5.

44 Organometallic Chemistry
Notes from Spring 2020.

Last Updated: May 14, 2020.

Contents

1 Introduction 2

2 Transition Metals and Electron Counting 3

3 Bonding 5

4 Geometry 8

5 Ligand Substitution 10

6 Cross Coupling 15

7 Hydrogenation 20

8 Hydrofunctionalization 22

9 Hydroformylation 25

10 Oxygenases and Oxidases 27

11 Allylic Substitution 31

12 C-H Activation 33

13 Metathesis 36

14 Carbenoids 38

1
1 Introduction

These notes were typeset from the online 5.44 lectures, taught by Alison Wendlandt. Special

thanks to Andrew Lin for providing his LaTeX formatting template.

Question 1. What is Organometallic Chemistry? Why study it?

Organometallic Chemistry is very diverse - it is about transition metal complexes that actually

perform chemical reactions, involving intermediates containg carbon-metal bonds. The diversity

of the eld is due to many factors, including the sheer number of metals there are, the range

of oxidation states these metals have, and the diverse ligands that accompany the complexes

themselves.

One of the primary uses of organometallics is in stabilizing reactive substrates - for example, benzyne

and cyclobutadiene can be stabilized respectively by Zirconium and Cobalt (Buchwald, JACS, 1996,

118, 1028 and Maitlis, J. Organomet. Chem 1965, 4, 173).

Another reason is to promote otherwise unfeasible reactions - for example, the activation of

methane in sulfuric acid (Periana, Science, 1998, 280, 560):

Pt2+
200!
CH4 + H2SO4 +
C MeOSO3 H + 2 H

These unfeasible reactions can also generate complex molecular structures, for example, the

Pauson-Khand reaction:

C2H2 + C2H4 + CO [Co]! 2-cyclopentenone

Finally, due to the wide variety of potential organometallic catalysts, they can also be used to tune

reactivity, for example, with chiral ligands that favor formation of one enantiomer over another.

2
2 Transition Metals and Electron Counting

Denition 2
A transition metal is any element with an incomplete d-orbital shell, or which can give rise to

cations with an incomplete d-orbital shell, typically those in groups 3-11. The early transition

metals are those in earlier groups, and the late transition metals are those in later groups.

Some trends amongst these transition metals are as follows:

 As we move from early to late transition metals, more covalent bonds are created - the early

transition metals are harder, more ionic, and more electrophilic. Thus, the early transition

metals interact stronger with hard ligands - those with lower HOMOs, and these are mostly

ionic interactions. On the other hand, the late transition metals interact stronger with soft

ligands - those with higher HOMO energies, and these are mostly covalent interactions.

 Electronegativity trends follow the normal trend across a period - it increases from left to

right. However, in transition metals, the electronegative often increases from top to bottom,

in contrast with the trend observed in the main group elements.

 As we go from top to bottom, the metals become less metallic, and thus form stronger

bonds to carbon. This is because the metals become more electronegative, becoming closer

to carbon's electronegativity, contributing to stabilization.

Not only does the identity of the metal matter, but its oxidation state does too. Even though the

oxidation state is just a formality, it does help us count d-electrons.

Note 3
In general chemistry, we were taught that the electron conguration for, for example, Nickel,

was 4s 2 3d 8: However, this is for gas phase Nickel - in complexes, such as Ni(cod)2 = Ni(C8H8)2;
10
the electron conguration is 3d : The group number thus directly correlates to the d electron

count - the number of electrons that are not involved in principal bonding interactions.

By modifying the central d-electron count, we can change properties of the complexes themselves.

d 0 is the maximum oxidation state, where the metal center itself has no electrons - thus it cannot
do oxidative addition nor backbonding. d
10 is the minimum oxidation state. An odd d-count leads
to unpaired electrons, which leads to paramagnetism (though an even d-count does not necessarily

lead to dimagnetism!)

3
To count the actual number of valence electrons in the metal center, we dene X type and L type
ligands. L type ligands are those that make a neutral ligand fragment upon disassociation, while

X type ligands are those that create an anion upon disassociation. For example, CO is an L type

ligand, and Cl is an X type ligand. L type ligands contribute 2 electrons to the metal center, while

X type ligands contribute 1. These can also be combined - for example, Cp is a L2X ligand. There

are two main methods for counting them - either the ionic or the covalent method.

In general, a complex will have 18 electrons or less, with 18 electrons being the most stable (the

18-electron rule.) This has its basis in the number of valence orbitals - 1 s orbital, 3 p orbitals,

and 5 d orbitals combine to hold 18 electrons.

Remark 4. Most complexes have a coordination number of 6 - an octahedral coordination. The

highest transition metal coordination number observed has been 9 - in ReH92 (though other
complexes involving f-block elements have been found to have higher CN)

The 18 electron rule, however, is just emperical, and it has been found that the p orbitals don't

contribute much to the actual metal bonding. There are some important exceptions, such as the

following:

 Notably, that d8 complexes will very often by 16 electron square planar, due to the square

planar d-orbital splittings.

 Bulky ligands will lead to more ligands being unfavorable (such as in Pt(PtBu3)2 being a 14
electron complex)

 Agostic interactions, where a metal orbital interacts with a C-H sigma bond in the ligand

(such as inW(CO)3[PCy3]2 being a 16 electron complex)

 -donor interactions (such as in CrO4 2 being a 16 electron complex)
 High-spin complexes, where spin-spin pairing interactions or negative spin-spin pairing inter-

actions make it more favorable to have 17 electrons rather than 18 (such as Mn(H2O)62+
having 17 electrons)

 Extra electrons residing in the ligand (resulting in more than 18 electrons)

 Dierent bonding modes (for example, Cp interacting with every carbon, 3 carbons, or one,

corresponding to L2 X, LX, and Xligands). This property is called ligand uctionality.

4
3 Bonding

To start o our discussion of bonding, let's go back to the basics of MO theory. Two atomic

orbitals combine to form a more stable bonding orbital and a less stable antibonding orbital, and

the net stabilization is proportional to the orbital overlap and the dierence in the energy levels of

the atomic orbitals. Based on this model, we can make two generalizations:

1) When electronegativities are similar between a metal and a ligand, then there is more covalent

bonding. (recall from the previous lecture that going down a group increases electronegativity)

2) A larger electronegativity dierence will lead to stronger ionic interactions.

The simplest molecular orbital diagram of a metal-ligand complex is far more complex than the

simple diatomic bonding. For example, the PtH62 octahedral complex's MO diagram is composed

from metal valence orbitals - in order of energy (from lowest to highest), ve d orbitals (splitting
into a set of two eg obritals and three t2g orbitals), one s orbital (labelled a1g ), then three p

orbitals (labeled t1u ). The hydride ligands combine to form symmetry-adapted linear combinations

(SALCs), composed of one a1g orbital, three t1u orbitals, and two eg orbitals.
Since only orbitals of similar symmetry can interact, thes orbital interacts with the a1g orbital,
creating the lowest energy level (degeneracy = 1). The p orbitals likewise interact with the t1u

orbitals, creating the second lowest energy level (degeneracy = 3). The two d-orbitals with eg

symmetry interact with the SALC eg orbitals, creating bonding and antibonding interactions, with

the bonding interaction being the next energy level (degeneracy = 2). The t2g orbitals remain

nonbonding, being the fourth energy level and the HOMO (degeneracy = 3). The eg antibonding

orbital is then the LUMO.

Note 5
This orbital description actually provides an explanation of the 18 electron rule! The nonbonding

and bonding energy levels provide enough degeneracies for exactly 18 electrons, and having more

than 18 electrons means that the antibonding LUMO will have to be lled.

Now we'll see how metal-ligand interactions change the shape of the energy diagram. The simplest

case is a -donor, where the interaction is just a covalent interaction, between an unlled metal
orbital and a ligand orbital, that lowers the overall energy of the metal.

Many ligands have both  and  donation/accepting properties (or both)! CO is the archetypal

-acceptor ligand. The CO has an sp lone pair that will interact in a -donor interaction with an
unlled metal d orbital, and the empty  antibonding orbital can also interact with a lled metal

5
orbital to its energy as well. Essentially, CO donates its lone pair to an unlled orbital, and the

metal donates electron density from its d orbital back to the ligand. This property is known as

metal-ligand backbonding and results in a higher bond order for the M-C bond.
This also explains why CO is a bad lewis acid, but a very strong binder to complexes. It also

decreases the bond order of CO itself (since electron density is pushed into an antibonding orbital).

This can be observed with IR spectroscopy - we see lower wavenumbers in complexes (2125-1850

cm
1) as compared to free CO (2143 cm
1 ). In fact, IR spectroscopy with metal carbonyls can

be used as a metric for electron richness!

This technique is put into practice with the Tolman Electronic Parameter, which measures the
frequency of the CO stretch of the Ni(CO)3 L complex. If the ligand is able to donate eletron

density into the metal, then the CO stretching frequency should decrease, since the eects of

backbonding will be stronger. On the other hand, if the ligand accepts electron density, then the

stretching frequency will be higher. (Tolman, CR, 1977, 77, 424).

Another class of  complexes is with alkenes, under the Chatt-Dewar-Duncanson model. Our

alkene has a bonding orbital of  symmetry and an antibonding orbital of  symmetry, which

respectively interact with a lled and unlled metal orbital. This diers from the CO interactions,

since now both interactions come from the  bond itself. This means that alkenes can turn out to

be both -donor and -acceptor ligands, since we don't know if the metal donating to the alkene
is more prominent than vice versa!

However, we can predict these properties. If the metal is in a high oxidation state, then there is

not much productive orbital overlap between the lled metal orbital and the unlled alkene orbital,

so in this case the alkene would be a -donor, and the alkene bond length will be approximately
that of the free alkene. If the metal instead is electron rich and have a low oxidation state, then

the alkene will act as a -acceptor and the C-C bond is then lengthened.
On one extreme, we have just a simple metal-alkene interaction being a -donor. On the other

extreme, we have a -acid (acceptor), consisting of a metal cyclopropane. This scale also manifests
itself in terms of reactivity. The  -donor alkene will become more electrophilic, and the  -acceptor

will become more nucleophilic (due to lower and higher electron density, respectively). For example,
2+
Pd can be used to activate alkenes towards hydration, similar to the Wacker process. On the

other hand, a titanium-alkene complex can be used to make an alkene attack an aldehyde! (Bercaw,

JACS, 1983, 105, 1136)

6
 Note 6
This continium is similar in -complex ligands, such as H2 . On one extreme, the metal bonds

directly to the sigma bond between the hydrogens, and on the other extreme, the two hydrogens

are each separately bonded to the metal. The scale can similarly be measured via bond lengths,

with free hydrogen being 0.74 Å, and the hydrogen in metal complexes being anywhere from

0.82 Å to 1.5 Å, depending on the degree of electron density. If you think about this from a

MO standpoint, the two situations are almost the same.

Not only does this happen with hydrogen, but it can also happen with any M-X bond. Then,

a second equivalent of metal can bind to the M-X bond, and we commonly depict this as a

bridging ligand, though this depiction is more accurate.

Our last class of interactors are -donor ligands. This happens when the ligand itself has an extra

lone pair - for example, halides, amides, etc. There is a -donor interaction between one lone pair
and a metal d orbital, and a -donor interaction from a p orbital into the d orbital. Since this ligand

will only donate electron density into the metal, these ligands bind quite strongly to early transition

metals to stabilize them, since they have unlled orbitals and want more electron density. However,

since the oxidation state formalism does not reect the true electron density of a complex, the 18

electron rule often fails, as oxo and nitro ligands donate more electron density than would seem.
3+
For example, Jacobsen's epoxidation catalyst is a Mn complex that only have 14 d electrons, but

is stabilized by 2 N and 2 O ligands, making it relatively stable.

7
4 Geometry

The observed geometries for metal complexes are as follows:

 ML3 : trigonal planar, T-shaped, trigonal pyramidal

 ML4 : tetrahedral, square planar

 ML5 : trigonal bipyramidal, square pyramidal

 ML6 : octahedral, prismatic (having ligands at the verticies of a triangular prism)

VSEPR, as in general chemistry, provides a rst approximation to transition metal complexes;

however, its use is largely inadequate. This is emphasized by the fact that these complexes often

do not adopt the sterically most favorable conformation. Instead, we can develop a model of

geometry from the valence electrons and d electron count.

Recall from the earlier lecture that an octahedral complex shows a splitting into a set of two eg
orbitals and threet2g orbitals, with the eg orbitals being antibonding and t2g being nonbonding. On
the other hand, a tetrahedral complex also splits into these orbitals, but now the e orbitals become

nonbonding and the t2 orbitals become weakly antibonding. These qualitative MO diagrams can

be constructed by assuming that the d-orbitals aligned with metal-ligand bonds will be increased in

energy.

Now how do these considerations play into geometry? In octahedral complexes, we want to ll

the nonbonding orbitals and none of the antibonding orbitals. The ligands contribute 12 electrons,
and so we need 6 more to fulll the 18 electron rule and ll the non-bonding electrons, making d
6
octahedral complexes highly favored.

Similarly, in a tetrahedral complex, we would need a d 4 complex to ll only the nonbonding orbitals,
making that preferrential. A completely lled d 10 complex also prefers a tetrahedral conguration,
due to spin-spin pairing that ultimately lowers the energy of the system (keeping in mind that the

antibonding orbitals are only weakly antibonding). Half-lled orbitals also lower the energy of the

system, so d 5 high spin complexes likewise prefer the tetrahedral geometry.

In a square planar complex, the orbitals become further split, with the dxy orbital being slightly

bonding, the dxz and dy z orbitals becoming nonbonding, the dz 2 orbital being signicantly lowered

in energy (still weakly antibonding), and the dx 2 y 2 orbital increased in energy. Then, d 8 complexes
will ll all the bonding, nonbonding and weakly antibonding orbitals, making them prefer the square

planar complex.

8
If there are stronger metal-ligand bonds in octahedral complexes, then the bonding orbital becomes

even lower in energy and the antibonding d-orbitals becomes higher in energy, ultimately increasing

, the ligand eld splitting parameter. This change can be quantied per ligand and put into

a spectrochemical series, which states how much a ligand aects the splitting of a complex, and

these orbital diagrams ultimately allow us to predict the spin states of octahedral complexes. This

spin-state dependence on the ligand is most commonly observed in the 3d transition metals, since
the metals in later periods will tend to have larger splittings just by themselves. However, these

electronic properties can still have a profound ineucnce on reactivity, even in the metals of later

periods.

9
5 Ligand Substitution

Now that we have an understanding of the fundamental structure of metal complexes, we can look

at the fundamental elementary steps involved in actual organometallic mechanims, as follows:

Name Reaction
e
Ox. State

Ligand exchange/substitution MLn )disassociation
association
* MLn 1 + L 2 0
Red. elimination/Ox. addition Ln MXY )reductive elimination
*L M + X Y
oxidative addition n
2 2
metathesis
-bond metathesis/transmetallation XM1Ln + M2R ) * Ln M 1R + M 2X 0 0
migratory insertion
Migratory Insertion/Deinsertion Ln MLR ) deinsertion * Ln M L R 2 0
We'll examine all of these steps in greater detail in due time, but for now, we'll focus on ligand

substitution. Though it seems like a basic reaction, it is often the rst step in many catalytic

reactions. Additionally, coordinateively unsaturated complexes can be made reactive towards OA

or migration reactions. So, the prescence of excess phosphine or CO can cause these metals to

become saturated and essentially poisoned.

A latent coordination site is perhaps the most important property of catalysts.

There are two major mechanims for ligand substitution, the associative and dissociative mechanism.

The associative mechanism has the following reaction equation and rate law:

@P k1 k2 [ML1 ][L2 ]
ML1 + L2 )kk1*
1
ML 1L2 k2! M
L1
L2 @t = k 1 + k2
which is derived from the steady state approximation. Here we have a second order dependence at

all concentrations.

The dissociative mechanism has the following reaction equation and rate law:

@P k1 k2 [ML1 ][L2 ]
ML1 )kk1*
1
M + L 1 k2! M
L2
L2 @t = k 1 [L1 ] + k2 [L2 ]
which can likewise be derived from the steady state approximation. If the end step is the RDS,
@P k1 k2 [ML1 ][L2 ]
then k2  k 1 and the rate law simplies to
@t = k 1 [L1 ] : On the other hand, if the rst
@P
step is the RDS, then k2  k 1 and the rate law simplies to
1
@t = k1 [ML ]:
These rate laws give us ideas on how to dierentiate between the three possible cases (associative,

dissociative with rst step RDS, dissociative with second step RDS). If the rate is independent of
2
[L ], then we have a dissociative mechanism with rst step RDS. If running the reaction in the
1
prescence of added L results in inhibition, then we have a dissociative mechanism with second

step RDS, otherwise, it is an associative mechanism. The enthalpy and entropy of activation

10
also provides valuable data - the
S y will be slightly negative and the
H y will be positive for a

dissociative mechanism with rst step RDS, and vice versa otherwise.

We can also predict a priori which molecules are likely to undergo associative/dissociative mechan-

ims from the degree of unsaturation. If we have an 18-electron octahedral d6 complex or a 18-

electron d 10 tetrahedral complex, then a disassociative pathway is preferred, since the intermediate
is a stable 16-electron intermediate rather than a 20-electron intermediate.

Example 7
In the substitution reaction Ni(CO)4 +PPh3 ! CO+Ni(CO)3PPh3; the rate law is unimolec-
ular in the nickel complex, and has a negative entropy of activation and a positive enthalpy of

activation ( 8 eu and 24 kcal/mol). This data strongly points towards a disassociative mech-

anism with rst step RDS, and is also supported by the fact that Ni(CO)4 is an 18 electron

complex.

Remark 8. The bond strength of the Ni-CO bond is on the order of 25 kcal/mol, so the
enthalpy of activation tells us that the transition state is very product-like.

Light, heat, bond strength, and steric properties can all inuence the rate of ligand disassociation.

Example 9
Sterics can inuence the disassociation equilibria heavily. For example, let us consider the

disassociation of phosphine ligands from a nickel complex, NiL4 )K * NiL3 + L: The data is
d

shown below:
ligand P(OEt)3 P(OTol)3 P(OiPr)3 P(O o-Tol)3 PPh3
cone angle 109 120 130 140 145
Kd (M)  10 10 6
10 10 3
10 5 4
10 2 no NiL4 detected
For a review of steric eects, see Chem. Rev. 1977, 77, 313.

11
 Example 10
Another example is in the Buchwald-Hartwig reaction, between an aryl halide/psuedohalide

with an amine, in the prescence of palladium and diphosphine ligands. The mechanism is

an oxidative addition of the aryl halide to the palladium (the rate limiting step), substitution

of the halide with  OtBu, substitution of the OtBu ligand with the amine, and then nally

reductive elimination to regenerate the palladium catalyst and the product. Thus, to increase

the rate of the reaction, we would attempt to speed up the oxidative addition step by using

small electron-rich ligands.

However, one thing complicates this analysis is that the resting state of the palladium is tetra-

coordinate, while only the dicoordinate ligand is catalytically active. Thus, replacement with

the small ligands actually favors the tetracoordinate palladium, so the reaction is slower.

To x this, we use instead even bulkier ligands to favor dicoordinate palladium, such as dialkylaryl

or ferrocenium ligands. This leads to a faster and milder reaction, since the bulky ligands highly

disfavor tetracoordinate palladium.

Ref: JACS 1998, 120, 24, 7369

What factors aect ligand association? Just as the 18-electron complexes favor dissociation to

16-electron intermediates rather than association to 20-electron intermediates, 16-electron square

planar complexes favor association to 18-electron complexes. The nucleophile attacks the dz 2
orbital, creating a square pyrimidal complex that then isomerizes through a trigonal bipyramidal

intermediate to a square pyrimidal complex that has the leaving group in the axial position, which

then gets eliminated into the square planar complex.

Interestingly, this reaction is relatively stereospecic, which can be explained by the principle of

microscopic reversibility in where the leaving group should leave through the axial path. If the

trigonal bipyrimidal intermediate is long lived enough, however, Berry pseudorotation, an axial-

equitorial isomerization, can take place, and the reaction no longer is stereospecic.

There are more factors governing ligand association than dissociation, which should be expected

given that this is a bimolecular reaction. One factor is the metal itself, and the extent of interaction

- namely, the hard/soft interactions, the size of the metal, and the strength of M-L bonds. For

example, the rates of associative substitution of X(CN)42 decreases steadily as you go down

group 10. Other properties include the steric/electronic properties of the departing ligand, and

other ligands on the metal.

12
One specic consequnece of other ligands is known as the trans eect, where the ligand trans

to the departing ligand can have a large eect on the rate of substitution, in both kinetic and

thermodynamic terms (i.e. the metal-ligand bond strength depends on the ligand trans to it.) In

general, the rate of reaction is higher when the ligand is a good  donor or  acceptor. The

justication for this property is that the reaction undergoes a late pentacoordinate transition state,

and the trans ligand is less destabilized there as compared to the square planar complex. This

is because greater backbonding can be achieved from a pentacoordinate equatorial position (as

compared to square planar), and the  donation `competes' for d -orbital overlap with the trans
ligand.

Example 11
In the reaction below (L is trans to Cl), the rate of reaction is almost 104 greater if L H rather
than L Cl:
LPtCl(PEt3)2 + py ! LPt(PEt3)2(py)
EtOH

Another consequence of reactivity is from ligand steric eects.

Example 12
(Et3P)2PtClAr; where Ar C6H4R2; the rate of reaction is about 100000
In the substitution of

times faster when R H as compared to R Me:

A third example of inuences of reactivity is from the `ring slip,' where a Cp ligand changes its

hapcity.

Example 13
In the substitution of PPh3 in Cp?Ir(NHPh)(Me)(PPh3); the rate law is proportional to the

incoming ligand and the iridium complex, and there is a negative entropy of actiation, imply-

ing the associative mechanism. However, this is an 18 electron complex, which contradicts

our assumptions. What has been proposed to explain this is that the Cp ligand equilibriates

from 5 hapticity to 3 hapticity, in which the latter 16-e complex actually participates in the
substitution.

13
 Example 14
Some interesting patterns are observed based on the nature of the Cp ligand. TheCp? ligand
+
reacts about 100 times slower, while Cp PPh3 reacts about 100 times faster, Cp NO2 reacts
4 6 8
about 10 faster, permethylated indene reacts about 10 faster, and indene reacts about 10 :

The indene ring systems have extremely high rate constants, since aromaticity is retained, thus

also giving this eect the name the indenyl eect. The nitro and phosphonium substituted

cyclopentadienes are also able to stabilize the system when the hapticity changes. On the other

hand, the addition methyl groups destabilize the system formed, since the cyclopentadiene has

somewhat of a negative charge, and carbanions favour less substitution.

Finally, we have one more example for the impact of sterics.

Example 15
In this study, a platinum-based polymerization catalyst was studied. It involved a platinum

center coordinated to a diamine ligand, ethylene, and a methyl group, and it catalyzed the

polymerization of ethylene. The mecahnism was for intramolecular joining of the ethylene and

methyl fragments, and then repeated addition of ethylene to the complex to create a growing

alkyl chain. However, this suered from problems in that the joined longer chain could under

-hydride elimination or chain transfer, before an equivalent of ethylene bound to it again.

To solve this, the group used bulkier diamine ligands to slow down the rate of the side reaction.

This trick was simple, yet eect, increasing the average molecular weight about 4-fold. (JACS,

1995, 117, 6414).

14
6 Cross Coupling

Cross coupling reactions are essentially nucleophilic substitution reactions, that have emerged as

one of the most important catalytic processes in organic chemistry. The general cross-coupling

reaction is between an electrophile, typically an aryl halide, with a nucleophile, typically an alkyl

metal, in the prescence of a metal catalyst, resulting in formal nucleophilic substitution.

Note 16
These cross couplings have names associated with them, based on their discoverers, as follows:

 Kumada Coupling: Nuc = XMg-R, Cat = Ni/Pd

 Negishi Coupling: Nuc = XZn-R, Cat = Pd

 Stille Coupling: Nuc = R3 Sn  R, Cat = Pd

 Suzuki Coupling: Nuc = (RO)2 B  R, Cat = Pd + base

 
Hiyama Coupling: Nuc = R3 Si  R, Cat = Pd + F

 Sonagashira Coupling: Nuc = [Cu]  C 


 C  R, Cat = Pd/Cu

Despite the variety of cross-coupling reactions, these all follow a similar mechanism. Specically,

we start with a Ln PdII precatalyst, which gets activated (usually with loss of ligands) to generate
a Pd0 active catalyst that participates in a catalytic cycle. The aryl halide undergoes oxidative
0 II
addition to the Pd complex, turning it into Ln Pd ArX. Afterwards, transmetallation with the

metal alkyl results in the formation of Ln PdIIRAr; which then undergoes reductive elimination to
generate the product and the active catalyst.

The rst step of this process is oxidative addition, which is just a general term for the formal oxida-
0
tion stage change. The process is favored by electron rich metals in low oxidation states (Pd ) and

a low coordination number, as well as electron rich ligands, and is also favored thermodynamically

by the formation of strong bonds. It is observed that the rate of addition decreases in the order I

> OTf  Br  Cl > OTs > OMs and the reason for this is due to the lower bond strength of the
C X bond as one goes down this order.
However, new ligands have been developed to speed up the oxidative addition step, so even oxidative

addition of aryl chlorides can occur at room temperature. Aryl iodides instead are not as good in

these cycles, since they don't perform well in transmetallation. These ligands were developed by

Buchwald (JACS 1998, 120, 9722) and Fu (ACIE, 1998, 37, 3387) to be used in the Suzuki

coupling, with the conditions being 1.5% Pd2 dba3 , 3.6% phosphine, 2.0% Cs2CO3 in dioxane.

Buchwald's ligand ultimately became a diaryl phosphine amine, gave 96% yield, and Fu's PtBu3 gave

15
86% yield. These bulkier ligands promote the formation of the activate catalyst (low coordination

number) which promotes oxidative addition.

There are actually many mechanisms for oxidative addition itself, as follows:

 Concerted 3-center addition, which are generally seen in nonpolar substrates. A 3-membered

2-electron transition state is seen, and can be described as rst generating an encounter
complex consisting of a  interaction from the ligand -bond into the metal d orbital,

which then proceeds to an actual transition state with an interaction between a metal d
orbital and the antibonding orbital. As the backbonding character increases to be greater

than the  bonding character, then the transition state is favored.

 SN 2-like addition, which is more commonly seen in polar substrates. Essentially, the metal

Ln M adds to the nucleophile A B in a polar fashion, making Ln M A+ + B : For sp3 elec-

trophiles, this is truly SN 2-like, and can be common for Pd/Pt. This mecahnism can be

distinguished with stereochemical studies, for example, with the reaction of (Ph3P)4Pd and
HDCClPh; where we see inversion in the chiral center and have a trans sterechemistry of the
alkyl group to the chloride. (Stille JACS, 1974, 96, 5956). This mechanism is also possible

forsp2 electrophiles, where we just have more of a SN Ar-like, highly charged transition state,
shown with Hammett parameter data  = 8:8! The order of reactivity is like the normal

SN 2 addition, with the least sterically-hindered reagents and the ones with best leaving group

quality reacting the fastest.

 Radical mechanism, which is generally seen with alkyl halides, and in Ni/Ir systems. The metal

adds to the electrophile in a radical fashion, creating a radical which then adds back again to

the metal. The order of reactivity is opposite that of the SN 2, with it primarily determined by

R X bond strength. The evidence for this mechanism comes from a halomethyl cyclopropyl

radical clock, in where the generation of a radical results in ring opening and the formation

of a propenyl moiety with the metal. This has also been observed with sp2 substrates.
 Bimetallic mechanisms - not discussed

Reductive Elimination is the microscopic reverse of oxidative addition, so the factors which promote

reductive elimination often disfavor oxidative addition - factors which include having metals in high

oxidation states, those that are sterically hindered. Generally, sp2 substrates react much faster

than alkyl substrates, though electronic eects can also inuence rate.

The other step in the catalytic cycle is transmetallation, which is now often the slow step. Its

mechanism is highly dependent on nucleophile, and is much less well-understood. This step is often

thermoneutral and reversible, so they are driven by subsequent reactivity of the intermediates.

16
The majority of studies have focused on the transfer from Sn to Pd. It is thought that the metal

accepting the R group (Pd) is thought to be the electrophile, while the Sn-species is the nucleophile.

The Hammett parameter is  = 1:2; indicating that the migrating group is experiencing a bit of a
negative charge in the transition state. (Stille JACS 1983, 105, 6129), and explains why weakly

nucleophilic transition metal reagents such as boron and silicon require activation (reacting with

base and uoride respectively to form negatively charged complexes, that increase the rate of

addition).

There are two main mechanisms for transmetallation. The closed transition state is one of -bond
metathesis, with a four membered ring. The other mechanism is known as the open transition
state, where there is just a transfer of the R group to the metal with inversion of conguration.

Which mechanism predominates is highly dependent on solvent and reaction conditions in general,

and stereochemical probes can be used. In general, highly polar solvents favor an open transition

state, while nonpolar solvents favor the closed transition state.

The rates of transmetallation have been experimentally determined to follow the order as follows:

Alkynyl > Vinyl > Aryl > Benzyl  Me, Bu. This explains why we use SnBu3 in the Stille Coupling
- so that the transmetallation actually transfers the group we want! The rate is proportional to

the electron richness (the s character), and the prescence of coordinating funcitonal groups also

increases the rate. It is additionally also highly sensitive to sterics. Due to this sensitivity to sterics,

the aryl chlorides and bromides react much faster than the iodide, and the bulky ligands still help

with the rate by reducing the coordination number.

The nal step in cross-coupling is in precatalyst activation. One method is to reduce it via trans-

metallation, making a bialkyl product after reductive eliminaiton. Another method is with tertiary

aliphatic amines, and an alcohol, where the amine displaces a halide from the palladium, then un-
0
dergoes beta-hydride elimination to form an imine. Then reductive elimination of HX forms the Pd

species. A third method is to reduce the metal with phosphine - essentially, having the phosphine

attack the metal, then having the displaced leaving group remove the phosphine, ultimately oxidiz-

ing the phosphine and reducing the metal. A fourth method (developed by Buchwald) is to use a

sacricial amine aryl ligand, whereupon treatment with base allows it to facilly undergo reductive

elimination. (Buchwald, 2008, 130, 6686) A method similarly based on these sacrical ligands is

based on the indene system seen in Hazari, Nova, ACS Cat, 2015, 5, 5596.

Now we're going to discuss the details of the cross-coupling reactions we discussed earlier - Kumada,

Negishi, Stille, and Suzuki couplings.

The Kumada Coupling is normally beset with issues of low functional group compatibility, since it

has a grignard, though it is great if it works, since many other transmetallations proceed through

17
the grignard. The coupling is promoted usually by Ni/Pd, less commonly Fe, and NiCl2 dppp =

NiCl2 (Ph2 PCH2 CH2 PPh2 ) is the catalyst system of choice due to the large bite angle that promotes

reductive elimination. In this case, aryl chlorides generally react faster than Ar-Br/I systems. It's the

most useful for constructing sp2 sp2 or sp2 sp3 linkages, and they are stereospecic with respect
to the electrophile, but not necessarily stereospecic with respect to the nucleophile (dependent

on conditions). Generally, the congurational stability of the nucleophile trends opposite to its

nucleophility - lithium reagents are least stable and most nucleophilic, while boron reagents are

most stable and least nucleophilic. This leads to fast racemization, which can be exploited in an

enantioconvergent synthesis (for example, with (ppfa)PdCl2 (Hayashi, J. Organomet Chem, 2002,

653, 41).

The Negishi coupling is between Al, Zr, Zn, etc, which are much milder and much more functional-

group compatible. This is helpful in sp2 sp3 couplings where grignard decomposition is a problem.
sp3 sp3 coupling also work (Knochel, ACIE, 1995, 34, 2723), but they are more challenging, since
oxidative addition is much slower, and beta-hydride elimination competes with the transmetallation

step. Thus, the prescence of a tethered olen or even alkene ligands can actually promote the

reductive elimination step, by reducing electron density in the metal. (ACIE, 1998, 37, 2387).

Secondary alkyl electrophilies can also be used with py-box ligands, which essentially occupy coor-

dination sites to prevent beta-hydride elimination, and this discovery also led to the development

of an enantioselective enantioconvergent Negishi reaction. (JACS 2003, 125, 14726)

The Stille coupling is between a trialkyltin nucleophile, and has a wide scope. The electrophile can

be any sp2 or sp group, and the electrophile can be any aryl or vinyl group. It is almost 100%

functional group compatible, the organostannaes are easy to make, and are air and moisture stable,

making it the coupling of choice by synthetic organic chemists. Mechanistic studies can be found

in ACIE 2009, 43, 4704 and ACS Cat. 2015, 5, 3040, while a synthetic review can be found in J.

Organomet. Chem. 2018, 869, 106.

Remark 17. Lots of people believe that tin is toxic, due to SnMe3: However, most of the tin used
is actually SnBu3; which is much less toxic.
The Suzuki coupling is between boronic acids, with an additional base catalyst (that potentially

generates the borane or can replace the halide with a hydroxy group after oxidative addition). The

nucleophile can be sp2 or sp3 ; including primary alkyl groups, and many types of boranes can be
used. Alkyl-alkyl couplings are also possible (with Ni cat. and a diamine ligand). The advantage

of Suzuki couplings are the large varieties of boronic acids commercially avaliable, though they are

not very stable, thus promoting their late-stage usage.

Question 18. So which reaction should we use for our bond formation?

18
  In a disubstituted alkene product, if we don't have many reactive functional groups, then we

can simply use a Kumada coupling. Note that Pd often oers better stereoselectivity for

cis-alkenes, otherwise, nickel is ne. Otherwise, Negishi or Stille couplings will work.

 For trisubstituted products, we may have problems with stereochemistry, but Kumada, Suzuki,

Stille couplings are all ne.

 For aryl-alkyl couplings, Kumada couplings work great, but as usual, Suzuki/Negishi condi-

tions are necessary for functional group tolerance. Also note that in Suzuki/Negishi condi-

tions, either can be the nucleophile or the electrophile.

 For aryl-vinyl couplings, Suzuki and Stille couplings work ne, and if there are base-sensitive

substrates, then using the Stille coupling is necessary.

 For alkene-alkene couplings, Negishi or Stille couplings work ne.

 For aryl-aryl couplings, essentially everything works.

Now we'll take a brief detour to look at the other types of nucleophiles that can be used in coupling

reactions. One of these is with enolates, as seen in JACS 1999, 121, 1473. These can also be

done enantioselectively, since the mechanism is very similar - the only dierence being a substitution

instead of a transmetallation. Many substitution modes are possible - either a 3 or O coordination.

The most important non-metallic coupling is the Heck Coupling. It involves a similar mechanism

as the standard cross-coupling reaction, and is the fastest with monodenate phosphine ligands.

The mechanistic changes involve coordination of the alkene to the metal, syn-migratory insertion,

molecular rotation, beta-hydride elimination (generating the product), and nally base-promoted

reductive elimination. A cationic mechanism has also been proposed, in where bidentate ligands

are used instead.

These mechanisms can inuence regiochemistry. The neutral mechanism sees migration to the

carbon that is least sterically hindered, while the cationic mechanism sees migration to the position

with lowest electron density. (ACE, 1995, 28, 2 and TOC 1992, 57, 1480)

We can also do cross-couplings with non-carbon based nucleophiles - namely, making C-O and C-N

bonds. The mechanism has been studied extensively (JACS, 2006, 128, 3584). Here, there is a

precoordination of the amine to the metal, which then reacts with the base to deprotonate the

amine. Now, we may produce beta-hydride elimination products, can dimerize, or actually produc

our desired product. This reaction has been pretty well studied and its scope includes cyclic/acyclic,

secondary, primary, and aromatic amines, as well as ammonia surrogoates, amides, carbamates,

and sulfonamides. These are summarized in the `user guide' Buchwald Chem Sci, 2011, 2, 27.

Finally, Cu-catalyzed Ulman/Goldberg reactions are also possible - see Chem Rev 2008, 108, 3054.

19
7 Hydrogenation

Heterogeneous hydrogenations are very commonly used in industry, for example, in the Haber-

Bosch and Fischer-Tropsch proceses as well as the reduction of adiponitrile. These are often more

advantageous due to the ease of product separation, recyclability of the catalyst, reactivity, and

turnover number.

Homogeneous hydrogenations, which we will focus on, also have some advantages - namely, milder

reaction conditions and ease of reactivity/selectivity control (for example, dierentiation between

alkene isomers, 1,2/1,4 addition, and ketone/imines.) Diastereoselective and enantioselective hy-

drogenations are also possible with homogeneous hydrogenations - in fact, enantioselective hydro-

genations are the most important enantioselective catalytic reaction in industry. Finally, mechanistic

studies are also more tractable.

The rst use of hydrogenation catalysts was in Wilkinson's reduction of an alkene, which reacted

much faster than the standard PtO2 reduction, with the catalyst Rh(PPh3 )3 Cl. In the mechanism,

named the dihydride pathway, one phosphine ligand is displaced with a solvent ligand, to which
hydrogen is oxidatively added to the complex. (Alternatively, an associative mechanism can also

take place, but is about 104 times slower, shown by subjecting an olen to a 1:1 mixture of H2
and D2; and noting that there were very few cross-products.) Then, the solvent is displaced by an

alkene, to which the metal coordinates. The rate determing step is then the insertion of the hydride

into the alkene (and the direct formation of a C-M bond), and afterwards, reductive elimination

occurs to give the reduced product.

Since H2 is a nonpolar substrate, the mechanism for oxidative addition is through the formation of

the  complex, which undergoes concerted addition to the cis-metal dihydride complex. Depending
on the conditions, the  complex may or may not be long lived. A certain tungsten complex (Kubas

complex) actually has the  complex as its most stable form!

After the hydride is added to the metal, then the substrate coordinates to the metal and then

reversibly undergoes migratory insertion. In this process, there is no change in oxidation state,

and the migrating group must be cis to the unsaturated ligand. Further, it is syn-selective (metal

and hydrogen add to same side). Because insertion is very fast, there are very few hydride olen

complexes. Insertion of C-H bonds is often much faster than C-C bonds, due to better orbital

overlap.

In general, terminal alkenes, cis alkenes, and unhindered alkenes react faster in hydrogenation

reactions with a homogeneous catalyst - sterics determine selectivity (this is not necessarily true

with solid-phase catalysts, where electron-rich alkenes react faster). Cationic catalysts can improve

20
this selectivity even more, for example in the Osborn-Schrock Rh catalysts or the Crabtree Ir

catalysts. Wilkinson's catalyst works well for most types of alkenes, the Schrock does not work

well with internal alkenes, and the Crabtree catalyst works for every type of alkene.

Using these catalysts, we can also perform directed hydrogenation. We can often get dr's of 99:1

when performing hydrogenation with directing groups (OMe, C=O, etc), even when sterics are

unfavorable. In comparison, the standard Pd/C gives only about 4:1 dr. One thing to watch out

for is that the catalysts can also act as Lewis acids promoting elimination of the products. This

problem can be worked around with by using THF as solvent (not done often!)

These catalysts have a dierent mechanism than the Wilkinson's catalyst. Here, the alkene co-

ordinates rst to the metal, and the directing group coordinates as well (this is the resting state

without hydrogen.) The hydrogenation of the metal is the rate limiting step, and the product is

formed with a cis-migratory insertion and reductive elimination. (Halpern, Science 1982, 217, 401)

These catalysts can also be modied to be enantioselective. The classic example is the reduction by

enamines by P-chiral C2 symmetric (cod)Rh(DIPAMP); used in the Monsanto process of L-DOPA.

Another more interesting example is with Ru(BINAP)(OAc)2 complexes reported by Noyori, (JACS,

1986, 108, 7117), which oer greatly increased enantioselectivity, even with 0.01% catalyst!

The Noyori catalysts proceed through a dierent mechanism, called the monohydride mechanism

(JACS 1991, 113, 589). Here, the alpha hydrogen comes from H2 , and the beta hydrogen comes

from solvent. The mechanism proceeds rst by loss of one acetate ligand and association by the

carboxylic acid substrate moiety. The hydrogen binds to the metal in a sigma complex, and the

other acetate ligand promotes heterolytic cleavage of the dihydrogen. This protonated acetate

ligand then is replaced with the solvent, and hydride is inserted into the substrate. Reassociation

of acetate and reductive elimination yields the product. Note that Ru stays in the +2 state through
the whole cycle.

With Ru(BINAP)Cl2 , then ketones with close chelating groups can also be reduced. With a diamine

ligand and base, then any ketones, even ; unsaturated ketones, can be even favored over alkenes!

(JACS 1995, 117, 10417) Now, if this is combined with chiral phosphine and chiral diamine ligands,

then we can get enantioselective reduction of ketones!

The mechanism for the diamine-mediated reduction starts o by making the trans-dihydride com-

plex with the loss of the two chlorine ligands. It's then proposed that the ketone doesn't directly

interact with the metal, but rather a hydrogen-bond based activation by the diamine. The hydro-

gen atom that reduces the ketone then comes from the ruthenium, and the one that bonds to the

oxygen comes from the nitrogen.

21
8 Hydrofunctionalization

Hydrofunctionalization has similar mechanisms to hydrogenation. For example, in hydrosilylation,

organosilanes similarly undergo oxidative addition, then coordination of the alkene. If there is a

C-H insertion this is known as the Chalk-Harrod mechanism, and reductive elimination creates

the silane at the less substituted position. Alternatively, C-Si insertion can also occur ( modied
Chalk-Harrod mechanism) that reductively eliminates to give the same product (but can also

eliminate on the other side to give rise to observed side products).

Typically, Rh, Ir, Pd, Pt, Cu, Au, Fe can be used, with Rh, Pd, Pt being the most common

(specically H2 PtCl6 ). The reactions often use much milder conditions to aect reduction of

carbonyl equivalents. (Org. Lett. 2007, 9, 1113)

Hydrosilyation can also be done to alkynes to add in a trans-fashion, once again with silicon adding

to the less hindered side. This can also be done to internal alkynes. Treatment with uoride after

hydrosilyation allows for conversion of an alkyne to a trans-alkene. (Trost, JACS, 2002, 128, 9328)

Enantioselective hydrosilyation has also been developed. Using a Pd - MOP catalyst with HSiCl3 ,

a SiCl3 group is added to the more substituted position, upon which substitution with EtOH and

a subsequent Tamao oxidation yields enantioselective hydroxylation.

Note 19
Classic chelating phosphines, such as DPPB, BINAP, etc don't work in these transformations.

We can also catalyze hydroboration, even though it is already pretty selective. We do this for

better terminal selectivity or enantioselectivity, or even just for reaction speed-up. For example,

hydroboration-oxidation of allylic alcohols will lead to the anti-product without a catalyst, but to

the syn product with the catalyst.

The mechanism for one of these catalysts, for hydrogenation, is also a monohydride mechanism,

but it is somewhat dierent. We start with (Ph3 P)3 RuHCl, which undergoes standard loss of ligand

to form (Ph3 P)2 RuHCl. Then, coordination to the alkene, insertion of the hydride into the alkene

happens as usual, and then a equivalent of hydrogen add to the metal to make an octahedral

complex. Reductive elimination then results in regeneration of the four-coordinate metal and the

hydrogenated product.

More specically, we can investigate the hydride migration step. For highly charged early metals,

such as Zr(IV),d 0 ; the equilibria lies in favor of migratory insertion, since the alkene interaction is
just a weak sigma donor interaction. This allows for the Schwartz reagent ZrCp2 ClH to be used

22
in a variety of transformations of alkenes, for example, with reaction with Br2 , HCl, or H2 O2 to

respectively form the bromide, reduced product, or the hydroxy compounds. It can also be used

for transmetallation purposes.

For late, low valent metals such as Pt(II), we instead have stronger -backbonding between the
alkene and the metal, and hence the equilibria lies more towards the hydride complex. Thus, we

can use these catalysts to aect isomerization between two olens through addition then beta-

elimination, or in polymerization. We can also use these low-valent metals to insert CO into the

alkene as well.

We have one `nal' mechanism, seen in the reduction of anthracene to 9,10-dihydroanthracene,

under H2=CO and Co2(CO)8 catalysis. This mechanism rst sees formation of the HCo(CO)4

intermediate, which then reacts with the antracene system to create an antracene radical and

Co(CO)4: Donation of another hydrogen and reformation of the dimer concludes the cycle.
In general, these reactions can proceed because metal-hydrogen bonds have low BDE's compared

to C-H bonds, with the average being about 60-65 kcal and the rst row metals having slightly

lower BDEs.

Another reaction is in using a MnIIITPPCl catalyst in the prescence of NaBH4 and O2: There are
both reductants and oxidants, since this reaction was designed to be a structural model for P450

(which uses O2 and NADH.) In the oxidation of cyclohexene, what is found is that 20% of the

allylic alcohol is made, and 80% of cyclohexanol is made. On the other hand, without NaBH4 ,

80% of the ketone is made, 18% of the allylic alcohol is made, and 2% of the epoxide is made.

This motif of oxidation-reduction hydration ultimately led to the Mukaiyama hydration, which

adds water across a double bond with Co(acac)2 , PhSiH3 , and O2 . These are much better chemos-

electively and tolerating as compared to acid-catalyzed hydration, and adds to the more substituted

position, as usual.

The mechanism for this reaction has not been fully understood, but is thought to involve oxidation

of the Co(II) catalyst into a Co(III) hydride. Then, the alkene adds to the cobalt, and a radical

is formed after cobalt leaves, which then reacts with the oxygen to form a peroxide radical, which

reacts with the catalyst. Reduction with the silane results in regeration of the active Co(III) hydride.

Since this reaction generates a radical, some applications have been developed, for example, in

hydroazidation. With the reaction of an alkene, catalysts, and tosyl azide, then we can get azidation

at the more substituted position. We can also perform hydroalkylation and alkene isomerizations

with these systems as well. These radical-based reductions are often much more mild, and can

even perform reductions on vinyl halides.

23
Note 20
Hartwig's rst edition textbook (published in 2010) says that radical-based reactions are a thing

of the past, but much work has been done in the last few years. For a review, see Shanvii ACR

2016, 116, 8912.

24
9 Hydroformylation

Hydroformylation is the addition of CO and H2 across a double bond, generally making a linear and

a branched isomer. It's catalyzed by platinum group metals, Pt, Ir, or Rh, and sees wide usage

on an industrial scale (Oxo process). The rst discovery was with a Co2 (CO)8 catalyst, in a 1:1

H2/ CO mixture. Our modern rhodium catalysts operate at lower temperature and pressure, and

also give better selectivity ratios. We can also use cationic co-catalysts: see Science 2020, 367,

542.

Note 21
The H2/ CO mixture is known as syngas, and is a waste product in petrochemistry, so it is

avaliable relatively cheaply.

In general, the active form of the catalyst is (Ph3 P)2 (CO)2 RhH, which can undergo equatorial-axial

inversion. Without Ph3 P, the reaction actually proceeds much faster, but gives worse selectivity

ratios. The general mechanism proceeds as follows:

First, a CO ligand is lost, and then the alkene coordinates to the Rh metal. Insertion of the hydride

into the alkene, followed by binding of CO or a phosphine ligand, regenerates the Rh catalyst.

Then, CO undergoes migratory insertion, upon which addition of H2 regenerates the catalyst and

produces the product through reductive elimination.

Originally, it was thought that the ratio of eq-ax isomers of the catalyst lead to the selectivity, and

this inspired the development of diphosphines with large bite angles. This hypothesis is not true,

but these ligands still oered improved selectivity (Buchwald JACS 1993, 115, 2066).

Since internal olens can also undergo isomerization, then a mixture of internal and terminal alkenes

can also be used in hydroformylation. This is important since these mixtures are produced from

the industrial SHOP process.

Though hydroformylation commonly favors the linear product, branched-selective hydroformylation

is also possible, and we can use chiral ligands, such as BINAPHOS in order to enforce enantiose-

lectivity. (Takaya, JACS, 1993, 115, 7033). This has been used in syntheses, such as the one of

ambraticin by Jacobsen.

Let's investigate the migratory insertion of the carbon monoxide. The question is - does the R

group migrate to the CO, or does the CO migrate to the R group? Caldeazzo demonstrated that

it was actually the R group that migrated, with stereochemical studies. (Caldeazzo J. Organomet.

Chem. 1967, 10, 101.)

25
They investigated the insertion of CO in MeMn(CO)5 . The rst experiment was to run the reaction

under a labelled CO atmosphere, and they saw that the resulting acyl group was always cis to the

new CO ligand. Thus, CO-insertion is intramolecular, and isomerization is negligible.

The second experiment was to show that the principle of microscopic reversibility holds, by showing

a labelled product (on the acyl) would result in cis-CO ligand to the methyl group in the starting

material.

The third experiment was when one of the CO ligands cis to the acyl in the product was labelled.

The isolated starting material were both the cis and trans isomers, and are observed in a ratio of

2:1. This shows that the methyl migrates, since otherwise we would only see cis starting materials.

The fourth experiment was when the starting material only had one of the cis-CO ligands labelled.

The resulting products were the trans isomer, cis isomer, and the isomer where labelled CO was

inserted, in a ratio of 1:2:1. Likewise, this is consistent only with the methyl migration.

Another example of hydroformylation is in Methanol carbonylation, which is used industrially to


produce acetic acid. Rh/Ir are the common catalysts of choice, with Rh(CO)2 I2 being the catalyst.

The mechanism starts with the formation of hydrogen iodide, which reacts with methanol to make

methyl iodide. This methyl iodide then reacts with the Rh catalyst to undergo a SN 2-like oxidative

addition (RLS for Rh). Insertion of CO (TOLS for Ir) followed by reductive elimination results in

the production of acyl iodide and the rhodium catalyst, and the acyl iodide reacts with the water

to form acetic acid and HI.

Since CO insertion/migration is known for every transition metal, then we can also use these

methods in synthesis. For example, if we have a C-O coupling reaction, if we run it under CO, we

can get an ester.

There are a few other reactions worth mentioning, for example, the insertion of CO into epoxides

under Cr(salen) catalysis (JACS, 2002, 124, 1174), the Pauson-Khand reaction, and hydroesteri-

cation/hydroamidation with (Ph3 P)2 PdCl2 , CO, and MeOH. The mechanism of hydroesterication

involves insertion of CO to form an acyl chloride, then substitution of the chloride to form the ester

by methanol, coordination by the alkene, and then migration to form the ester product which is

reduced with reductive elimination.

26
10 Oxygenases and Oxidases

The next few lectures will focus on oxidative transformations aected by organometallic complexes.

We'll rst focus on biological enzymes oxygenases and oxidases, which dier depending on whether
the oxidant (usually O2 ) gets incorporated into the substrate or just serving as an electron acceptor.

Schematically, for an oxygenase, we can represent this as the formation of a metal-oxo complex,

which then transfers the oxygen atom to the substrate. For an oxidase, the metal is also oxidized

by oxygen without O transfer, which then goes on to oxidazes the substrate.

The archetypical oxygenase are the cytochromes, for example, P450, which has motivated the

development of many catalysts. One example of the use of P450 is in the biosynthesis of Taxol,

where an unactivated C-H bond is oxidzed to a C-OH bond with retention of stereochemistry.

These cytochromes are based on a Fe(II)-heme complex. The general mechanism is for coordination

of oxygen to make a hydroperoxyl radical Fe(III) complex, which then is reduced (by a coreductant,

typically NADH) to the hydroperoxide Fe(III) complex. Then, addition of protons results in the

loss of water and oxidation to the Fe(IV) complex, which then transfers the oxygen to a species of

interest with concurrent reduction to Fe(III), which is then reduced back to Fe(II).

Note 22
Since the reduction of oxygen to water is a 4 electron reduction, but the incorporation of an

oxygen atom is only a 2 electron oxidation, commonly coreducants are needed. A way to get

around this is to use H2 O2 as the oxidant instead.

We can use our own synthetic Fe(III) porphyrins for epoxidation/hydroxylation. For example, the

reaction of 1,3-cyclohexadiene, with oxidant PhIO and 10% Fe(III)TPP  Cl, results in the formation
of PhI and monoepoxidation. Chiral porphyrin catalysts are also possible, but enantioselectivities

are not great since most porphyrins are mostly at. One example is reported in Collman, Science,

1993, 261, 1404.

Salen-type ligands were also shown to be able to support high oxidation states. For example, use

of 4% of Mn-salen complex and PhIO oxidizes cyclohexene to cyclohexene oxide (Kochi, JACS,

1986, 108, 2309), through an intermediate Mn-oxo species. The salen ligands can be made to

be chiral, resulting in enantioenriched oxidation, known as the Jacobsen/Katsuki epoxidation.

This reaction has wide scope, and performs best on cis alkenes (since trans alkenes may cause

stereochemical scrambling). Sometimes, 4-phenylpyridine N-oxide is added to improve the reaction

ee, where it acts as a coordinating ligand. A review can be found in CR 2005, 105, 1563.

27
It is thought that the enantioselectivity of this reaction derives from the side-on approach of the

alkene to the Mn-oxo complex (from the alkene HOMO to the oxygen lone pair LUMO), and

proceeds through stepwise radical intermediates. It is generally found that those ligands with

electron-donating substituents give the highest ee's (JACS, 1991, 113, 6703).

Remark 23. The Sharpless Ti-based enantioselective epoxidation is often complementary to the
Jacobsen epoxidation.

Aziridinations occur in a very similar way to epoxidations. For example, a Mn 


 N bond is formed

from photolysis of a Mn-azide, and in the prescence of triuoroacetic anhydride, cyclohexene is

oxidized to N-triuoroacyl cycloaziridine. Another example is in the oxidation of a phenyl silyl enol

ether, which then undergoes ring-opening to make N-triuoroacyl 1-aminoacetophenone. A review

can be found in Dubois, Carriera, ACR, 1997, 30, 372. The development of TFAA-free aziridation

came when using copper catalysts with PhI 

 NTs as oxidant to create tosylaziridines. (Jacobsen,

JACS, 1993, 115, 5326).

Two mechanisms for the copper-catalyzed reaction have been proposed. The rst mechanism
+
involves a reduction of Cu(I) into a Cu 
 NTs intermediate, which then oxidizes the alkene. The

other proposed mechanism is one where the copper generates a nitrene-like species, CuN(Ts)(IPh),

which oxidizes alkenes without changes in copper oxidation state.

Remark 24. The aziridination reaction has not been fully developed, and has limited scope (working
only on styryl substrates and needing a protecting group). The state-of-the-art aziridations are
based on rhodium systems.

We also have some systems based on non-heme oxygenases. The biological system is the oxidation

of methane to methanol with oxygen and NADPH as reductant, with a hydroxo-bridged Fe(II) dimer

serving as catalyst. Addition of oxygen to this species makes a peroxo-bridged Fe(III) dimer, upon

which protonation and loss of water creates a Fe(IV) oxo dimer (also represented as Fe(III)/Fe(IV-

oxyl species). Abstraction of hydrogen generates a substrate radical, from which a radical rebound

results in hydroxylation and generation of the Fe(III) hydroxy dimer, which is reduced back to Fe(II)

with NADPH. (Lippard, CR, 2003, 103, 2385).

Non-heme ligands have been used in synthetic oxidation systems, which are often stereoretentive

due to a radical rebound. These reactions often have yields that are moderate (51%) since the

catalyst actively decomposes the hydrogen peroxide cocatalyst, and the catalyst system has to be

recycled 3x. (White, Science, 2007, 318, 783). Another example of directed C-H hydroxylation is

in the oxidation of a gibberilic acid, which uses a carboxylic acid-directed oxidation in 52% yield.

Now let's look at oxidases. One biological example is galactose oxidase, which converts galactose

28
into its aldehyde. The mechanism involves a Cu(II) catalyst with an organic tyrosyl radical cofactor.

The alcohol coordinates to the Cu(II), and then there is a reaction in where the hydride transfers

one electron to the tyrosyl radical and one to the Cu(I), forming an aldehyde in the process - a

mechanism for a 2-electron oxidation with Cu. Coordination of oxygen makes the hydroperoxyl

Cu(II) complex and releases the aldehyde product, then hydroxyl transfer regenerates the tyrosyl

radical, and then substitution liberates hydrogen peroxide.

Remark 25. In many of these biological reactions, the desired product is actually not the organic
product, but rather H2 O2 , which is useful cellularly.

Oxidations based on this protocol have been developed using TEMPO instead of tyrosyl radical.

The typical conditions are 5% Cu(OTf ), 5% bpy, 5% TEMPO, 10% NMI, and air. (Stahl JACS,

2011, 133, 16901). The mechanism has been resolved (see JACS, 2013, 135, 6, 2357) and is very

similar to biological oxidases. This protocol is useful for selectively oxidizing primary alcohols.

One of the most important organometallic oxidases is in the Wacker oxidation, which oxidizes

alkenes to ketones and is used heavily industrially to oxidize ethylene. The catalysts are PdCl2

and CuCl2 , and the reaction is done with O2 , H2 O, HCl. Note that this is an oxidase - the oxygen

incorporated in the product comes from the water, not the oxygen.

The mechanism involves the coordination of the alkene, and then a nucleopalladation step that

involves the outer-sphere attack of water onto the alkene to create a Pd-alkyl species with loss

of HCl. Beta-hydride elimination makes a palladium hydride and our product that enolizes to the

ketone. The palladium hydride reductively eliminates HCl and is reoxidized by copper chloride,

which is oxidized by oxygen itself.

A cis-nucleopalladation step (where the hydroxyl is rst bound to the palladium) has also been

proposed, which was disproven by Stille. The key experiment was the stereocontrolled addition of

cis-deuterated ethylene in the prescence of CO. It ends up producing the trans-deutero product (a

4-membered lactone), showing that trans-nucleopalladation occurs.

Other oxidants such as benzoquinone (as seen in Tet, 1994, 35, 6481) can also be used. Enan-

tioselective variants (based on sparteine) have also been developed. A cascade reaction has also

been developed, using CO and MeOH to transform 5-hydroxy alkenes to tetrahydropyran moieties

with a methyl ester (JACS, 1984, 106, 1496).

Finally, usage of other nucleophilies has also been developed, for example, in the dihydroxyacetyla-

tion of indene with Pd(dppp)(H2 O)2 (OTf)2 and PhI(OAc)2 , HOAc (JACS 2008, 130, 2964). The

mechanism is thought to proceed via trans-attack of the acetic acid after alkene coordination, then

oxidation to Pd(IV), from which a second attack by the incorporated acetic acid unit regenerates

29
Pd(II) and releases an acetoxyonium ion that can be ring-opened to give the dihydroxyacetylated

product.

The usage of other nucleophilies has also led to oxidative annulation protocols (Stahl, JACS, 2007,

129, 6328), which undergoes cis-attack due to the intramolecular nature of annulations. Overall,

cis/trans attack and regioselectivity itself depends on the catalyst system and solvent (JACS 2005,

127, 2868).

Hydroamination reactions (not oxidative) bear a similar mechanism to these Wacker-style oxida-

tions. Instead of a beta-hydride elimination at the end, the alkyl substitutent is simply protonated.

A palladium hydride mechanism has also been proposed (Exercise for the reader). A review: Beller,

ACIE, 2004, 43, 3368.

Finally, though this lecture has focused on palladium, rhodium catalysts have been developed (Hall,

JACS 2019, 141, 739).

30
11 Allylic Substitution

Allylic substitution, like the Wacker oxidation, involves coordinated attack of a nucleophile from

the outer-sphere. It is not an oxidation in itself, but has a similar mechanism. (For a review: CR

1996, 96, 395)

The general mechanism is a metal reacting with an allylic halide to generate the metal- 3 allyl

complex (through a alkene complex followed by an `oxidative addition'), which then is subjected

to nucleophilic attack (a `reductive elimination') to generate a Pd-alkene complex, that ultimately

disassociates to the product.

What are the stereochemical outcomes of this reaction? Trost found that if the nucleophilic is soft

(enolates, amines, etc), then retention of stereochemistry occured, through a double-substitution

mechanism. On the other hand, when gringnards, organometallics, hydrides, or other hard nu-

cleophilies are used, then inversion is seen. The mechanism of oxidative addition remains with

inversion, but now the transmetallation occurs and cis-attack occurs for an inner-sphere cis-attack.

There is also a possibility for alkene isomerization, via 3 =1 slip, that can result in the formation
of the trans-alkene from a cis allylic alcohol, through a Pd-alkyl intermediate. The rate of isomer-

ization is slow for sterically hindered alkenes, but for unhindered alkenes, the syn-form (that leads

to the trans-alkene) is favored. Thus we can exploit this to generate E alkenes selectivity.

Additionally, linear products are preferred for steric reasons in the outer-sphere trans-attack mech-

anism. However, if there is a strong electronic bias, then internal substitution is instead preferred.

Essentially, if a cation in the internal position is favored (or the outer group is an EWG), then

sometimes the internal isomer is formed.

For an inner-sphere cis-attack mechanism, then the nucleophile will substitute at the more internal

position, allowing the Pd to be less hindered.

The synthetic methodology that has been developed using allylic carbonates is known as the Tsuji-
Trost reaction (Tsuji, TL, 1982, 23, 4809). Further, beta-ketoesters can also be used to mask

the carbonate. This methodology has been used in many natural product syntheses. Further, allylic

epoxides can also react, with concurrent ring opening (Trost, ACIE 1997, 36, 1486).

Rh and Ir-catalyzed allylic substitutions have higher regioselectivity for the more substituted po-

sition. This has been justied by proposing that the cationic site adjacent to the metal is more

stable. An enyl complex (where the metal complexes to the alkene and has a sigma bond to the

alkyl as well) is proposed instead, and this is slow to exchange to the other enyl complex. This

was proposed since branched/linear allylic starting materials sometimes gave dierent product ra-

31
tios. Further, crystollography showed dierent bond lengths, that would not be present in the allyl

complex.

Now let's look at how we can introduce asymmetry in these reactions. Supposing we have a

symmetric allyl complex, attack at the two positions will generate two dierent enantiomers, and

one can be selectively favored. Another method of introducing asymmetry is the creation of a new

stereocenter from a prochiral molecule. A third method is substitution of diasterotopic leaving

groups. Essentially, we have a symmetric alkene with two leaving groups, of which one selectively

leaves to generate chirality. A fourth method is when we have an unsymmetric -allyl complex,
where one face of palladium coordination is favored. This depends on the rate of isomerization of

the palladium coordinated complexes and the relative rates of reaction of each.

Examples can be seen in Stoltz, ACIE, 2006, 45, 3109 and Trost, JACS, 1999, 121, 3543, where

a Bu4 NCl is used to facilitate isomerization of the two palladium disasteromers.

32
12 C-H Activation

C-H activation is the process of modifying C-H bonds to C-X or C-C bonds. It's been called the

holy grail of chemistry - basically, how do we change something so inert (BDE 90-105 kcal/mol,

pKa 45-50) into something reactive? Despite the relative stability of C-H bonds, they are actually

pretty easy to break - just combust it in oxygen! The problem becomes how we just selectively

activate one 1 C-H bond. (Review: ACR, 1995, 28, 3, 154.)

Some typical reactions we want to perform include the conversion of methane to methanol, or the

cleavage of a CH3 bond in the prescence of a CH bond (which is easier to cleave). We also need

to be wary of overoxidation, since the product C-H bonds are weaker than the starting material.

Recall that metal sigma complexes have two binding modes, on a continuium from backbonding

to complete sigma-donation. The former is known as Bergman-type and is exlemplied by Ir(I)

complexes, while the latter is known as Shilov-type and is exlemplied by Pt(II) complexes.
In Shilov complexes, C-H cleavage occurs in a heterolytic manner, and there is no oxidation state

change of the metal. For example, this happens in the oxidation of methane to methanol in the

prescence of K2 PtCl2 , with K2 PtCl6 as the stoichiometric oxidant.

The mechanism of this transformation starts with Cl2 Pt(OH2 )2 , where methane displaces water

(!!) [due to the softer nature of Pt] to make a sigma complex, which then becomes deprotonated

by chloride to generate Cl2 Pt(OH2 )Me . The oxidant oxidizes the Pt to Pt(IV) with the addition of

two chloro ligands, and then water is lost, generating Cl4 PtMe . Then, an SN 2-like pathway (shown
2
by stereochemical studies) with water results in the loss of methanol and creation of PtCl4 , which

then goes on to regenerate the catalyst. H/D exchange studies has shown that primary C-H bonds

react faster than secondary and tertiary C-H bonds.

This method, however, is not very practical due to the use of K2 PtCl6 stoichiometric oxidant, which

has given hope that it could be replaced with another oxidant. Indeed, the Catalytica process using

sulfuric acid as the oxidant has been developed. It involves the reaction of methane with H2 SO4

(as the solvent) to generate methanesulfonic acid, in the prescence of a PtCl2 (diamine) catalyst.

The mechanism for this process is very similar, with a starting (diamine)Pt(OSO3 H)2 catalyst

that has one of the sulfonic acids displaced by methane, which then generates the methyl complex.

Sulfuric acid then oxidizes the Pt to a 6-coordinate complex, from which either reductive elimination

or a SN 2-like pathway occurs to generate methanesulfonic acid.

A similar C-H activation of benzene has been developed by Crabtree, in where benzene (as the

solvent as well) in the prescence of Pd(OAc)2 , PhI(OAc)2 , AcOH becomes oxidized to acetoxyben-

33
zene. (J. Mol. Catal. 1996, 108, 35). The I(III) oxidants are decently mild (especially compared

to the Catalytica process).

To avoid solvent-quantities of substrate and site-selectivity, directing group strategies have been

used, with the I(III) catalysts. For example, in the acetyloxylation of 4-azaphenanthrene, the 5-

position gets selectively aectyloxylated, due to the coordination of the nitrogen. Other groups such

as Br, F, and OMe can also be incorporated. Further, other directing groups such as oximes and

acids can also be used (Sanford, JACS, 2004, 126, 2300).

Using I(II) reagents, we can also make C-C bonds. For example, 3-methyl,2-phenylpyridine, under
+ 
the action of Ph2 I BF4 attaches a phenyl group to the 2' position of the phenyl group. The methyl

group doesn't provide any selectity, rather, it is just there to prevent overoxidation (restricting

bond rotation). These I(II) reagents result in the transfer of the less hindered aryl group, so a
+ 
MesIPh BF4 reagent would result in the transfer of a phenyl group. (JACS, 2005, 127, 7330).

C-H activation can also be achieved with a Pd(0)/Pd(II) cycle, instead of the Pd(II)/Pd(IV) cycles

discussed earlier. Benzoquinone and silver oxide is used, in the prescence of a boronic acid, which

ultimately gets incorporated in the same 2' position in the systems above. (Yu, 2006, 128, 73).

We may also induce enantioselectively with chiral amino acids (ACIE, 2008, 47, 4882).

Now, how do we get non-directed C-H activation? Electron-rich substrates like an indole will react

with, for example, Pd(OAc)2 at the 2 position, which can further go on to react with I(II) to

incorporate aryls. (JACS, 2000, 128, 4972). Electron-decient arenes will also perform the same

reaction. For example, the reaction of with F5 C6 H with 4-bromotoluene results in coupling of the

H-Br bond. (Fagnou, JACS, 2006, 128, 16496)

The mechanism for concerted arylation involves rst oxidative addition as usual, then a displace-

ment of the halide with carbonate, creating a bidentate carbonate ligand. The relatively acidic

peruoroarene then is able to be deprotonated and concertedly attached to the palladium while the

carbonate leaves, after which reductive elimination can take place. This mechanism is supported by

higher reactivity or more acidic species, as well as large KIEs, and the fact that bromide or acetate

ligands also worked.

If we use the substrate as a cosolvent, then electron-neutral arylation can also take place. We can

then cause a dehydrogenative cross-coupling between an electron rich indole at the 3 position with

p-diuorobenzene. (Science, 2007, 376, 117)

For allylic activation, Trost found that Pd(II) salts could also aect it, but in stoichiometric amounts

(ACR, 1980, 13, 385). The challenge was to have both the electrophilic C-H activation and the

nucleophilic formation of the -allyl complex.

34
We can get around this by using benzoquinone and MnO2 as oxidant, and AcOH as nucleophile,

in order to impart allylic acetyloxlation. (Buckwall, Chem. Eur. J, 1998, 4, 1083). Wacker-

type products can also result, depending on conditions. Specically, the amount of Wacker-type

products was much greater in pure AcOH as compared to a 1:1 AcOH/DMSO mixture. Thus, it

was hypothesized that DMSO-like ligands could improve selectivity, and indeed it did (though to

the branched isomer) (White, JACS, 2004, 126, 1346). An example is in the total synthesis of

Hippolachnin A, which used a late stage acid-directed C-H cyclization (JACS 2016, 138, 7, 2437).

Now let's take a look at Bergman-like complexes, those with pi-backbonding character. These

often involve Ir and Rh complexes, and were some of the rst complexes discovered to induce

C-H activation. Essentially, starting with a 3 or 4 coordinate ligand, treatment with heat and

light causes the loss of ligand and formation of a 2-ligand 16 electron complex, which then can

coordinate with a C-H bond either in a concerted or radical fashion (JACS, 1983, 105, 3929).

Kinetic studies showed that there was a strong preference for activation of less hindered positions,

with the dierence being more profound for Rh rather than Ir. These catalysts can then be used

to cause dehydrgenation (JACS 1979, 101, 7738)

Using similar systems, other methods such as selective C-H borylation have been developed. Mech-

anistically, the Iridium precatalyst (diamine)Ir(cod)(Bpin)3 loses a cod ligand, and then oxidative

addition/coordination of the aryl hydride results in a 7-coordinate Ir catalyst. Reductive elimination

of PhBpin followed by addition of B2 pin2 makes the product and regenerates the catalyst (Science

1997, 297, 211 and 2000, 207, 1995). Ortho-directed alkylation systems based on carbonyl and

imine chelation (Murai, Nature 1993, 366, 529 and JACS 1996, 118, 493) have also been reported.

35
13 Metathesis

Metathesis is a catalyzed cleavage and reassemblage of alkenes, esssentially shifting substituents

between alkenes. It is almost always under thermodynamic control just due to its reversible nature.

It was rst discovered when looking at Ziegler-Natta polymerization catalysts, where ring-opened

products were found in the polymerization under strong lewis acids. Though these lewis-acid based

systems are sometimes used in industry, they often have low functional group compatability and

control. As a result, more mild homogeneous catalysts based on Mo, W, Ta, Ti, or Ru have been

developed.

The mechanism for metathesis starts from a metal alkylidene, which rst loses ligand and creates

of a metal-carbon double bond via reaction with an olen. It then reacts with another olen to

create a 4-membered metallocycle via a [2+2] addition. A retro [2+2] provides for the product and

another metal alkylidene. This can react with another equivalent of the alkene then loses ethylene

under another retro [2+2], thus, ths loss of ethylene can drive the reactions forward. This is why

the reactions are sometimes done in slight vacuum.

The metathesis relies heavily on metal alkylidenes, which were rst synthesized by Schrock starting
+ 
from Cp2 TaMe3 , which was reacted with CPh3 BF4 and base to produce Cp2 Ta( 
 CH2 )Me

(Schrock, JACS, 1975, 97, 6577).

These carbenes are electrophilic at the metal and nucleophilic at the carbon. As a result, they

are highly reactive towards polarized  systems like carbonyls and can act as olenating reagents,
for example, in the Tebbe reagent and the Petathesis reagent that have Cp2 Ti 
 CH2 as its active

catalyst (Tebbe, JACS 1978, 100, 3611). These reagents are more nucleophilic and less basic than

Wittig reagents, and hence provide a method for the reaction of less reactive functional groups.

Grubbs was able to show that the Tebbe reagent was able to act as a metathesis catalyst in the

polymerization of norbornene, through a metallocyclobutane. (Grubbs, JACS, 1980, 102, 6876).

From this, metathesis catalysts such as Schrock's catalyst (based on Mo) and Grubbs' catalyst

(based on Ru) were developed.

In general, as we move further to the right of the periodic table and metals become less oxophilic,

we get higher olen/carbonyl selectivity. Thus, Ti/W catalysts are generally unusable, Mo catalysts

are usable but react with air and water and have low functional group tolerance, and Ru catalysts

work ne.

The active catalyst can be generated from the precatalyst through the disassociation of the phos-

phine ligand. In the second generation catalysts the carbenes slows both the disassociation, but

36
slows the association even more, thus shifting the equilibrium to the active form of the catalyst.

The 3rd generation catalyst is also slow to initiate but more stable. There are also fast-activating

catalysts, for example, by using pyridine as ligand in ROMP.

Since metathesis is often based on equilibriating mixtures, how do we predict which products get

formed? Grubbs proposed a classication of olens into four types:

 Type 1: rapid homodimerization, which can react further.

 Type 2: slow homodimerization, which cannot react further.

 Type 3: no homodimerization.

 Type 4: spectator olen that does not deactivate the catalyst.

These classications dier based on catalysts. Terminal olens, and allyl silyl olens are type 1,

while allyl alcohols and styrenes are either type 1 or 2. 1,1-disubstituted olens are often type 3 or

4, while vinyl nitro olens are almost exclusively type 4. (JACS, 2003, 125, 11360)

Now based on these classications, we have some rules for cross-metathesis:

1. Two type 1 olens reacting will lead to statistical mixtures. To achieve selectivity, we need a

large excess (about 10-fold) of one of the alkenes.

2. Reactions between two non-type 1 olens will lead to non-selective outcomes (bad!)

3. Reactions between two dierent types of olens are generally good. Even though they initially

may be poorly selective, the homodimers could still react and form the desired product when

ethylene is removed from the system.

We can use metathesis catalysts to aect ring-opening, as well as enantioselectivity (resulting from

ring-closing). In general, medium sized rings (5, 6, 7) are the most reactive, while macrocyclizations

can be more challenging, but still possible.

A technique known as Relay ring-closing metathesis is also possible for hard macrocyclizations.

Essentially, you append on a (CH2 )3  CH 

 CH2 unit to the terminal olen, which the ruthenium

reacts with. Loss of cyclopentadiene then makes the ruthenium bind to the original olen, improving

reactivity. (Porco, ACIE, 2004, 43, 3601).

A modication is in ene-yne metathesis, in where we essentially have metathesis with an alkene

and an alkyne. The ruthenium catalyst can react with either component and join them together,

leading to the same product.

37
14 Carbenoids

Note: This is the nal lecture of the class.

Carbenes are classied into two types based on electronic structure - Fischer carbenes and Schrock
carbenes. Fischer carbenes often are nucleophilic at the metal and electrophilic at the carbon, and

have EDG substituents on the carbene and -acceptor ligands on the metal. Additionally they

often have low oxidation state metals. In essence, Fischer carbenes are carbenes that favor a

negatively-charged metal.

Schrock carbenes are the opposite - they are nucleophilic at carbon instead and often have H and R

groups on the carbene. They have -donor ligands instead and are often in a high oxidation state.
We actually have a third type of carbenes, known as carbenoids. These carbenoids have an EWG

acceptor such as a carbonyl, and have never been isolated. Their reactivity instead resembles

that of singlet carbenes. This is because the carbene essentially interacts only through a weak

-backbonding interaction and a strong  donor interaction. Since the backbonding interaction is

weak, the carbene is still quite electrophilic.

The discovery of these carbenes was accidental - when a Wol rearrangement was attempted to

be performed on an -diazoketone with a copper catalyst and ethanol, the copper instead was able

to make a carbene and insert itself into ethanol, making unexpected products. (JACS 1952, 74,

5376).

In general, copper and rhodium are used for the generation of carbenes. Cu(I) is often used for

the addition of carbenes to -bonds, and can do this enantioselectively. We have a few types

of precursors such as -diazoketones (acceptor), -diazo-1,3-diketones (acceptor/acceptor), and

-diazoarylketones (acceptor/donor).

The mechanism starts with Cu coordinating to one of the olens, then to the diazo reactant

(through the nitrogen). Loss of N2 then generates the Cu carbenoid, which then can react to form

a cyclopropane. Coordination of another alkene then regenerates the catalyst.

Rh(II) dimers are instead more eective in C-H insertion reactions, in addition to cyclopropanation.

The carbene interacts with the  orbital of the rhodium dimer and the rhodium interacts with the
 orbital of the carbene. Though the basic C-H insertion (for example, with decomposition of a

diazoketone with Ag2 O) is feasible without catalyst, using rhodium improves the selectivity much

more. Rhodium often forms 5-membered rings, with the more electronically activated position

reacting with the carbenoid species. The C-H insertion is stereoretentive (JACS, 1985, 107, 196).

38
Two mechanisms have been proposed. The rst involves coordination of the rhodium to the carbon

bonded to the diazo group, which then has a rate-determining loss of N2 to form the carbenoid.

Now, the path to form the product can proceed through a 3-center-2-electron pathway, or instead by

breaking the Rh-Rh bond (forming a Rh(III)/Rh(I) pair) and causing subsequent oxidative addition

and reductive elimination to form the product. The rst mechanism is generally favored.

These carbenoids can take on a few forms, based on many heterocycles such as proline-derived

and phthalimido-derived substituents replacing the acetates from Rh2 (OAc)4 . These catalysts can

improve selectivity and induce enantioselectivity.

Intermolecular carbenoid C-H insertion is much harder, but possible. There may be issues with

chemoselectivity, functional group tolerance, or carbene dimerization. These carbenoids are often

donor/acceptor carbenes, and can insert into substrates such as Boc-piperidine or cyclohexane.

These insertions can create substrctures that resemble products from Aldol addition, Michael

addition, or Mannich reactions, just to give a few examples. O-H and N-H bond intermolecular

insertions are often possible, with Fu ligands.

Rh dimers have also been seen to induce nitrene insertions, for example. We can use carbamates

to insert and form 5-membered rings, while using sulfamides favors forming 6-membered rings.

This allows for syn-selective 1,2 and 1,3 amination from the alcohol (JACS, 2001, 123, 6955).

Using Rh2 (esp)2 catalysts on, 1,3 diamidosulfamates can also be formed. This catalyst can also

promote intermolecular stereoretentive insertion. Other catalyst systems can also be used to induce

enantiospecicity.

39