On the mechanism of Lewis base catalyzed aldol addition reactions: Kinetic and spectroscopic investigations using rapid-injection NMR

Published on Web 07/31/2009

On the Mechanism of Lewis Base Catalyzed Aldol Addition

Reactions: Kinetic and Spectroscopic Investigations Using

Rapid-Injection NMR

Scott E. Denmark,* Brian M. Eklov, Peter J. Yao, and Martin D. Eastgate

Roger Adams Laboratory, Department of Chemistry, UniVersity of Illinois,

Urbana, Illinois 61801

Received April 6, 2009; E-mail: sdenmark@illinois.edu

Abstract: The mechanistic foundations of the Lewis base catalyzed aldol addition reactions have been

29

31

investigated. From a combination of low-temperature spectroscopic studies ( Si and P NMR) and kinetic

analyses using a rapid-injection NMR apparatus (RINMR), a correlation of the ground states and transition

structures for the aldolization reactions has been formulated. The aldol addition of the tert-butylsilyl ketene

acetal of tert-butyl propanoate with 1-naphthaldehyde is efﬁciently catalyzed by a combination of silicon

tetrachloride and chiral phosphoramide Lewis bases. The rates and selectivities of the aldol additions are

highly dependent on the structure of the Lewis bases: bisphosphoramides give the highest rate and

selectivity, whereas a related monophosphoramide reacts slowly and with low selectivity. The monophos-

phoramide shows no nonlinear behavior. All of the additions show a ﬁrst-order kinetic dependence on silyl

ketene acetal and 1-naphthaldehyde and a zeroth-order dependence on silicon tetrachloride. The kinetic

order in catalyst is structure dependent and is either half-, two-thirds-, or ﬁrst-order. All of the phosphoramides

are saturated with silicon tetrachloride in some form, and the resting-state species are mixtures of monomeric

and dimeric, pentacoordinate cationic, or hexacoordinate neutral complexes. These data allow the

formulation of a uniﬁed mechanistic scheme based on the postulate of a common reactive intermediate for

all catalysts.

Introduction

bonyl subunit, the asymmetric aldol addition reaction has been

both an engine and a proving ground for new methodological

The evolution and development of catalytic enantioselective

and diastereoselective) aldol addition reactions represents the

advances. For example, the study of the structure and reactivity

(

3

of metal enolates, the design and development of chiral Lewis

apotheosis of organic synthesis methodology. The generality,

versatility, and selectivity associated with this process have been

thoroughly chronicled in countless books, reviews, and authori-

4

acids based on nearly every element in the periodic table, and

the most recent frenzy of disclosures on direct aldolization via

5

1

enamine catalysis amply illustrate this point. However, despite

tative summaries. Stimulated by the challenges posed by nature,

the enormous impact of these advances for addressing practical

problems of efﬁciency, generality, and selectivity, the extent

of fundamental mechanistic understanding of the newly devel-

oped processes pales by comparison. Remarkably, the literature

in this ﬁeld is replete with transition structure-based rationaliza-

tions for the stereochemical course of these reactions without

even the most basic knowledge of the reaction parameters, i.e.,

the catalyst resting state, the kinetic equation, and the turnover-

limiting and stereochemistry-determining steps. Although

notable exceptions exist, this situation arises because the

advances in analytical chemistry make it much easier to measure

generations of synthetic organic chemists have constructed an

impressive ediﬁce of knowledge that constitutes insightful,

elegant, and practical solutions to the structural and stereo-

chemical problems presented by polypropionate-derived natural

2

products.

Beyond its obvious utility for the synthesis of natural and

non-natural compounds bearing the signature ꢀ-hydroxy car-

(

1) For reviews on catalytic asymmetric aldol reactions, see: (a) Modern

Carbonyl Chemistry; Otera, J., Ed.; Wiley-VCH: Weinheim, 2000;

Chapt. 8. (b) Carreira, E. M. In ComprehensiVe Asymmetric Catalysis;

Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer Verlag:

Heidelberg, 1999; Vol. III, Chapt. 29.1. (c) Carreira, E. M.; Fettes,

A.; Marti, C. Org. React. 2006, 67, 1–216. (d) Carreira, E. M. In

Catalytic Asymmetric Synthesis, 2nd ed.; Ojima, I., Ed.; Wiley-VCH:

Weinheim, 2000; Chapt. 8B2. (e) Braun, M. In StereoselectiVe

Synthesis, Methods of Organic Chemistry (Houben-Weyl), Vol. 3,

Edition E21; Helmchen, G., Hoffman, R., Mulzer, J., Schaumann, M.,

Eds.; Thieme: Stuttgart, 1996; p 1603. (f) Nelson, S. G. Tetrahedron:

Asymmetry 1998, 9, 357. (g) For a deﬁnitive treatise see: Modern Aldol

Reactions; Mahrwald, R., Ed.; Wiley-VCH: Weinheim, 2004.

(3) (a) Williard, P. G. In ComprehensiVe Organic Synthesis, Vol. 1,

Additions to C-X Bonds, Part 1; Trost, B. M., Fleming, I., Eds.;

Pergamon: Oxford, 1991; Chapt. 1.1. (b) Seebach, D. Angew. Chem.,

Int. Ed. Engl. 1988, 27, 1624–1654. Collum, D. B.; McNeil, A. J.;

Ramirez, A. Angew. Chem., Int. Ed. 2007, 46, 3002–3017.

(4) (a) Yamamoto, H., Ed. Lewis Acids in Organic Synthesis; Wiley-VCH:

Weinheim, 2000. (b) Santelli, M.; Pons, J.-M. Lewis Acids and

SelectiVity in Organic Synthesis; CRC Press: Boca Raton, FL, 1996.

(5) (a) Berkessel, A.; Groger, H. Asymmetric Organocatalysis; Wiley-

VCH: Weinheim, 2005. (b) Dalko, P. I.; Moisan, L. Angew. Chem.,

Int. Ed. 2004, 43, 5138. (c) Mukherjee, S.; Yang, J. W.; Hoffmann,

S.; List, B. Chem. ReV. 2007, 107, 5471.

(

2) (a) Norcross, R.; Paterson, I. Chem. ReV. 1995, 95, 2041. (b) Paterson,

I.; Cowden, C. J.; Wallace, D. J. In Modern Carbonyl Chemistry; Otera,

I., Ed.; Wiley-VCH: Weinheim, 2000, Chapt. 9. (c) Florence, G. J.;

Gardner, N. M.; Paterson, I. Nat. Prod. Rep. 2008, 25, 342–375.

1

1770 9 J. AM. CHEM. SOC. 2009, 131, 11770–11787

10.1021/ja902474j CCC: $40.75  2009 American Chemical Society

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

Figure 1. Grand uniﬁed mechanistic scheme for phosphoramide-catalyzed aldolization.

an enantiomer ratio or diastereomer ratio than to determine a

rate equation.

Commencing in 1996, investigations in these laboratories

reaction involved the use of preformed trichlorosilyl enolates

6

,8

9

10

of esters, ketones, and aldehydes. These reagents undergo

high-yielding and selective additions to both aldehydes and

ketones under catalysis by Lewis bases, primarily chiral

phosphoramides or N-oxides. In addition to a thorough explora-

6

sought to introduce a new paradigm for catalysis of aldol

additions by harnessing the concept of “Lewis base activation

of Lewis acids”. In the intervening years, the theoretical

foundations, methodological development, and generality of this

concept have been extensively documented, primarily as il-

1

tion of reaction scope, extensive mechanistic investigations

revealed the following generalizations: (1) the reaction displays

ﬁrst-order dependence on enoxytrichlorosilane and aldehyde,

and ﬁrst- or second-order dependence on catalyst, (2) the

addition takes place via the simultaneous operation of two

mechanistic pathways involving either one or two phosphor-

amides bound to a pentacoordinate siliconium ion organizational

center, (3) the aldol addition occurs through the reversible albeit

unfavorable formation of an activated complex, and (4) the

turnover-limiting and stereodetermining step is the aldol addi-

tion. Finally, the effects of catalyst loading, rate of addition,

solvents, and additives have been studied and together allow

the formulation of a uniﬁed mechanistic picture for the aldol

addition, Figure 1.

7

lustrated in manifold applications of aldolization reactions. This

report discloses our efforts to shed light on the origin of the

remarkable catalytic effect of Lewis base activated Lewis acids

by laying a concrete mechanistic foundation. We describe herein:

(

1) spectroscopic studies on the interaction of the Lewis bases

phosphoramides) with the weak Lewis acid, silicon tetra-

chloride, (2) establishment of the resting states of the adducts,

and (3) the full kinetic analysis of the catalytic reactions. These

studies have allowed the formulation of a uniﬁed theory of

catalysis that has implications beyond the aldol addition reaction.

Background

1.2. Enoxytrialkylsilanes. In recent years, all of the method-

ological efforts in this area have focused on the use of Lewis

base activated Lewis acids to effect highly general and selective

1

. Summary of Catalytic Enantioselective Aldolization

with Chiral Lewis Bases. 1.1. Enoxytrichlorosilanes. The origi-

nal incarnation of the Lewis base catalyzed aldol addition

12

Mukaiyama-type directed aldol additions of enoxysilane

13

14

9a,15

16

derivatives of esters, nitriles, ketones,

aldehydes,

17

18

unsaturated esters, ketones, and amides. All of these reactions

(

6) Denmark, S. E.; Winter, S. B. D.; Su, X.; Wong, K.-T. J. Am. Chem.

Soc. 1996, 118, 7404–7405.

7) (a) Denmark, S. E.; Beutner, G. L. Angew. Chem., Int. Ed. 2008, 47,

(11) (a) Denmark, S. E.; Pham, S. M.; Stavenger, R. A.; Su, X.; Wong,

K.-T.; Nishigaichi, Y. J. Org. Chem. 2006, 71, 3904–3022. (b)

Denmark, S. E.; Bui, T. J. Org. Chem. 2005, 70, 10393–10399. (c)

Denmark, S. E.; Pham, S. M. HelV. Chim. Acta 2000, 83, 1846–1853.

(d) Denmark, S. E.; Su, X. Tetrahedron 1999, 55, 8727–8738.

(12) (a) Mukaiyama, T. Org. React. 1982, 28, 203–331. (b) Mukaiyama,

T.; Kobayashi, S. Org. React. 1994, 46, 1–103. (c) Mukaiyama, T.;

Matsuo, T. In Modern Aldol Reactions; Mahrwald, R., Ed.; Wiley-

VCH: Weinheim, 2004; Vol. 1, Chapt. 3.

1

560–1638. (b) Denmark, S. E.; Fujimori, S. In Modern Aldol

Reactions; Mahrwald, R., Ed.; Wiley-VCH: Weinheim, 2004; Vol. 2,

Chapt. 7. (c) Denmark, S. E.; Stavenger, R. A. Acc. Chem. Res. 2000,

3

3, 432.

(

8) (a) Denmark, S. E.; Fan, Y. J. Am. Chem. Soc. 2002, 124, 4233–

4

2

235. (b) Denmark, S. E.; Fan, Y.; Eastgate, M. D. J. Org. Chem.

005, 70, 5235–5248.

9) (a) Denmark, S. E.; Fujimori, S.; Pham, S. M. J. Org. Chem. 2005,

7

0, 10823–10840. (b) Denmark, S. E.; Pham, S. M. J. Org. Chem.

003, 68, 5045–5055. (c) Denmark, S. E.; Stavenger, R. A.; Wong,

(13) (a) Denmark, S. E.; Beutner, G. L.; Wynn, T.; Eastgate, M. D. J. Am.

Chem. Soc. 2005, 127, 3774–3789. (b) Denmark, S. E.; Chung, W.-j.

J. Org. Chem. 2008, 73, 4582–4595.

2

K.-T.; Su, X. J. Am. Chem. Soc. 1999, 121, 4982–4991. (d) Denmark,

S. E.; Stavenger, R. A. J. Am. Chem. Soc. 2000, 122, 8837–8847. (e)

Denmark, S. E.; Stavenger, R. A.; Wong, K.-T. Tetrahedron 1998,

(14) Denmark, S. E.; Wilson, T. W.; Burk, M. T.; Heemstra, J. R., Jr. J. Am.

Chem. Soc. 2007, 129, 14864–14865.

5

4, 10389–10402. (f) Denmark, S. E.; Su, X.; Nishigaichi, Y. J. Am.

Chem. Soc. 1998, 120, 12990–12991.

10) (a) Denmark, S. E.; Ghosh, S. K. Tetrahedron 2007, 63, 8636–8644.

(15) Denmark, S. E.; Heemstra, J. R., Jr. Org. Lett. 2003, 5, 2303–2306.

(16) Denmark, S. E.; Bui, T. J. Org. Chem. 2005, 70, 10190–10193.

(17) Denmark, S. E.; Beutner, G. L. J. Am. Chem. Soc. 2003, 125, 7800–

7801.

(

b) Denmark, S. E.; Bui, T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,

5

439–5444. (c) Denmark, S. E.; Ghosh, S. K. Angew. Chem., Int. Ed.

001, 40, 4759–4762.

(18) Denmark, S. E.; Heemstra, J. R., Jr. J. Org. Chem. 2007, 72, 5668–

5688.

2

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11771

A R T I C L E S

Denmark et al.

share a common pathway in which an in situ-generated chiral

Lewis acid catalyzes the addition of one of the above-mentioned

nucleophiles to a range of aldehydes. The conceptual basis for

the activation of the nascent Lewis acid (silicon tetrachloride)

derives from the rehybridization of the Lewis acid-base adduct,

spectrometer and ready for acquisition, all while coordinating

the injection event with data acquisition. Care must then be taken

to ensure that the data are collected in a manner that produces

integral values that are not only internally consistent (resonance

to resonance within a single spectrum) but also consistent among

the plethora of spectra collected over the course of an individual

experiment. To this end, we have developed a rapid-injection

7a

Scheme 1, and has been discussed in detail elsewhere. During

the development of the various aldol additions catalyzed by this

unique Lewis acid, a working hypothesis for the mechanism of

the process was formulated that borrowed heavily from our

understanding of the cognate reactions of enoxytrichlorosilanes

described above. However, from simple empirical analysis of

the rates, selectivities, substrate scope, and response to experi-

mental variables, it was clear that the two manifolds were not

completely congruent. In addition, no deﬁnitive mechanistic

studies supported the validity of those analogies. Accordingly,

we undertook the challenge of placing this newer family of

catalyzed aldol additions on the same concrete foundation that

now exists for the enoxytrichlorosilane additions.

1

system that, when coupled with a high-sensitivity H NMR

2

0

probe, allows us to accurately inject a known volume of a

reagent solution into a sample tube inside the magnet (and

controlled by the spectrometer console), even while the tube is

2

1

spinning at -78 °C.

This system allows reactions to be monitored under conditions

that are identical to, or very closely resemble, those employed

in the preparative system, minimizing the danger that the results

will not be representative of those from preparative reactions.

In our case, four minor adjustments were made to the preparative

system to allow the reactions of interest to be studied repro-

ducibly: (1) the overall concentration (0.076 M) was dropped

to one-third that of the preparative conditions, (2) chromium(III)

Scheme 1

2

3

tris(dipivaloylmethane) [Cr(dpm) , 5 mM] was added as a

paramagnetic relaxation agent to allow a more rapid acquisition

cycle while ensuring integral accuracy, (3) the injected reagent

was dissolved in CDCl , not CD Cl , and (4) the reactions were

3 2 2

performed at -60 °C. For these studies, 1-naphthaldehyde was

chosen as the aldehyde component due to the exquisite

selectivities observed with this substrate, as well as its lower

rate of reaction.

Employing this system to study the asymmetric Lewis base

catalyzed aldol reaction allowed us to study the reaction under

the preparative conditions. That is, each sample tube can be

2

. Objectives of This Study. To investigate the validity of

2

3

the working hypothesis outlined in Scheme 1, the following

reaction parameters needed clariﬁcation: (1) the effect on

reaction rate and enantioselectivity of different catalyst struc-

tures, (2) the molecularity of the transition structure, (3) the

stoichiometry, structure, and equilibrium position of the Lewis

base-Lewis acid association, and (4) the overall rate equation

and partial orders with respect to all reaction components.

Ultimately, our goal was to formulate a uniﬁed mechanistic

picture that could explain the origin of catalysis and selectivity.

We describe herein the successful realization of the ﬁrst of these

two highly ambitious goals. The second, i.e., rationalization of

selectivity, will require extensive computational analysis of the

structures that have been established in this study.

ﬁlled with all of the reagents (1-naphthaldehyde, SiCl

DIPEA, and the Lewis basic catalyst) dissolved in CD Cl with

added Cr(dpm) , Scheme 2. All concentrations but one were

4

,

2

3

held constant, while the concentration of the reagent being

studied was varied from 0.25, 0.5, 1, and 2 times the normal

reaction concentration (for an 8-fold range). All of the samples

were stored under an inert atmosphere at -78 °C, before being

placed into the precooled (-60 °C) spectrometer. Once the

sample was in the magnet, the injector was inserted into the

sample tube while it was spinning, and the entire apparatus was

temperature equilibrated for a period of 5-10 min, which

allowed the sample in the injector to cool. A vigorous ﬂow of

nitrogen (between 10 and 20 L/min) through the probe was

maintained to ensure rapid equilibration. The magnetic ﬁeld was

shimmed in the normal fashion, the spectrometer was pro-

3

. Challenges and Solutions: Development and Imple-

mentation of RINMR Spectroscopic Analysis. Monitoring the

progress of a dilute (ca. 40 mM) reaction that has a half-life of

less than 90 s at -78 °C is a daunting task. Only spectroscopic

(20) The temperature range is determined by the conditions that the probe

can support. The Varian 10 mm broadband probe used in these studies

has operating limits of -130 and +100 °C.

1

9

methods are viable at these temperatures. Although IR

spectroscopy is popular for reaction monitoring, at concentra-

tions below 0.1 M, the low sensitivity of this technique is

prohibitive. Monitoring the progress of these reactions by NMR

spectroscopy is ideal as it allows exquisite sensitivity, the ability

to operate over a wide range of temperatures, and the observa-

tion of any well-resolved resonance(s), whether correlated to

starting material, intermediate, or product.

(

21) The details of the design, implementation, calibration, and use of this

system are not the subject of this publication and will be published in

due course. For descriptions and results from pioneering RINMR

studies, see: (a) McGarrity, J. F.; Prodolliet, J.; Smyth, T. Org. Magn.

Reson. 1981, 17, 59. (b) Palmer, C. A.; Ogle, C. A.; Arnett, E. M.

J. Am. Chem. Soc. 1992, 114, 5619. (c) Reetz, M. T.; Raguse, B.;

Marth, C. F.; H u¨ gel, H. M.; Bach, T.; Fox, D. N. A. Tetrahedron

1

992, 48, 5731.

(

22) The additive Cr(dpm)

3

was chosen over the more popular Cr(acac)

3

The difﬁculty in this case is rapidly delivering a known

amount of a temperature-equilibrated reagent into a temperature-

equilibrated substrate while the sample is spinning in the NMR

due to its decreased Lewis acidity, minimizing the chance that it would

interfere with the reaction. Control experiments showed no measurable

inﬂuence of this additive on the rate of reaction, as well as no reactivity

4

in the absence of SiCl (see Supporting Information). Levy, G. C.;

Edlund, U.; Hexem, J. G. J. Magn. Reson. 1975, 19, 259–262.

(

19) Reaction calorimetry at this temperature is challenging and lacks

reproducibility, and ex situ analysis of a quenched aliquot requires

that no signiﬁcant warming of the sample occur before the quench, a

very difﬁcult task in this temperature regime.

(23) Commercial 1-naphthaldehyde is contaminated with ca. 10% 2-naph-

thaldehyde, which reacts faster than 1-naphthaldehyde. The 1-naph-

thaldehyde used in these kinetic studies was puriﬁed to g99.8% purity

before use. Details are in the Supporting Information.

1

1772 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

grammed to automate the injection of a precisely known amount

of the precooled solution of the silyl ketene acetal, and the

sequence was started. After the spectrometer collected ﬁve

spectra without the silyl ketene acetal, the spectrometer would

trigger the injection event and resume acquiring data at set

intervals (in this case, 3-s intervals). Typically, 0.200 mL of

the silyl ketene acetal solution was injected into a 3.0 mL

sample.

that require our attention are magnetization recovery and

appropriate signal collection, both of which are related to the

T

1

relaxation times of the samples.

Optimization of the data collection began by determining the

9

0° pulse (pw90) for a sample at ca. 50% conversion so that

all species, including the product, were observable in solution.

The optimized pw90 value was determined to be 30 µs with a

transmitter power of 60 db. A T

recovery showed that this sample’s longest T

1

determination by inversion

value was

1

Scheme 2

displayed by the aldehyde proton and was on the order of 3 s.

Consequently, the relaxation delay (d1) was set to 15 s minus

the acquisition time (at; d1 ) 15 - at), to allow 5 times the

1

measured T between pulses. Unfortunately, these conditions

did not allow data collection to occur rapidly enough to obtain

good initial reaction rate data. Even at -60 °C, the reaction

progressed too quickly to be monitored under these conditions.

1

Next, methods to shorten the time between pulses were

implemented. Decreasing the pulse width from 90° to 45° would

allow one pulse every 12 rather than every 15 s, a rather

insigniﬁcant change. Reducing the pulse width further would

begin to severely impact the resulting signal-to-noise and was

not considered. Ultimately, the addition of a paramagnetic

Figure 2 shows an excerpt of the H NMR data obtained from

a typical run. The aldehyde (initial concentration, 39 mM)

displays two resonances (δ ) 9.3 and 10.3 ppm) that are well-

resolved and that allow the concentration change in this starting

material to be easily monitored. Similarly, a resonance corre-

sponding to the product aldolate can be observed (δ ) 6.2 ppm).

Orienting experiments demonstrated that the concentration

decrease displayed by both the aldehyde and silyl ketene acetal

directly led to a corresponding increase in the aldolate; no

intermediate species were observed. In practice, the disappear-

ance of 1-naphthaldehyde was routinely monitored.

relaxation agent, chromium(III) dipivaloylmethane [Cr(dpm)

3

],

, Cr(dpm)

times, but it is less Lewis acidic than Cr(acac)

25

gave the desired results. Like its congener Cr(acac)

reduces T

3

1

3

because of the increased steric bulk about the central chromium

atom.

A great deal of mechanical and electrical optimization of the

injection system was performed during its design and initial

testing to ensure reproducibility and accuracy of the obtained

The optimal concentration of Cr(dpm)

gated. As discussed above, in the absence of Cr(dpm)

3

was brieﬂy investi-

, the

was 3 s, requiring 15 s between pulses. At 1 mM in

3

longest T

Cr(dpm)

pulses. At 5 mM in Cr(dpm)

1

1c

data. Although some of these details have been reported,

a

3

, the longest T

1

dropped to 1.4 s, requiring 7 s between

, the longest T was 0.5 s, requiring

2.5 s between pulses. We decided to use 5 mM Cr(dpm) and

a total time between pulses of 3 s. A control experiment with

5 mM Cr(dpm) but without SiCl showed no consumption of

either the aldehyde or silyl ketene acetal over 20 min at -60

°C. A typical SiCl -catalyzed reaction was complete within this

full account will be published in a more suitable venue.

Optimization of the NMR parameters themselves is critical to

the collection of reproducibly accurate integral data and must

be discussed in some detail.

3

1

3

4

The techniques necessary to collect quantitative NMR data

2

4

are well-known and well-developed. The two main factors

4

Figure 2. A portion of the NMR spectral data collected over the course of 900 s. Data were collected every 3 s, but only every third spectrum (every 9 s)

is shown here for clarity. The disappearance of 1-naphthaldehyde is clear (δ ) 9.25 and 10.3 ppm), as is the appearance of the product aldolate (δ ) 6.2

ppm).

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11773

A R T I C L E S

Denmark et al.

time frame. This observation, combined with the similarities in

Scheme 3

3

observed rates of reaction with and without the added Cr(dpm) ,

provides convincing evidence that the relaxation agent does not

affect the outcome or rate of the reaction. In practice, a 3-s

data acquisition without a relaxation delay between pulses was

1

employed. This setup provides 6 times the longest T value

between pulses and also ensures that all of the available signal

has been collected. These parameters allowed the collection of

very reproducible data sets and were used throughout the study.

After data collection was complete, integral regions were cut

to include 5 times the line width, and an automated routine was

employed to tabulate the integral data from all of the spectra

collected (measurements were made on 3-s intervals until 20%

conversion or, for slow reactions, 1200 s).

When combined, the addition of the relaxation agent, the

optimized data collection, and integral determination routines

allowed the collection of exquisitely reproducible data sets.

Results

1

. Enantioselective Aldol Additions. The reactions studied in

Table 1. Catalyzed Additions of 1 to 2 with Chiral Lewis Bases

Scheme 3)

this work are the enantio- and diastereoselective aldol addition

reactions between silyl ketene acetals (1a/1b) and benzaldehyde

2a) or 1-naphthaldehyde (2b), Scheme 3. The aldol additions

(

entry ketene acetal catalyst (mol %) time, h product temp, °C yield, %

er

(

1

2

3

4

5

6

7

8

9

1

1a

1b

5 (5.0)

1

3a

3b

-78

94

92

97

90

89:11

91:9

96:4

of both acetate- and propanoate-derived silyl ketene acetals were

studied with the two different aldehydes to demonstrate the

generality of the trends observed in phosphoramide structure.

The full kinetic analysis was carried out only for the reactions

of 1b and 2b. Several phosphoramides were used as catalysts:

hexamethylphosphoric triamide (HMPA), the monophosphor-

amide 5, and the bisphosphoramides 4a-4d wherein the chain

length between the phosphoramides varied from 3 to 6 meth-

ylene units. The results of the catalyst dependence are collected

in Table 1.

4b (2.5)

4c (5.0)

4d (2.5)

HMPA

5 (1.0)

4a (1.0)

4b (1.0)

4c (1.0)

4d (1.0)

a

88:12

3

-78

80

77

83

98

78

77:23

93:7

84:16

97:3

72:28

a

0

a

Data taken from ref 13a.

The addition of 1a to 2a was examined with four different

catalysts (Table 1, entries 1-4). Each phosphoramide catalyzed

the reaction in high yield (89-94%) but with only a modest

effect of catalyst structure on enantioselectivity. These selectivi-

ties closely matched those obtained previously for the addition

of allyltributylstannane to benzaldehyde with SiCl . Similarly,

4

the addition of 1b to 2b proceeded in good yield and with high

anti diastereoselectivity, despite the use of a 9:1 mixture of (E)-

1

27

rosilyl enolates, allylation with allyltrichlorosilanes, and

2

8

epoxide ring-opening with silicon tetrachloride ), important

insights on the nature of phosphoramide·Si interaction were

secured by carrying out nonlinear effect studies. Thus, to glean

as much information from this process as possible, a nonlinear

effect study was undertaken for the addition of 1a to 2a.

Both (R)-5 and (S)-5 were independently prepared by the

previously described method and were assured to be of >99.9%

29

26

2

/(Z)-2. Although we knew that the reaction catalyzed by 4c

30

enantiopurity by CSP-SFC analysis. The enantiopure catalysts

were then combined in nominal ratios to provide enantiomeric

mixtures (60:40, 70:30, 80:20, and 90:10) for the nonlinear

study, and the enantiomeric ratios were independently conﬁrmed

by SFC analysis (59.3:40.7, 68.8:31.2, 78.3:21.6, 88.9:11.1).

The aldol addition was then carried out as described above,

and the enantiopurity of the aldol product was determined by

CSP-SFC analysis. The reactions were carried out in triplicate

was extremely fast, we did not know the rates of the reactions

catalyzed by the other phosphoramides; consequently, all the

reactions were run at -78 °C for 3 h. Interestingly, the

enantioselectivities were signiﬁcantly more sensitive to the tether

length. As has always been observed, 4c (n ) 5) gave the highest

enantioselectivity (97:3). Bisphosphoramides with shorter chain

lengths (4a and 4b) showed moderately lower selectivities (93:7

and 84:16), while the catalyst with a longer chain length (4d)

and monophosphoramide 5 showed signiﬁcantly lower selectivi-

ties (72:28 and 77:23). The greater sensitivity of the 1b/2b

combination to changes in catalyst structure suggested that the

full kinetic analysis of this system would likely be more

informative.

(

yields all >90%), and each product was analyzed three times.

When the averaged results are represented graphically, ee of

(27) Denmark, S. E.; Fu, J.; Coe, D. M.; Su, X.; Pratt, N. E.; Griedel,

B. D. J. Org. Chem. 2006, 71, 1513–1522.

(

28) Denmark, S. E.; Barsanti, P. A.; Beutner, G. L.; Wilson, T. W. AdV.

Synth. Catal. 2007, 349, 567–582.

2

. Nonlinear Effect Studies. In previous mechanistic studies

of Lewis base catalyzed reactions (aldol additions of trichlo-

(29) (a) Guillaneux, D.; Zhao, S.-H.; Samuel, O.; Rainford, D.; Kagan,

H. B. J. Am. Chem. Soc. 1994, 116, 9430–9439. (b) Fenwick, D. R.;

Kagan, H. B. Top. Stereochem. 1999, 22, 257–296. (c) Girard, G. L.;

Kagan, H. B. Angew. Chem., Int. Ed. 1998, 37, 2923–2959. (d) Avalos,

M.; Babiano, R.; Cintas, P.; Jimenez, J. L.; Palacios, J. C. Tetrahedron:

Asymmetry 1997, 8, 2997–3017.

(

24) For example, see: (a) Gerritz, S. W.; Seﬂer, A. M. J. Comb. Chem.

2

000, 2, 39–41. (b) Hoye, T. R.; Eklov, B. M.; Voloshin, M. Org.

Lett. 2004, 6, 2567–2570.

(

25) Stille, D.; Doyle, J. R. Inorg. Synth. 1986, 24, 183–184.

26) Denmark, S. E.; Wynn, T. J. Am. Chem. Soc. 2001, 123, 6199–6200.

(30) Denmark, S. E.; Su, X.; Nishigaichi, Y.; Coe, D. M.; Wong, K.-T. J.

Org. Chem. 1999, 64, 1958.

1

1774 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

the catalyst vs ee of the product, a linear relationship is clearly

observed, Figure 3.

The absence of a nonlinear effect is surprising given that an

effect has been seen in all previous studies. In light of the full

kinetic studies described below, the implications of this outcome

will be deferred to the Discussion section.

The partial reaction order of each reagent was determined

by the initial rates method. The concentration of each individual

reactant was varied over an 8-fold range, while holding all other

reagent concentrations constant and observing the changes in

concentration over the ﬁrst 5-20% conversion. As shown in

Figure 4, both silyl ketene acetal 1b (12, 24, 48, and 96 mM)

and 1-naphthaldehyde (2b) (9.2, 20, 39, and 78 mM) displayed

ﬁrst-order behaviors. The reaction orders for silicon tetrachloride

and the bisphosphoramide catalyst 4c were more surprising.

Varying the concentration of silicon tetrachloride (44, 87, 174,

and 348 mM) had no effect upon the rate of the reaction, while

the partial reaction order in 4c (0.18, 0.36, 0.72, and 1.43 mM)

was 0.5, Figure 5. Taken together, these two results provide

the greatest insight into the structure of the catalyst resting state

and the active catalytic species. However, the complete picture

still requires the analysis of the other catalysts, as described

below.

3

. RINMR Kinetic Analysis of the Addition of 1b to 2b.

3.1. Partial Order in Each Reagent with Catalyst 4c. Optimiza-

tion of the NMR parameters occurred before beginning to

measure rates of reactions and was discussed above. It was

decided early on to monitor the disappearance of aldehyde, as

the two well-resolved resonances at δ ) 9.3 and 10.3 ppm could

be monitored and averaged to measure the changes in concen-

tration for each run.

Although not contributors to the stoichiometric process, the

partial orders in all of the other reaction components were also

obtained. Varying the concentration of diisopropylethylamine

(

3

added as an acid scavenger) and Cr(dpm ) (the added para-

magnetic relaxation agent) had no effect upon the rate of

reaction. Interestingly, varying the concentration of chloride by

the addition of tetrabutylammonium chloride (4, 8,16, and 32

mM) gave rise to a negative fractional order (-0.2), whereas

the addition of tetrabutylammonium triﬂate (4, 8, 16, and 32

mM) did not affect the rate of reaction.

3

.2. Partial Order in Catalysts. 3.2.1. HMPA. Because the

complexation behavior of HMPA and SiCl has been studied

in depth, and HMPA is known to be an active, albeit sluggish

4

31

Figure 3. Nonlinear effect study of acetate aldolization with catalyst 5.

Figure 4. Partial reaction order in silyl ketene acetal 1b and 1-naphthaldehdye (2b).

4

Figure 5. Partial reaction order in SiCl and 4c.

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11775

A R T I C L E S

Denmark et al.

Figure 6. Partial reaction order in HMPA.

Figure 7. Partial reaction order in monophosphoramide 5.

catalyst for this aldol reaction, learning how varying its

concentration affects the rate of reaction would be instructive.

Indeed, following the analysis above, HMPA (4, 9, 17, and 35

mM) displayed an interesting fractional order dependence (0.64),

Figure 6. This result was most certainly not expected. If the

working hypothesis for the catalytic cycle outlined above was

correct, HMPA should have displayed a second-order behavior.

The approximate two-thirds order is intriguing and will be

discussed below.

resting state. Thus, to provide insight into the structure and

composition of the species formed by combining the phos-

31

phoramides with SiCl

4

, a series of ²⁹Si and P NMR spectro-

scopic experiments were performed with mixtures of each of

the catalysts 4a-4d and 5 together with SiCl . The species

4

detected under these conditions are not rigorously the catalyst

resting states because no catalytic process is operating. However,

the rate and concentration of the catalytic reactions preclude

any chance of obtaining information at natural abundance of

29

Si. Thus, all conclusions drawn from these studies must be

3

.2.2. Monophosphoramide 5. Although catalyst 5 afforded

tempered with that caveat. Nevertheless, very useful information

could be gathered that merits presentation and analysis.

NMR spectra recorded at room temperature showed that the

species observed in all of these mixtures were ﬂuxional. When

the spectra were recorded between -70 and -100 °C, sharp, well-

low enantioselectivities in the aldol additions, it is nonetheless

of mechanistic interest. In previous studies of the reactions of

enoxytrichlorosilanes and allyltrichlorosilanes, two independent

mechanistic pathways involving one and two phosphoramides

could be discerned. The latter, higher order pathway afforded

greater rates and selectivities (which stimulated the development

of the bisphosphoramides). Surprisingly, this catalyst displayed

a ﬁrst-order dependence when its concentration was varied (1.6,

resolved signals could be observed. Again, Cr(dpm)

3

was added

29

to the samples to reduce the long Si T

1

times. Silicon-29 chemical

shifts are highly dependent on coordination number and fall into

three major regionssfour coordinate (+10 to -30 ppm), ﬁve

coordinate (-50 to -120 ppm), and six coordinate (-150 to -200

ppm); thus, Si-29 spectra are ideally suited for the analysis of the

3.2, 6.3, and 12.7 mM, Figure 7). As was the case with HMPA,

this result is an interesting departure from the expected second-

order behavior. Any proposal for a catalytic cycle will have to

explain why each of the catalysts discussed so far displays

different partial reaction orders.

32

possible coordinate structures of the SiCl

4

catalyst.

4

.1. HMPA/SiCl . Previous investigations of the SiCl /HMPA

4

system have shown that even when HMPA is present in

substoichiometric amounts, three major species are present in

solution, Figure 9: a cationic, ﬁve-coordinate complex, [trans-

3

.2.3. Bisphosphoramide Catalysts 4a, 4b, and 4d. As noted

previously, catalyst 4c, with a ﬁve-methylene chain length,

showed 0.5 partial reaction order. Increasing the chain length

to six methylenes gave the same 0.5 partial reaction order for

catalyst 4d, Figure 8, with similar initial rates. Unexpectedly,

by decreasing the chain length to three and four methylenes

+

-

33

HMPA

coordinate complexes, one cationic, [mer-HMPA

-206 ppm, s), and one neutral, [trans-HMPA ·SiCl

2

·SiCl ] Cl (-120.5 ppm, J ) 15 Hz), and two six-

3

+

-

3

SiCl

3

] Cl

(

2

4

] (-207.8,

3

1

s), Scheme 4. This observation reveals the strong thermody-

namic preference for the formation of these complexes. In fact,

no free HMPA was observed in solution until more than 3.0

equiv of HMPA had been added. Moreover, no species with

a single bound phosphoramide was identiﬁed; all of the species

observed in solution had either two or three phosphoramides

bound to a central silicon atom.

(

4a and 4b), the partial reaction order changed to unity! These

catalysts also showed remarkably faster initial rates than 4c and

d. In fact, for catalyst 4a the initial rates were almost too fast

for the kinetic studies (50% conversion in 300 s with 0.08 mol

4a), and some caution must be exercised in interpreting the

4

3

4

%

results for this catalyst. Catalyst 4b showed slightly slower initial

rates (but still much faster than 4c and 4d), and its results can

be interpreted with a much higher degree of conﬁdence. It should

(

32) (a) Kennedy, J. D.; McFarlane, W. In Multinuclear NMR; Mason, J.,

Ed.; Plenum Press: New York, 1987; Chapt. 11. (b) Kobayashi, S.;

Nishio, K. Tetrahedron Lett. 1993, 34, 3453. (c) Olsson, L.; Ottosson,

C. H.; Cremer, D. J. Am. Chem. Soc. 1995, 117, 7460. (d) Arshadi,

M.; Johnels, D.; Edlund, U.; Ottoson, C.-H.; Cremer, D. J. Am. Chem.

Soc. 1996, 118, 5120.

be noted that the partial orders in 1b, 1-naphthaldehyde, SiCl

diisopropylethylamine, and Cr(dpm) were obtained with catalyst

c only. It is assumed that the orders in these reagents are

preserved across the entire series of catalysts.

. Catalyst Resting State. The interpretation of the rate data

presented above is intimately tied to the identity of the catalyst

4

,

3

4

31

(

33) In our earlier publication, this resonance was erroneously assigned

to be the cis complex, which is inconsistent with the 15 Hz coupling

constant.

34) Conversely, free SiCl

until 3.0 equiv of HMPA had been added.

was always observed in the ²⁹Si NMR spectrum

4

(

31) Denmark, S. E.; Eklov, B. M. Chem. Eur. J. 2008, 14, 234–239.

1776 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

1

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

Figure 9. ²⁹Si NMR spectra of a mixture of SiCl

at -70 °C.

4

2 2

and HMPA in CD Cl

-

205.5 (t, J ) 15 Hz) and -207.0 ppm (s) and two broad signals

at -205 and -206.5 ppm. The species corresponding to the triplet

resonance at -205.5 ppm is deﬁnitively doubly ligated, but the

coordination environment of species corresponding to the broad

singlet at -207.0 ppm is less well-deﬁned. On the basis of the

coupling constants, the signal at -205.5 ppm is assigned to the

trans-SiCl

4

·5

2

complex and that at 207.0 to cis-SiCl

4

2

·5 . The ill-

deﬁned signals are also clearly six-coordinate species and thus could

be the trisphosphoramide-bound cations that were identiﬁed in our

studies with HMPA and SiCl .

4

In view of the lack of a nonlinear effect in reactions catalyzed

by 5, the composition of the complexes formed from a nearly

racemic mixture of 5 was of interest. The spectra shown in the

2

9

bottom of the two chemical shift regimes in Figures 10 ( Si

Figure 8. Partial reaction order in bisphosphoramides 4d, 4a, and 4b.

31

NMR) and 11 ( P NMR) display one additional signal each.

Thus, both cationic, ﬁve-coordinate and neutral, six-coordinate

compounds can exist as roughly equal mixtures of homo- and

heterochiral complexes.

Scheme 4

4

.3. Bisphosphoramide 4c/SiCl

analysis of a 2.7:1 mixture of SiCl

also showed three phosphoramide-bound species in addition to

free SiCl (-19 ppm), Figure 12. All three species appear as

triplets and therefore are all coordinated to two phosphorus

4

. ²⁹Si NMR spectroscopic

and 4c in CD Cl solution

4

2

4

3

5

moieties. The ﬁrst species (δ ) -118.4 ppm (t, J ) 14 Hz))

falls in a region that is indicative of a ﬁve-coordinate silicon

complex. Because the coupling constant is nearly identical to

that seen in the ﬁve-coordinate complex with the monophos-

phoramide 5 (and is consistent with a trans arrangement of the

ligands), the signal is assigned to the dimeric trichlorosilyl

4

.2. Monophosphoramide 5/SiCl

4

. The ²⁹Si NMR spectro-

and 5 in CD Cl solution

scopic analysis of a 4:1 mixture of SiCl

showed a number of phosphoramide-bound species, Figure 10. The

ﬁrst resonance appears at δ ) -118.8 ppm (t, J ) 15 Hz), which

is within the ﬁve-coordinate region of the Si spectrum, and must

be bound by two phosphoramides as seen in the triplet coupling

pattern. This signal is assigned to the cationic species [trans-

4

2

29

(

35) Since the phosphoramide/silicon ratios of these three species are the

same, the relative abundances will not change with the concentration

of either phosphoramide or silicon.

+

-

SiCl

3

·5

2

] Cl , Scheme 5. The other peaks fall into the six-

coordinate regime, of which the two major signals appear at δ )

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11777

A R T I C L E S

Denmark et al.

Scheme 5

Figure 10. Portions of the ²⁹Si NMR spectra of SiCl

gave an apparent diastereomer ratio of ca. 1:1.

4

2 2

and 5 in CD Cl at -70 °C. (Top) Enantiopure (R)-5. (Bottom) 54:46 mixture of (R)- and (S)-5, which

Figure 11. ³¹P NMR spectra of SiCl

and 5 in CD Cl at -70 °C. (Top)

4 2 2

Enantiopure (R)-5. (Bottom) 54:46 mixture of (R)- and (S)-5, which gave

an apparent diastereomer ratio of ca. 1:1.

2

+

-

cation, [trans,trans-SiCl

3

·4c]

2

2Cl (see Figure 14). The other

two phosphorus-bound species fall within the six-coordinate

regime at δ ) -204.8 (t, J ) 15 Hz) and -205.1 ppm (t, J )

5

Hz). These two species are each bound to two identical

phosphorus atoms with different geometrical arrangements

around the silicon octahedron.

The ³¹P NMR spectrum of this mixture is also informative,

Figure 13. First, even at this ratio (2.7 SiCl

4

/4c) only a trace of

Figure 12. ²⁹Si NMR spectra of SiCl

and 4c in CD Cl at -70 °C.

4 2 2

4

c is observed, showing the high thermodynamic preference

for complexation of the Lewis base. Second, four discrete

signals, the resonance at -204.9 ppm is assigned to the

[trans,trans-SiCl ·4c] complex and the resonance at -205.2

ppm to the [cis,cis-SiCl · 4c] complex (see Figure 14). The

mixed dimer [trans,cis-SiCl ·4c] would display equal intensity

signals for the two silicon nuclei and can therefore be eliminated.

The expected monomeric chelate cis-SiCl ·4c can be ruled out

by the magnitude of the coupling constant. Neutral bis-

phosphoramide-SiCl complexes display a strong preference

to maintain a trans geometry that would disfavor the formation

resonances are visible in the chemical regime of complexed

4

2

31

phosphoramides. From these data, four limiting structures for

neutral, six-coordinate complexes can be considered, Figure 14.

Although the spectroscopic data alone cannot distinguish

between monomeric and dimeric complexes, the dimeric

complexes would explain the half-order kinetic behavior of 4c

4

2

4

2

4

(

vide supra). On the basis of the magnitude of the two coupling

4

3

constants ( J31P,29Si) and the nonequivalent intensities of the two

1

1778 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

³J31P-29Si (14 Hz) can rule out any cis-conﬁgured complexes of

this type as well as the phosphoramide-bridged structure 7. In

addition, 7 is asymmetric, so in the absence of rapid ligand

topoisomerization at -70 °C, this species can also be ruled out.

4

.4. Bisphosphoramide 4d/SiCl

4

. The ²⁹Si NMR spectrum of

Cl

a 3.0:1 mixture of SiCl to bisphosphoramide 4d in CD

4

2

solution, Figure 15, looked similar to that from bisphosphor-

amide 4c but differed markedly in the ratios of the signals. As

before, a strong signal at δ ) -118.6 ppm (t, J ) 14 Hz)

revealed the dominant presence of a cationic, ﬁve-coordinate

complex. Two additional, weak and featureless signals at δ )

-

178 and -206 ppm suggest the mutual coexistence of neutral,

six-coordinate complexes. The splitting patterns of the six-

coordinate species were hard to discern. Interestingly, the

relative amounts of these species were different for 4d and 4c,

such that bisphosphoramide 4d clearly favored a ﬁve-coordinate,

cationic resting state. Also, for 4d a single peak is dominant in

Figure 13. ³¹P NMR spectra of SiCl

and 4c in CD Cl at -70 °C.

4 2 2

of cis-SiCl

reconciled with the data from the Si spectrum. The dimeric

complexes in Figure 14 have nominal D symmetry, and thus

all the phosphorus nuclei are homotopic. However, we cannot

rule out ground state conformations of lower symmetry.

Therefore, combining the kinetic data with the spectroscopic

observations, all of the silicon species are proposed to be bridged

dimers.

4

·4c. The observation of four ³¹P signals is not easily

29

31

the P spectrum together with a larger amount of free 4d, Figure

1

6. As with 4c, the structure is assigned to a cyclic, dimeric,

3

+

-

ﬁve-coordinate cation, [trans,trans-SiCl

3

·4d]

2

2Cl .

4

.5. Bisphosphoramides 4a/SiCl and 4b/SiCl

4 4

. The ²⁹Si

spectra of the shorter tethered bisphosphoramides 4a and 4b

were signiﬁcantly different compared to those of 4c and 4d.

The spectra from the n ) 4 tethered 4b were of slightly better

quality and will be discussed ﬁrst.

The available data cannot eliminate the closely related bridged

2

+

-

The ²⁹Si spectra of 4b in the presence of 3.0 equiv of SiCl

structure [trans,trans·6] 2Cl . This chloride-bridged species

also has D symmetry and is thus a possible candidate for the

resonance at -118.5 ppm. However, the magnitude of the

4

3

contained no species in the ﬁve-coordinate regime (-115 to

-125 ppm), but a signal at δ ) -185.5 ppm (t, J ) 2 Hz)

4

Figure 14. Proposed structures for the complexes of 4c with SiCl .

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11779

A R T I C L E S

Denmark et al.

Figure 15. Portions of the ²⁹Si NMR spectra of SiCl

4

2 2

and 4d in CD Cl at -70 °C.

Figure 16. ³¹P NMR spectra of SiCl

and 4d in CD Cl at -70 °C.

4 2 2

Figure 17. Portions of the ²⁹Si NMR spectra of SiCl

70 °C.

4

2 2

and 4b in CD Cl at

-

appeared along with several peaks between δ ) -204 and

-

206 ppm, Figure 17. The signal at δ ) -185.5 ppm in the

2

9

Si spectrum is assigned to the cis-chelated, monomeric,

neutral, six-coordinate species cis-SiCl · 4b. The assignment

4

of the signals in the six-coordinate region is more ambiguous,

particularly because their splitting patterns cannot be dis-

cerned, but they are tentatively assigned to cyclic dimers

[

trans,trans-SiCl

4

· 4b]

2

and [cis,cis-SiCl

4

· 4b]

2

, similar to

is very

3

1

those for 4c. The P spectrum of 4b with SiCl

4

complex, and at this time, assignment of the individual signals

to speciﬁc structures is not possible, Figure 18 (see Figure

1

9 for proposed structures).

29

Catalyst 4a showed a Si NMR spectrum similar to that from

b: a signal at δ ) -186.3 ppm (broad singlet) and two signals

4

3

6

between δ ) -205.4 and -206.0 ppm, Figure 20. The

structural assignments are the same as those for 4b. The broad

singlet at δ ) -186.3 ppm would likely show a triplet splitting

with better resolution. The coupling constant must be smaller

than 2 Hz, that of 4b, which is indicative of a smaller bite angle

Figure 18. ³¹P NMR spectra of SiCl

and 4b in CD Cl at -70 °C.

4 2 2

the concentration of added base nor Cr(dpm) was noted. Most

3

interesting, however, was the dependence on the catalyst.

31

in the chelated structure. The P NMR spectrum of this mixture

is also too complex to be interpreted, Figure 21.

HMPA showed a two-thirds order, with a slow overall rate and

a resting state composed of a cationic 3:1 complex of

+

-

[

HMPA

3

·SiCl

3

] Cl and hexacoordinate neutral 2:1 complexes

Discussion

2

·SiCl

4

]. The monophosphoramide 5, also a slow-acting

1

. Brief Summary of Results. To facilitate the following

catalyst, displayed a ﬁrst-order kinetic dependence and a resting

state comprising an equilibrium mixture of doubly ligated,

cationic, ﬁve-coordinate and doubly ligated, neutral, six-

coordinate silicon species. Among the tethered bisphosphor-

amides, 4c, the catalyst of choice for maximizing enantio-

selectivity, showed a rapid rate of reaction, with a half-order

kinetic dependence. The resting state is comprised of a

mixture of two neutral, six-coordinate complexes with each

discussion of the mechanism and catalytic cycles, a brief

summary of both the kinetic experiments and resting-state

observations would be helpful. The aldol reaction (with catalyst

4c) displayed ﬁrst-order dependences on both aldehyde and silyl

ketene acetal but zero-order dependence on SiCl . A small

4

inverse dependence on added chloride ion was noted, but no

dependence at all on added triﬂate ion. No rate dependence on

1

1780 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

Figure 20. Portions of the ²⁹Si NMR spectra of SiCl

70 °C.

4

2 2

and 4a in CD Cl at

-

4

Figure 19. Proposed structures for the complexes of 4b with SiCl .

Figure 21. ³¹P NMR spectra of SiCl

and 4a in CD Cl at -70 °C.

4 2 2

silicon center bound to two phosphoramides along with

a small amount of the doubly ligated, cationic, ﬁve-coordinate

species, which is proposed to be the catalytically relevant

species. Similarly, 4d mimics the overall rate and half-order

kinetic dependence observed for 4c, but the doubly ligated,

cationic, ﬁve-coordinate resting structure predominates. Phos-

phoramides 4a and 4b are the most powerful catalysts,

providing much faster overall rates of reaction (albeit with

attenuated enantioselectivities) and ﬁrst-order kinetic de-

pendencies. The resting states are mixtures of monomeric

and dimeric neutral, six-coordinate complexes.

Figure 13. Under the initial rate (and preparative reaction)

conditions, the concentration of SiCl is always much larger

4

than the concentration of the Lewis base. Therefore, since all

of the available phosphoramide is bound in some form with

SiCl

4

, the rate of the reaction is dependent not upon the

but instead upon the concentration of

concentration of SiCl

4

the Lewis base, which in turn dictates the concentration of the

active, cationic, catalytic complex.

The results of the control experiments were all anticipated.

Varying the concentration of diisopropylethylamine or

Cr(dpm) demonstrates that their addition does not affect the

3

outcome of the reaction. Both had a beneﬁcial effect upon

reproducibility, however, as addition of the base prevented

2. Partial Order in Stoichiometric Reactants. The ﬁrst-order

behavior for both the silyl ketene acetal 1b and 1-naphthalde-

hyde is not surprising for a bimolecular reaction and demon-

strates that the rate-limiting step is either the aldolization event,

a desilylation event, or a catalyst turnover event. While the latter

two cannot be ruled out by these data, we believe the aldolization

step is turnover-limiting. Both of the post-aldolization steps are

3

8

the Brønsted acid catalyzed aldol reaction from occurring,

while addition of the chromium reagent decreased the

relaxation delays required to repeatedly obtain quantitative

integration data.

3

7

believed to be rapid when compared to the aldolization step.

Addition of tetrabutylammonium triﬂate or chloride was

performed to probe the proposed cationic nature of the catalyst.

If the catalyst is cationic, the addition of exogenous chloride

ions should inhibit the reaction, while the addition of triﬂate

should not affect the rate of reaction. Indeed, the addition of

tetrabutylammonium chloride does inhibit the reaction, and this

inverse partial order in added chloride supports the cationic

nature of the intermediates. The addition of tetrabutylammonium

triﬂate had no effect upon the rate of reaction.

Furthermore, it is difﬁcult to imagine a scenario wherein the

aldolization could be reversible yet still give such high enan-

tioselectivities.

The insensitivity of the reaction rate to the concentration of

SiCl

quantities) provides an important insight into the resting structure

of the phosphoramide catalyst. Zeroth-order behavior in SiCl

implies that 4c is saturated under these conditions, a conclusion

4

(even though this reagent is required in stoichiometric

4

3

1

supported by the lack of free 4c in the P NMR spectrum,

(

36) The small signals at δ ) -180.6 ppm for 4b and δ ) -183.2 ppm

for 4a are too small to be interpreted but may correspond to conformers

(37) Turnover-limiting desilylation is rarely seen. For a thorough discussion

of the consequences of silicon group transfer in Mukaiyama aldol

additions, see ref 1b.

of cis-SiCl

4

·4b and cis-SiCl ·4a frozen out at low temperature.

4

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11781

A R T I C L E S

Denmark et al.

3

. Partial Order in Catalysts and Their Resting Struc-

ﬁve-coordinate and two six-coordinate), each with two bound

phosphoramides. Because the reaction displays ﬁrst-order

kinetic dependence on the phosphoramide, the transition

structure must also contain two molecules of 5. However,

the arrangement of the two ligands in the silicon coordination

sphere cannot be deduced from these data. Interestingly, 5

affords a much lower enantioselectivity compared to 4a-4c,

suggesting that the additional geometrical and conformational

degrees of freedom available to the monomer lead to

kinetically competent but sterically less biased catalysts.

The studies with HMPA conﬁrm our understanding of the

tures. Taken together, these two sets of data, i.e., the kinetic

29

analyses and Si NMR studies of the catalysts, offer the greatest

challenges to interpretation but also provide the most insight.

One of the most important facts required for a detailed

knowledge of a catalytic reaction mechanism is the molecularity

of the catalyst in the turnover-limiting transition structure. This

knowledge can be obtained by determining the resting state of

the catalyst (NGS) and the kinetic order in catalyst. The

molecularity in the transition structure (NTS) is then simply

3

9

expressed in eq 1.

ground- and transition-state structures for 5. The existence of a

^NTS ⁾^NGS × kinetic order

(1)

29

3:1 HMPA/SiCl

4

resting state according to Si NMR and solid-

31

state characterization, combined with the observed two-thirds

kinetic order in HMPA, corroborates the notion that the active

catalytic species is a cationic, ﬁve-coordinate siliconium com-

plex bearing two phosphoramides.

Thus, for the purposes of the following discussion, the

general mechanistic scheme shown in Figure 22 applies. In

this formulation, the ﬁrst steps involve the pre-equilibrium

4

activation of SiCl by the Lewis base (LB) (Figure 22, eq 1)

3

.2. Nonlinear Effects. For a catalytic system in which the

and the activation of the aldehyde by association with the

chiral siliconium complex (Figure 22, eq 2). We posit that

the rate-determining and stereodetermining step is the attack

of the silyl ketene acetal on the activated aldehyde complex

Figure 22, eq 3), which is followed by a rapid desilylation

Figure 22, eq 4). The catalyst turnover step involves the

resting state possesses two catalyst molecules and still shows

ﬁrst-order kinetic dependence on catalyst (i.e., NTS ) 2), the

absence of nonlinear effects can be interpreted a number of

different ways. Because the resting state of the catalytically

active species with the monophosphoramide 5 is doubly ligated,

the ﬁrst-order dependence upon phosphoramide can only be

explained if two phosphoramides are bound in the transition

structure; therefore, a mono-ligated pathway cannot explain the

lack of a nonlinear effect. Two limiting scenarios can be

formulated: either the heterochiral 2:1 complex [(R)-5,(S)-

(

chelation of the trichlorosilyl moiety to form the spectro-

scopically observed end product and release the Lewis base

catalyst (Figure 22, eq 5). The composition of the reactive

+

-

complex [LB

order kinetic dependence on aldehyde and silyl ketene acetal,

2) zeroth-order kinetic dependence on SiCl , and (3) NTS

n 3

SiCl · RCHO] Cl is established by (1) ﬁrst-

+

-

5

·SiCl

3

] Cl does not form (K ) 0), or the hetero- and

(

4

,

+

-

homochiral complexes [(R)-5,(R)-5·SiCl

3

] Cl have identical

the molecularity of LB in the turnover-limiting step. The latter

will be established for each catalyst in the following sections.

29

reactivities (g ) 1). To determine which of these two

possibilities is correct, spectroscopic studies of the resting

structure with nearly racemic mixtures of the monophosphor-

29

amide were carried out. The Si NMR spectra of these mixtures

deﬁnitively demonstrate that both the hetero- and homochiral

complexes form in solution and are present in nearly equal

amounts. Thus, the most reasonable explanation for the lack of

a nonlinear effect is that both the homo- and heterochiral

complexes have identical reactivities. Inspection of species 8,

9

, and 10, Figure 23, illustrates the origin of this behavior. All

four complexes contain a phosphoramide moiety situated trans

to the bound aldehyde and a second phosphoramide located cis

4

0

to the bound aldehyde. In this model, the proximal cis-bound

phosphoramide is responsible for the orientation of the aldehyde

and also controls the topicity of attack on the carbonyl group.

The trans-situated phosphoramide is required for reactivity, but

it is located too far from the aldehyde to have any effect upon

the stereochemical outcome of the reaction. Thus, in the

homochiral complexes 8 and ent-8, the aldehyde is in the same

environment and is equally activated as in the corresponding

heterochiral complexes 9 and 10, respectively.

Figure 22. Generalized mechanism for the aldolization process.

3

.1. Monophosphoramide (5) and HMPA. The mechanism

for aldolization catalyzed by 5 is the easiest to explain and

aids in understanding the behavior of the other catalysts. The

2

9

Si NMR spectrum of 5 with SiCl

4

reveals that the resting

state of the catalyst is comprised of three silicon species (one

Figure 23. Reactive homo- and heterochiral mer-cis complexes of an

aldehyde.

(

38) The preparative reactions do not require the addition of an amine base

if the SiCl is distilled before use. Due to the nature of the RI-NMR

4

3.3. Bisphosphoramides 4a and 4b. Three- and four-meth-

ylene-linked catalysts 4a and 4b share the same kinetic

device, each sample must be momentarily exposed to the ambient

atmosphere, which can be avoided when these reactions are run

preparatively. Consequently, we decided to add the base to remove

any adventitious acid that may be formed from this exposure.

4

behavior and quite similar SiCl complexes, as seen in their

2

9

29

Si NMR spectra. Each species observed in the Si spectra

1

1782 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

is six-coordinate, with two bound phosphoramides. Unlike

the monomer, the bisphosphoramides can chelate the silicon

equation gives eq 2. This equation reduces to eq 3 for low

concentration of chloride, i.e., little ionization, which was

spectroscopically conﬁrmed (no species between δ ) -115 and

-125 ppm). Equation 3 ﬁts the experimental observation of ﬁrst-

order behavior in the catalysts.

2

9

3

atom. On the basis of the Si chemical shift and the J31P,29Si

,

the catalyst ground states are assigned to be cis-chelated,

neutral, six-coordinate 1:1 complexes of the bisphosphor-

4

amides and SiCl . The ﬁrst-order behavior means that this

chelated structure is preserved in the transition structure. This

observation provides further evidence that two phosphor-

amides are bound to silicon in the turnover-limiting transition

structure.

3.4. Bisphosphoramides 4c and 4d. Five- and six-methylene-

linked catalysts 4c and 4d behave the same way and can be

discussed together. Unlike the shorter-chained ligands 4a and

4

b, these phosphoramides prefer to form a ﬁve-coordinate,

doubly ligated, cationic species with SiCl . The half-order kinetic

4

behavior observed for 4c and 4d strongly suggests that a dimeric

resting state undergoes a dissociation leading to a monomeric

reactive complex with the aldehyde; i.e., the ground-state

3

structure contains two bisphosphoramide·SiCl units, and the

transition structure contains one. The simplest way to accom-

modate this mathematical mandate is to propose that the resting-

state structures are cyclic dimers, which break apart to give two

2

9

equivalent active species. From the Si NMR spectra, both

neutral, six-coordinate complexes, [trans,trans-SiCl ·4c] and

cis,cis-SiCl ·4c] , are present but only one cationic, ﬁve-

coordinate complex, [trans,trans-SiCl

4

2

[

4

2

+

-

3

·4c]

2

2Cl . It is this

latter species that is believed to undergo reversible binding of

the aldehyde and dissociation to a reactive monomeric complex,

vide infra.

Figure 24. Derivation of kinetic expression for a monomeric catalyst resting

state, 4a and 4b.

The analysis of 4d follows along similar lines. Of special

note is the relative simplicity of the Si and P spectra for 4d,

where essentially only one major ﬁve-coordinate cationic

rate ) k k k k [SiCl ][RCHO][1b][cat]total^/

2

9

31

1

2

3

4

-

4

-

{

k k k [Cl ] + k k k [Cl ][SiCl ] + k k k [SiCl ] +

-1 -2 -3

1

-2 -3

4

1

2

-3

4

2

+

-

complex, [trans,trans-SiCl

d shows a half-order kinetic dependence, it is proposed that

this species must be dimeric.

.5. Derivation of the Rate Equation for Aldolizations. Given

the knowledge of the resting states and the partial orders in the

various catalysts (as well as the orders in 1b, 2b, SiCl , and

N Cl ), it should be possible to derive a kinetic equation

3

·4d]

2

2Cl , is observed. Because

k k k [SiCl ][RCHO] (2)

1 2 3 4

}

4

-

For very small [Cl ], eq 2 simpliﬁes to

3

k

[RCHO][1b][cat]total

1

2

3

4

rate )

(3)

4

k k k + k k k k [RCHO]

1

2

-3

1 2 3 4

+

-

n-Bu

4

under the assumption of the general mechanistic scheme shown

in Figure 22. Two derivations are needed, one for a monomeric

resting state (for catalysts 4a and 4b) and one for a dimeric

resting state (4c and 4d).

For catalysts 4a and 4b the relevant pre-equilibria, mass

For catalysts 4c and 4d the relevant pre-equilibria, mass balance,

and rate expressions are shown in Figure 25. Solving the equation

for [cat]total gives a complex, ﬁve-term expression that can be

simpliﬁed by making some reasonable assumptions: (1) the term

[cat]_f_r_e_ecan be ignored because no free catalyst 4c or 4d is ever

4

1

3

1

+

balance, and rate expressions are shown in Figure 24. Re-

observed in the P NMR spectra, and (2) the terms [C ] and [D ]

+

29

arranging all of the equilibrium expressions in terms of [C ],

can be ignored because these species are never detected in the Si

+

solving for [C ], and substituting that expression in the rate

NMR spectra or during the analysis of reactions.

From there, three equations can be derived by assuming that

different species are the dominant contributors to the catalyst resting

state: (1) the resting state is a dimeric, neutral hexacoordinate

species A, (2) the resting state is a dimeric, cationic pentacoordinate

species B , or (3) the resting state is a combination of A and B .

Under assumption (1) the expression reduces to eq 4, which ﬁts

the partial order dependencies on 1b, 2b, catalysts (4c and 4d),

and chloride, except that chloride is integral inverse order. Under

assumption (2) the expression reduces to eq 3, which ﬁts the partial

order dependencies on 1b, 2b, and catalysts (4c and 4d) and but

shows no dependence on chloride. This result makes sense if the

resting state is already ionized. Under assumption (3) the expression

reduces to eq 6, which ﬁts the partial order dependencies on 1b,

(

39) Collum, D. B.; McNeil, A. J.; Ramirez, A. Angew. Chem., Int. Ed.

007, 46, 3002–3017.

40) For octahedral LB SiCl

2

3

·RCHO complexes, three constitutional

isomers exist, two cis (fac and mer) and one trans. Moreover, each

isomer can exist in many conformations that orient the aldehyde

differently with respect to the LB groups. Previous computational

studies revealed two important features that lead to our formulation

of the preferred mer-cis isomers shown in Figure 23: (1) the binding

of the aldehyde is more favorable in a trans position to an LB group

than to a Cl and (2) chlorines prefer to be trans if possible.

41) See Supporting Information for the full derivation of the kinetic

equations.

42) The inﬂuence of ligand structure on the enantioselectivity will also

not be discussed herein. For a substantive analysis of enantioselectivity,

a detailed understanding of the conformations of the various reactive

complexes is required, which will require extensive computational

modeling beyond the scope of the present mechanistic study. Although

the composition of the various species is now well in hand, their

stereochemical attributes are not discernable.

2+

(

2b, and catalysts (4c and 4d) and uniﬁes the two previous scenarios

for the chloride dependence. Thus, if k-2/k is large (i.e., a neutral,

2

dimeric resting state), then eq 6 reduces to eq 4, which predicts

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11783

A R T I C L E S

inverse ﬁrst order in chloride. However, if k-2/k

Denmark et al.

2

is small (i.e., a

cationic, dimeric resting state), then eq 6 reduces to eq 5, which

predicts no dependence on chloride. The observed -0.2 order in

chloride must mean that both cationic and neutral species are

29

present in the resting state (as detected by Si NMR) and are both

contributing to the formation of the active monomeric species.

Figure 26. Generalized catalytic cycle for aldolization catalyzed by

LB·SiCl .

4

with the rationale for identifying complex i as the active catalyst.

Coordination of i by the aldehyde through the oxygen lone pair

forms the reactive complex ii. This species may also be formed

directly from the dimeric resting states through binding of the

aldehyde, thus bypassing intermediate i. According to the rate

equation, the turnover-limiting (and stereodetermining) step is

the bimolecular collision between complex ii and the silyl ketene

acetal via transition structures approximated by depiction iii.

The origin of the high anti diastereoselectivity has been

discussed in detail previously and will not be elaborated further

here. The kinetically generated aldolate complex iv then

undergoes a rapid desilylation to form TBSCl, giving the

spectroscopically observable aldolate product v. We suspect that

the aldolate product is internally coordinated by the ester oxygen,

which may play a supporting role in facilitating the catalyst

turnover step. The Lewis acidity of the silicon atom in iv is

Figure 25. Derivation of kinetic expression for a dimeric catalyst resting

state, 4c and 4d.

4

2

1

/2

k k

^k4

2

3

1/2

- -1

rate ) k₅

[RCHO][1b][cat]_t_o_t_a_l[Cl ]

(

) (

)

2

k k

2 -3

k_-₄

-

(

4)

1

/2

^k3

k

4

1/2

certainly less than that in SiCl

Lewis base to dissociate at the end of the cycle. However,

the chelation of the SiCl subunit would further facilitate the

dissociation of the Lewis base for re-entry into the catalytic

4

, which is critical to allow the

rate ) k₅

[RCHO][1b][cat]_t_o_t_a_l

(5)

(6)

(

) (

)

2

k_-₃

k_-₄

3

1

/2

1

/2

^k3

k₄[RCHO][1b][cat]_t_o_t_a_l

4

3

rate ) k₅

cycle.

(

) (

)

1/2

2

^k^-3

k_-₄

k_-₂

- 2

From the rate equation and the foregoing discussion, it is

clear that the rate of the aldolization is dependent on the

[

Cl ] + 1

(

)

^k2

4

concentration of the reactive complex ii. Since the aldehyde

is consistent throughout, the concentration of ii is related to the

pre-equilibria established by the interaction of the phosphor-

4

. Analysis of Differences in Rates for Catalysts 4a-4d

and 5. 4.1. Postulate of a Common Catalytic Intermediate. On

the basis of all of the preceding discussion, a catalytic cycle

that represents the behavior of all of the catalysts studied herein

can be formulated, Figure 26. To facilitate discussion of the

factors that inﬂuence the differences in rates for the various

catalyst structures, it is instructive ﬁrst to focus on the common

features of the catalytic cycles that are believed to be operative

for all. The fundamental tenet of the mechanistic hypothesis is

that the reactive catalytic species is a cationic, ﬁve-coordinate

complex bearing two Lewis basic phosphoryl groups, i, Figure

amides with SiCl

4

and their afﬁnity for the aldehyde (eqs 2 and

3

in Figure 22). The zeroth-order kinetic dependence of the rate

on [SiCl ] implies that, under the reaction conditions (>50:1,

4

SiCl

4

/ligand), all of the available phosphoramide is coordinated

in some form. Indeed, even under the conditions of

P NMR observation (ca. 4:1, SiCl /ligand), little if any free

with SiCl

4

31

4

phosphoramide was detected. The relevant complexes formed

in these equilibria, as demonstrated by NMR spectroscopic

2

6. This species can be formed by dissociation of a chloride

(

43) Nevertheless, we did observe signiﬁcant curvature of the kinetic plots

beyond 30% conversion, which most likely is ascribed to sequestering

of the catalyst by the product aldolate, v.

ion induced by the Lewis basic ligands, the manner of which is

ligand dependent and will be discussed in detail below along

1

1784 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

4

Figure 27. Composition of complexes in bisphosphoramide/SiCl pre-equilibria.

Scheme 6

analysis of the mixtures of the bisphosphoramides 4 and SiCl

4

,

A similar collection of complexes representing the pre-

are shown in Figure 27. Although the composition of the

mixtures could not be rigorously quantiﬁed, the general trends

can be summarized as follows: (1) 4a consists of a roughly equal

mixture of I, II, and III; (2) 4b consists primarily of species II

and III along with other neutral, six-coordinate complexes and

a lesser amount of I; (3) 4c consists of a roughly equal mixture

of II, III, and V; and (4) 4d consists of a similar mixture of

complexes, but more strongly favoring the ﬁve-coordinate cation

V compared to 4c. Thus, two families of bisphosphoramides

can be identiﬁed: (1) short tethered ligands that form both

monomeric and dimeric, neutral, six-coordinate complexes but

do not form cationic, ﬁve-coordinate complexes in detectable

concentration and (2) longer tethered ligands that form only

dimeric complexes of both cationic, ﬁve-coordinate and neutral,

six-coordinate structures. In no case is the monomeric, cis-ﬁve-

coordinate cationic complex IV detected.

equilibrium of monophosphoramide 5 with SiCl is shown in

Scheme 6. By NMR analysis, the mixture consists of a roughly

4

equal composition of VI, VII, and VIII. Here again, the cis-

2

9

ﬁve-coordinate cationic complex IX is not detected in the Si

NMR spectrum.

To construct a self-consistent hypothesis for the different

catalytic activity of the various phosphoramides, one critical

assumption must be posited, namely that the cis-cationic

complexes of the type ii, Figure 26, are the kinetically most

competent for activation of the aldehyde toward nucleophilic

attack. This assumption is supported by three important lines

of evidence. First, the half-order kinetic dependence of reactions

in which the catalyst resting state is a dimer, even a cationic

dimer such as V, Figure 25, suggests that the dimeric species

are not kinetically competent and must dissociate to effect the

activation of the aldehyde. Second, even monomeric, trans-

cationic complexes such as VIII, though spectroscopically

detectable, are associated with the slowest of the aldolization

processes (i.e., catalysis by 5 and HMPA). Third, previous

(

44) Of course, the rate of the aldolization also depends on the intrinsic

reactivity of the different complexes. This feature is impossible to

determine without extensive computational study. However, assuming

that a cis-chelated cationic complex such as i is equally reactive for

all phosphoramides 4a-4c and 5 allows us to formulate an intriguing

hypothesis about the impact of the different ground states on the overall

reaction rates. In-depth computational analyses will be reported in due

course.

calculations of the various complexes of aldehydes with

+

[

SiCl

3

·4c] and ketones with enoxydichlorosilyl ·bis(N-oxide)

1

b,13a

cations

showed that electrophilic activation of the aldehyde

is strongest in such cationic, octahedral complexes when the

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11785

A R T I C L E S

Denmark et al.

Figure 28. Grand uniﬁed catalytic cycle.

carbonyl oxygen is bound through an sp-type orbital, trans to a

ligand oxygen. The high s-character of this orbital is a

consequence of the polarization of p-character toward the

The ground-state composition of the most enantioselective

catalyst 4c (n ) 5) is a ternary mixture of neutral and cationic

dimers II, III, and V. The half-order kinetic dependence requires

that the dimers must dissociate to form the reactive intermediate

ii. Since dimeric, cationic, ﬁve-coordinate complex [trans,trans-

4

5

chlorine atoms.

Combining the results from the kinetic analyses, spectroscopic

studies, and the postulate of ii as a common reactive intermedi-

ate, an explanation for the differential reactivity of the phos-

phoramides can be formulated, Figure 28. Implicit in this

2+

-

SiCl

3

·4c]

2

2Cl is present in high concentration, it is reason-

able to propose that this species is the immediate precursor to

ii on the basis of the Hammond postulate. Nevertheless, a more

signiﬁcant structural reorganization is needed compared to

catalysts 4a and 4b, and, indeed, 4c is a considerably slower

acting catalyst. The same analysis holds for the last bisphos-

phoramide, 4d (n ) 6), except that the resting-state composition

appears to have a greater proportion of the dimeric, cationic,

46

analysis is a corollary of the Hammond postulate, which states

that species that are close in structure are also close in energy;

therefore, the greater the structural reorganization needed to

change a resting state to a reactive complex, the greater the

energy of that reorganization, and therefore the less favorable

is that process.

ﬁve-coordinate species [trans,trans-SiCl ·4d]. Remarkably, this

3

is the slowest acting of the bisphosphoramides, yet it contains

the greatest amount of a cationic, ﬁve-coordinate species. This

highly counterintuitive observation provides strong support for

the notion that the trans-coordinated complexes are not kineti-

cally competent.

Finally, the behavior of the monophosphoramide 5 is

conﬂuent with the foregoing analysis. The composition of

the ground-state complexes with 5 is very similar to the

ground-state composition obtained with bisphosphoramide 4c;

namely, they are dominated by 2:1 complexes of both

cationic, ﬁve-coordinate and neutral, six-coordinate structures.

The ﬁrst-order kinetic dependence on [5] implies that two

molecules of the phosphoramide are present in the turnover-

limiting transition structure. However, this is by far the

slowest acting catalyst, which forces the conclusion that even

The composition of the ground-state complexes is dictated

by the structure of the bisphosphoramide according to the trends

described above. Thus, the shortest tethered bisphosphoramide,

4

a (n ) 3), forms a substantial amount of the neutral, cis-six-

coordinate complex I. Because of the ﬁrst-order kinetic depen-

dence on [4a], it is assumed that one molecule of 4a is also

present in the turnover-limiting transition structure. Moreover,

the conversion of I to ii (or its cationic, cis-ﬁve-coordinate

precursor) represents a minimal structural change and is

therefore the most energetically favorable for the various

intermediate complexes. Accordingly, reactions catalyzed by 4a

are found to be the fastest. A very similar analysis holds for

the next bisphosphoramide, 4b (n ) 4), which also shows ﬁrst-

order kinetic dependence on [4b] and is the second fastest acting

catalyst.

+

(

45) (a) Curnow, O. J. J. Chem. Educ. 1998, 75, 910–915. (b) Tandura,

S. N.; Voronkov, M. G.; Alekseev, N. V. Top. Curr. Chem. 1986,

(47) Magnetization transfer experiments with the HMPA ·SiCl

3

com-

32

plexes showed rapid reorganization of the ligands without dissocia-

·5 ] Cl complex is

2

kinetically accessible but thermodynamically highly unfavorable.

+

-

1

13, 99–189. (c) Bent, H. A. Chem. ReV. 1961, 61, 275–311.

tion. Thus, we conclude that the [cis-SiCl

3

(

46) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334–338.

1

1786 J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009

Rapid-Injection NMR Study of Aldol Addition Reactions

A R T I C L E S

+

-

a simple transfcis isomerization of [trans-SiCl

3

· 5

2

] Cl to

a chelated, trichlorosiliconium ion, the relative rates and kinetic

orders of all ﬁve catalysts could be understood.

+

-

a structure related to [cis-SiCl

3

· 5 ] Cl (IX) is energetically

2

47

unfavorable, most probably because of steric repulsion. This

conclusion is easily reconciled by the different steric

consequences of the reorganization of [trans,trans-

With a clearer understanding of the underpinnings of the

process, it is now possible to approach the much more difﬁcult

challenge of rationalizing the origins of variable enantioselec-

tivity for the different catalysts. These studies necessarily will

require extensive computational modeling and are in progress.

The explosive growth of the ﬁeld of asymmetric catalysis

has yielded spectacular advances in the introduction and

development of new, selective chemical reactions. During a time

of rapid expansion, the lure of harvesting the most readily

accessible results (low hanging fruit) is understandable and can

be seen in all corners of the discipline: transition metal, Brønsted

acid and base, and especially organocatalysis. However, the most

valuable bounty to be extracted from these enterprises is the

much harder won insight that comes from a fundamental

understanding of the origins of catalysis and selectivity. The

lessons gleaned from these investigations contribute to the

foundations of organic chemistry that will endure long after the

en Vogue infatuations are replaced by the next fad du jour.

2

+

2Cl^-(V) to [cis-SiCl

+

-

SiCl

3

· 4c]

2

3

· 4c] Cl (IV) compared

2

] Cl (VIII) to [cis-

] Cl (IX). The tethered bisphosphoramide has

+

-

to the change from [trans-SiCl

3

· 5

+

-

SiCl

3

· 5

2

already accommodated the steric energy of cis coordination

in its structure, whereas the monophosphoramides must

overcome the energetic cost of cis coordination of extremely

bulky ligands about the octahedral silicon center.

Conclusions and Outlook

The origin of the ability of chiral phosphoramides to enhance

the Lewis acidity of silicon tetrachloride has been investigated

in the context of enantioselective catalysis of Mukaiyama-type

aldol addition reactions. The empirical development of bisphos-

phoramides was guided by the knowledge that two Lewis basic

moieties are needed to facilitate ionization of a chloride ion to

create the trichlorosiliconium ion that acts as the catalytic

species. However, the kinetic behavior of the different bisphos-

phoramides (that vary the length of the connecting tether) was

unknown. By careful kinetic analysis of the aldolizations through

the use of rapid injection NMR spectroscopic analysis, a striking

difference in the catalytic activity and kinetic order of the

bisphosphoramides was revealed. From these kinetic studies,

Acknowledgment. We are grateful for generous ﬁnancial

support from the National Science Foundation (NSF CHE 0414440

and CHE 0717989). M.D.E. thanks Glaxo-Smith Kline for a

postdoctoral fellowship.

Supporting Information Available: Experimental details for

the preparative experiments, general description of the kinetic

experiments, all spectroscopic and kinetic data, and kinetic

derivations. This material is available free of charge via the

Internet at http://pubs.acs.org.

2

9

in combination with natural abundance Si NMR studies of

the resting states of the catalytic species, a clear trend emerged

that allowed the interpretation of the different kinetic orders.

By postulating a common catalytic intermediate comprised of

JA902474J

J. AM. CHEM. SOC. 9 VOL. 131, NO. 33, 2009 11787

Article Doi

On the mechanism of Lewis base catalyzed aldol addition reactions: Kinetic and spectroscopic investigations using rapid-injection NMR

DOI: 10.1021/ja902474j

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja902474j

Full text of DOI:10.1021/ja902474j

Products guided by the article

R&D Labs maybe for 36615-45-9

Relevant to this article

Hot Product