron deficient—a partial bond

lacks the full two-electron occupation—it is donation of density from the σC2−L

into the σ∗C−Nuc orbital that will stabilize the TS (Scheme 6.3). Support for the

Cieplak model was provided by experimental results for nucleophilic addition to

3-substituted cyclohexanones,61 reductions of 2,2-diarylcyclopentanones,62 and

especially le Noble’s extensive studies of reductions of adamantanones.63

Negative reaction to the Cieplak model by many computational chemists

was quickly forthcoming, resulting in modifications to the simple Felkin–Anh

model. Houk64 reported the three TSs for the reaction of NaH with propanal (the

HF/6-31G* optimizes structures are shown in Figure 6.14). The lowest energy

TS is 27a, where the methyl group is in the “inside” position. Assuming that the

methyl group is larger than a hydrogen atom, the Felkin–Anh TS, 27b, is 1.0 kcal



396



ORGANIC REACTIONS OF ANIONS



O



O



Scheme 6.3



27a



27b



27c



Figure 6.14 HF/6-31G* optimized TSs for the reaction of NaH with propanal.



mol−1 above 27a, and the TS where the methyl group is in the outside position

(27c) is 0.9 kcal mol−1 above 27a. For the Cieplak model to work, the CH bond

would have to be a better donor than the CC bond, an argument Cieplak60,61 and

others65,66 have made, but this contention has been disputed.67 Houk maintained

that the CH bond is a poorer donor than the CH bond and rejects the Cieplak

model. Rather, he argued that the methyl group, being a better donor, destabilizes

27b relative to 27a. Keep in mind that the Anh model is based on the anti group

being the best acceptor, that is, having the lowest 𝜎* orbital.

Frenking has also criticized the Cieplak model. His initial argument68 is

with the conceptual footing of the model itself. Stabilization of the Cieplak

TS comes by donation of density into the vacant antibonding orbital for the

forming bond. In other words, this is a HOMO(σC−L )–LUMO(σ∗C−Nuc ) interaction.

This is not the interaction favored by frontier molecular orbital (FMO) theory:

HOMO(σC−Nuc )–LUMO(π∗C−O ). Instead of using the Cieplak model, Frenking

argued that the preferred orientation of 1,2-addition is understood in terms of the

Felkin model, the effect of the conformation of the aldehyde, and the shape of the

carbonyl LUMO.

Frenking examined the TSs for the reaction of LiH with propanal,

2-chloroethanal, and 2-chloropropanal.69 He found three TSs (28a–c) very

similar to 27a–c. 28a is the lowest energy of the three, while 28b and 28c are 1.3

and 1.6 kcal mol−1 higher in energy, respectively. If the LiH fragment is removed,



ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS



O



O



Li



Me



H



H

H

28a



H



O



Li



H



397



Li



H



Me



H

H

28b



Me



H

H

28c



H



H



but the structure of the remaining aldehyde is kept unchanged, 28a remains the

lowest in energy, 28b is 1.2 kcal mol−1 higher and 28c is 0.7 kcal mol−1 above

28a. Frenking argued that this indicates that the energy of the conformers of the

aldehyde alone determine much of the energy difference between the TSs.

For the reaction of lithium hydride with 2-chloroethanal, the lowest energy TS

is 29a (Figure 6.15), with 29b and 29c lying 1.7 and 0.3 kcal mol−1 above 29a,

respectively. When the lithium cation is removed from each structure and its energy

recomputed, 29c has the lowest energy, with 29b and 29a are 3.4 and 6.4 kcal

mol−1 , respectively, above it. Frenking suggested that this indicates strong complexation energy in 29a, reminiscent of the Cram chelation model where the cation

associates with both the carbonyl oxygen and the electronegative substituent on C2 .

This chelation stabilizes 29a more than the Anh orbital interaction available in 29c.

Paddon-Row70 noticed the same favorable chelation in the most favorable TS for

the analogous reaction of LiH with 2-fluoroethanal.

Lastly, in the reaction of LiH with 2-chloropropanal, five TSs were located,

30a–e (Figure 6.16). 30a and 30b lead to the major product, while the other

three lead to the minor product. The two lowest energy structures correspond

with the Felkin–Anh major (30a) and minor (30c) TSs. Their energy difference

corresponds with the energy difference for their aldehyde fragments when LiH is

removed. 30b and 30d display the chelation effect, but unlike with 29, chelation

is not enough to make up for the favorable orbital interactions in the Felkin–Anh

approach.

Frenking next examined the reaction of LiH with cyclohexanone.68 The

Hartree–Fock (HF)/3-21G transition structures for axial (31ax) and equatorial

(31eq) attack are shown in Figure 6.17. Axial attack is lower than equatorial



29a



29b



29c



Figure 6.15 HF/6-31G* optimized transition structures for the reaction of LiH with

2-chloroethanal.



ORGANIC REACTIONS OF ANIONS



398



Li Me O

H



Li



Cl

H



H

30a

0.0

(−LiH) 0.0

(−Li+) 0.0



Cl

H



O



O



H Cl

Me H

30b

1.8

2.0

10.4



H

30c

1.3

1.9

1.8



H

H

Me



O



Li

Me



H H

30d

1.5

1.3

10.4



O



Li



Cl

H



Me



H

H Cl

30e

2.9

0.0

7.5



Li



H



Figure 6.16 Transition states for the reaction of LiH with 2-chloropropanal.

(a) HF/6-31G* relative energies; (b) relative energies without LiH; (c) relative energies

without Li+ .



2.06

(2.57)



2.03

(2.42)



31ax

(−LiH)

(−Li+)



0.0 (0.0)

0.0

0.0



31eq

1.4 (2.1)

−0.3

3.2



Figure 6.17 HF/3-21G optimized structures for the reaction of LiH with cyclohexanone.68

(a) MP2/6-31G*//HF/3-21G relative energies; (b) relative energies without LiH;

(c) relative energies without Li+ . Values in parenthesis are for the B3LYP/6-31G** optimized structures.72



attack by 1.4 kcal mol−1 . The energy difference between the two is only 0.3 kcal

mol−1 when the LiH fragment is removed, indicating that two ketone fragments

have similar energies. This is counter to Houk’s71 explanation for the preference

for axial attack. Houk argues that the equatorial TS is more strained, based on

geometrical arguments; however, the equatorial TS is actually slightly less strained

than the axial TS. The large difference in energies when Li+ is removed from

the structures suggests that the preference for axial attack must come from the

interaction of the nucleophile (H− ) with cyclohexanone. Frenking argued that the

carbonyl LUMO is distorted by hyperconjugation making the carbon lobe larger

on the side of axial attack. Very similar conclusions have been drawn by Luibrand

in a more recent B3LYP/6-31G** study of the reaction of cyclohexanone with

LiH or LiAlH4 .72 The B3LYP TSs are much earlier than the HF ones, but their

energy differences are comparable—both the methods predict the axial attack to

be lower in energy than the equatorial attack.



ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS



399



Frenking also examined the reaction of LiH with 3-fluorocyclohexanone, for

which four TSs were located. Fluorine can occupy either the equatorial (32) or the

axial (33) position and then attack can come from the axial or equatorial faces. The

Cieplak model does not distinguish between 32 and 33, predicting axial attack for

both. While axial attack is preferred in 32, equatorial attack is favored by 2.3 kcal

mol−1 over axial attack in 33. Frenking suggests that this difference is reflected in

the orbital coefficients of the LUMO of 32 and 33. A simpler explanation, one that

will be further explored next, is that electrostatic interactions between the hydride

and the axial fluorine destabilize 33ax relative to 33eq.

Li

H



Li



O



O

Li



F



O



32eq

2.7



O



F



Li



H



F

32ax

0.0



F



H



H

33ax

0.0



33eq

−2.3



Paddon-Row and Houk73 were the first to strongly advocate for the role of electrostatics in determining the stereo-outcome of 1,2-addition reactions. They examined the addition of LiH to a number of substituted-7-norbornanones, of which

34 is representative. The addition can come from the same side of substituents

(34syn) or from the opposite side (34anti). These two transitions states are shown

in Figure 6.18. At MP2/6-31G*//HF/3-21G, attack is favored from the syn face

by 4.0 kcal mol−1 over anti attack. When the LiH fragment is removed, the two

TSs differ by only 0.8 kcal mol−1 , with 34anti lower in energy. Natural population



(−LiH)

(−LiH, 0.5 e at H)



34anti

4.0

0.0

4.6



34syn

0.0

0.8

0.0



Figure 6.18 HF/6-31G*-optimized transition states for the reaction of LiH with 34. (a)

Relative energies at MP2/6-31G*//HF/6-31G*; (b) relative energies without LiH; (c) relative

energies with LiH and a −0.5 point charge at the position of the hydride.73



400



ORGANIC REACTIONS OF ANIONS



analysis suggests that the charge on the hydride is about −0.5. When the LiH fragment is removed and a -0.5 charge is placed at the position of the hydride, the syn

TS structure is 4.6 kcal mol−1 lower in energy than the anti one, almost perfectly

matching the energy difference between 34syn and 34anti. Since the formyl group

will polarize C2 and C3 (the substituted ring carbons) and will make them partially

positively charged, stronger electrostatic stabilization will occur in 34syn, where

the hydride atom is closer to these positive centers than in 34anti.

O



O

34



O



Paddon-Row extended this work with the investigation of the reaction of disubstituted norbornen-7-ones (35) with LiH or methyllithium.74 Again, there are two

pathways, one from the same side as the substituents (syn) and one from the side of

the alkene (anti). The energy difference between the TSs for these two paths with

either LiH or MeLi are listed in Table 6.9.

O

anti



syn



R

a: SiMe3

b: Me

c: H

R



R



d: CH2OH

e: CN



35



Syn attack of LiH is favored for all five cases (35a–e). This is contrary to the

predictions of the Cieplak model, which would favor anti attack for compounds

with electron-donating substituents, such as 35a. The Anh–Eisenstein model fairs

no better, as it predicts anti attack for compounds with electron-withdrawing substituents, such as 35e. Paddon-Row, however, argued that electrostatic effect can

account for this uniform attack direction. Minimization of steric interactions would

argue for the anti approach. However, the anti approach takes the nucleophile over

the electron-rich double bond. The resulting electrostatic repulsion disfavors the

anti path and the syn approach dominates. The very large energy difference between

the syn and anti TSs for LiH addition to 35e is understood also in terms of electrostatics. Here, the strongly electron-withdrawing cyano groups build up positive



ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS



401



TABLE 6.9 MP2/6-31G*//HF/6-31G* Energy

Difference between the anti and syn Transition States

(Eanti − Esyn ) for the Reaction of 35 with LiH or MeLia



35a

35b

35c

35d

35e

a Ref.



LiH



MeLi



10.14

8.32

12.38

7.83

34.78



−14.56

−14.06

−7.88

−11.28

7.19



74.



charge on their neighboring carbon atoms, generating favorable electrostatic interactions with the incoming nucleophile for the syn face.

For the reactions of 35a–e with methyllithium, steric interactions overcome

the electrostatic effects and the anti TSs are lower in energy than the syn TSs for

all cases but 35e. Here, the strong electron-withdrawing nature of the cyano substituents generates enough favorable electrostatic interaction to outweigh the steric

demands, and the reaction through the syn TS is the preferred path.

Marsella75 has proposed a scheme for predicting the preferred direction of attack

at carbonyls. He suggests visualizing the computed electrostatic potential on an

isosurface of the carbonyl π∗CO orbital (the LUMO). The lobe that is more positive

accurately predicts the correct face that is attacked for 13 different reactions.

The interplay between favorable electrostatic interactions and steric demands

was further demonstrated in the computational study of the reaction of LiH

with 2-silylethanal (Reaction 6.7) and 2-trimethylsilylethanal (Reaction 6.8). For

both the reactions, Fleming et al.76 located three TSs corresponding to different

rotamers about the C–C bond. The Felkin model, in which the largest (most

sterically demanding) group is placed anti to the nucleophile, suggests that TSs

36c and 37c should be the best. The Anh–Eisenstein model favors 36b and 37c;

here, the strongest donor is placed anti to the nucleophile and the larger silyl group

is placed in the “inside” position.

RCH2 CHO + LiH → RCH2 CHOH



R = SiH3,



(Reaction 6.7)



R = SiMe3,



(Reaction 6.8)



For Reaction 6.7, the lowest energy (MP2/6-31G*//HF/6-31G*) TS is 36a, with

the Anh TS lying 0.56 kcal mol−1 higher, and the Felkin TS 0.05 kcal mol−1 higher

still. In contrast, the relative ordering of the TSs is reversed for Reaction 6.8; the

Felkin TS is the lowest in energy, followed by the Anh TS lying 0.98 kcal mol−1

higher, and 37a is 1.15 kcal mol−1 above 37c. 36a is stabilized by electrostatic

attraction between the nearby electropositive silyl group and the hydride; these

electrostatics dominate any steric repulsions, and 36a is the preferred TS for Reaction 6.7. In 37a, the much larger trimethylsilyl group does not allow for the two



402



ORGANIC REACTIONS OF ANIONS



oppositely charged groups to closely approach each other. The sheer bulk of the

trimethylsilyl group makes 37c the most favorable TS for Reaction 6.8.

O



H



O



Li



H



SiH3



H

H

36a

0.0

O



H

SiH3



H



H H

36b

0.56

O



Li



H



H



H

37a

1.15



H H

36c

0.61

O



H H

37b

0.98



H



Li



H



H3Si



Li

SiMe3



H

H

SiMe 3



O



Li



H



Li



H



Me3Si

H H

37c

0.0



H



Another interesting variation on the Felkin–Anh model is the reduction of

ketones with very bulky α-substituents: phenyl (38), cyclohexyl (39), and t-butyl

(40). The TSs for the reactions of these ketones with LiH were examined at

MP2/6-311+G**//HF/6-31G*.77 The optimized geometries for the lowest energy

TS leading to the Cram and anti-Cram products are shown in Figure 6.19. For

the reactions of 38 and 39, the computations predict the Cram product to be the

major product, consistent with the experimental reduction of these ketones with

LiAlH4 . At first sight, this result appears consistent with the Felkin–Anh model:

TSs 38cram and 39cram have the largest group anti to the incoming hydride and

the methyl group occupies the “inside” position. However, the TSs leading to the

minor, anti-Cram, product (38anticram and 39anticram) have the methyl group

in the anti position and the large group in the “inside” position.

O



R



38: R = Phenyl

39: R = Cyclohexyl

40: R = t-Butyl



More unusual are the results for the reaction of 40. The two TSs, 40cram and

40anticram, are geometrically consistent with the Felkin–Anh model. Both have

the largest group anti to the hydride. The Felkin–Anh model predicts that the TS

with the methyl group in the “inside” position (40cram) should be lower than that

when the methyl group is in the “outside” position (40anticram). The MP2 energies

are in fact opposite: 40anticram is 3.42 kcal mol−1 lower in energy than 40cram,

and the major product found in the experimental reduction of 40 with LiAlH4 is the

anti-Cram product!



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

2 ASYMMETRIC INDUCTION VIA 1,2-ADDITION TO CARBONYL COMPOUNDS

Tải bản đầy đủ ngay(0 tr)

×