uva-dare (digital academic repository) group 9 open-shell ...can play a dominant role in directing...

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Group 9 open-shell organometallics: reactivity at the ligand

Dzik, W.I.

Link to publication

Citation for published version (APA):Dzik, W. I. (2011). Group 9 open-shell organometallics: reactivity at the ligand.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 16 Jul 2020

https://dare.uva.nl/personal/pure/en/publications/group-9-openshell-organometallics-reactivity-at-the-ligand(19e7a683-660b-440c-8dda-e54b36bd828d).html

Group 9Open-Shell Organometallics:Reactivity at the Ligand

Group 9 O

pen-Shell O

rganometallics: R

eactivity at the Ligand Wojciech I. D

zik 2011

Invitation

to the public defence of my Thesis

which will take place on:

Tuesday, 11th of January 2011 at 10:00 in the

Agnietenkapel Oudezijds Voorburgwal

231 Amsterdam

After the ceremony a reception will take place

Wojciech [email protected]

Paranymphs:

Erica [email protected]

Paweł [email protected] I. Dzik

Group 9 Open-Shell Organometallics:

Reactivity at the Ligand

Wojciech I. Dzik

Cover: Radical reaction (taking place on a redox non-innocent ligand) approaching the transition state. Photo: © Adam Gryko / istockphoto.com

Group 9 Open-Shell Organometallics:

Reactivity at the Ligand

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus

prof. dr. D.C. van den Boom ten overstaan van een door het college voor promoties

ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel

op dinsdag 11 januari 2011, te 10:00 uur

door

Wojciech Iwo Dzik

geboren te Columbus, Verenigde Staten van Amerika

Promotiecommissie Promotor: Prof. dr. J. N. H. Reek Co-promotor: Dr. B. de Bruin Overige Leden: Prof. dr. C. J. Elsevier

Dr. D. G. H. Hetterscheid Prof. dr. K. Lammertsma Dr. ir. J. I. van der Vlugt Prof. dr. K. Wieghardt Prof. dr. X. P. Zhang

Faculteit der Natuurwetenschappen, Wiskunde en Informatica The research described in this dissertation has been financially supported by the Netherlands Organization for Scientific Research – Division Chemical Sciences (NWO-CW) within VIDI project 700.55.426

Table of Contents Chapter 1 General Introduction: Redox Non-Innocent Behavior of

Alkene, Carbonyl and Carbene Ligands 1

Chapter 2 Activation of Carbon Monoxide by (Me3tpa)Rh and (Me3tpa)Ir

21

Chapter 3 Carbonyl Complexes of Rhodium with N-donor Ligands: Factors Determining the Formation of Terminal versus Bridging Carbonyls

49

Chapter 4 Base Assisted Selective Reduction of Organometallic RhII and IrII Radicals

69

Chapter 5 Selective C–C Coupling of Ir-Ethene and Ir-Carbene Radicals

87

Chapter 6 ‘Carbene Radicals’ in CoII(por)-Catalyzed Olefin Cyclopropanation

99

Chapter 7 Hydrogen Atom Transfer in Reactions of Organic Radicals with (por)CoII• and in Subsequent Addition of (por)Co–H to Olefins

121

Chapter 8 Perspective: Ligand Radicals in New, Catalytic Reactions 137

Summary 143

Samenvatting 149

Streszczenie 155

List of Publications 161

Acknowledgements 163

Chapter 1

General Introduction:

Redox Non-Innocent Behavior of Alkene, Carbonyl and Carbene Ligands

Chapter 1

2

1.1 Introduction

Radicals are often considered to be too reactive to be selective. For free radicals this might be true, but in the coordination sphere of a transition metal highly chemo-, regio-, diastereo- and enantioselective radical-type reactions become feasible. Selectivity is primarily a matter of relative rather than absolute rates, and hence also fast radical-type reactions can be (made) selective using the steric and electronic influence of ligand surroundings in transition metal controlled reactions.1

Nature is teaching us clearly that selective radical reactions are certainly possible and radical-type reactions are actually quite common in several biochemical processes. Many metallo-enzyme catalyzed transformations proceed via radical-type reactions involving open-shell intermediates. They are essential in mediating difficult reactions such as C–C or C–O bond formations or transformations, and despite their radical nature these reactions still proceed in the highly selective manner typical for enzymatic reactions.2

Selective radical-type atom transfer reactions most frequently require the presence of unpaired electron (radical) density at a reactive ligand site (e.g. oxo / oxyl radical ligands),3 as is clear from mechanistic studies of bio-catalytic processes mediated by among others cytochrome P450,

4 methane monooxygenases,5 class I ribonucleotide reductase6 or galactose oxidase.

CuN

ON

OO

S

H

HH

II CuN

ON

O

OHS

H

H

II

CuN

ON

HO

OHS

H

I

CuN

ON

O

OHS

O

H

IIO

HR

R

O2

RR

H2O2

RCH2OH

Tyr

Tyr

His

His

Figure 1. Cooperative ligand radical in the mechanism of alcohol oxidation by galactose oxidase.

The mechanism of Galactose Oxidase (GOase) is perhaps one of the most illustrative examples in showing how radicals can be controlled in the coordination sphere of a transition metal to achieve selectivity (Figure 1).7 This fungal enzyme chemo-selectively converts primary alcohols with O2 into aldehydes and hydrogen peroxide with impressive second order rate constants (~1.59 x 104 M−1 s−1).8 The active site of GOase consists of a redox active CuI ion in proximity to a redox active cysteine modified tyrosine. Oxidation of the active site by dioxygen produces a

General Introduction

3

tyrosinyl radical coordinated to CuII and a hydroperoxo ligand. In the key subsequent steps of the catalytic cycle the thus obtained hydroperoxo ligand deprotonates the alcohol to form free hydrogen peroxide and an alcoholate ligand. The tyrosinyl radical ligand subsequently abstracts a hydrogen atom from the α-CH functionality of the alcoholate to produce the aldehyde. Remarkably, the tyrosinyl radical is a reactive fragment in the catalytic cycle and functions as a cooperative ligand, assisting copper in the turn-over process. The ligand redox activity allows removal of the second electron from the substrate, and thereby a two-electron oxidation using a copper ion that is normally only capable of one-electron transformations. It is the interplay of the one-electron reactivity of the ligand radical and the one-electron reactivity of the metal which makes the overall two-electron oxidation process possible. The cooperative radical-type process makes the reaction fast and selective.

Metal-ligand bond homolysis within an enzyme pocket is another frequently observed pathway to start selective radical-type transformations. The typical radical-type reactions mediated by cobalamine9 (coenzyme B12) based dependent mutases (allowing carbon skeleton rearrangements)10 and diol dehydrase or ethanolamine deaminase11 are initiated by homolytic splitting of a weak cobalt(III)-carbon bond (Figure 2).12 The thus formed carbon radical serves as a hydrogen atom acceptor from the substrate CH group. After the first hydrogen atom transfer (HAT) step the open shell substrate rearranges through the migration of a functional group to the radical carbon followed by the second HAT to form the closed-shell rearranged product. The •CH2Ad radical remains in close proximity to the cobalt(II) centre reforming the cobalt-carbon bond after the catalytic turnover.

N

NN

N

H2NOC

CONH2

CONH2

CONH2

H2NOC

H

H2NOC

Co

NH

O

O

PO O

O

HO

OH

N

N

O

R

OH2C N

HO OH

N

N N

NH2

R = CH2Ad =

CoIII

N

CH2Ad

CoII

N

CH2Ad

CoII

N

CH3Ad

C C

XH

C C

X

CoII

N

CH3Ad

C C

X

C C

HX

Figure 2. Structure of coenzyme B12 and the generic mechanism for AdoCbl-dependent rearrangement reactions.

Chapter 1

4

The mechanisms of reactions catalyzed by galactose oxidase or coenzyme B12 are excellent examples of how radical-type reactions can be controlled in the coordination sphere of a transition metal, and serve as a clear inspiration for synthetic chemists. Interesting bio-mimetic approaches to GOase have already shown synthetic power in alcohol oxidation reactions,13 but future studies are likely to uncover a wide field of new opportunities. The ease of reversible generation of a carbon centered radical by the cleavage of the weak Co–C bond (used by Nature in the coenzyme B12) has found wide application in transition metal controlled radical polymerization.14 The latter example shows elegantly that the same phenomena (i.e. homolytic cleavage of a weak metal-carbon bond) that are used in enzymes for certain transformation in vivo can be successfully applied to mediate reactions (like polymerization of olefins) that occur only in chemical laboratories. Some other important examples of synthetic (catalytic) reactions that proceed via open-shell intermediates are atom transfer radical polymerization,14a atom transfer radical addition,15 and oxygenation reactions.16 For most of these reactions, atom or group transfer to a reactive metal-centered radical (metallo-radical) occurs at some stage of the catalytic reaction, but properly designed synthetic transition metal complexes should allow us to activate substrates in a similar (cooperative) way as Nature does. Especially in the field of organometallic chemistry, the playground of catalytic synthetic-organic transformations, controlling the radical-type reactivity of open-shell organometallic compounds can offer vast opportunities to achieve a wide variety of new, fast and selective catalytic transformations proceeding via radical-type rearrangements, and open-shell coupling, abstraction, addition, (single or multi-component) cyclization, and/or disproportionation reactions.

Due to the ligand redox non-innocence of many π-accepting ligands, organometallic complexes offer many possibilities to form ligand radicals, which can play a dominant role in directing the open-shell reactivity of the compounds. Such behavior offers further interesting and new opportunities to achieve selectivity and new reactivity in the field of open-shell organometallic catalysis.

In further sections the concept of ligand redox non-innocence will be described, followed by an overview of one-electron reactivity of olefins, carbonyls and carbenes where redox non-innocence plays a crucial role.

1.2 Ligand redox non-innocence

When considering the properties and reactivity of paramagnetic species the location of the unpaired electron(s) is an important issue to address. For many open-shell compounds the atoms or groups bearing the largest unpaired spin density are also the most reactive sites of the molecule. In this Chapter we are primarily considering low-spin complexes containing only a single unpaired electron (S = ½ systems). This does not mean that their electronic structure is always as simple as it seems, because the unpaired electron can be located at the metal, at the ligand, or in-between. Some species should be regarded as distinct metal or ligand radicals, whereas in other species the ligand and metal radical descriptions are limiting resonance structures of the real electronic structure (Figure 3).


5

Figure 3. Possible electronic structures of an open-shell complex: (a) metal radical; (b) covalent situation; (c) ligand radical.

In formal oxidation state counting, redox processes of transition metal complexes are always assumed to occur at the metal. Such formal oxidation states are appropriate for ‘classical’ Werner-type coordination compounds, and in combination with convenient ligand-field approximations the thus obtained d-electron configuration is a quite useful concept, applicable for spectroscopic interpretations or predictions of the magnetic behavior of these complexes. However, for complexes bearing redox active ligands, formal oxidation state changes do not always reflect true changes in the d-electron configuration of the metal. Such redox active ligands are usually ligands which undergo π-bonding with the metal and/or have an extended π-system enabling them to delocalize accumulated charges and unpaired spin density upon oxidation or reduction of the complex. In other words, the actual redox process may occur at a ligand site rather than the metal, hence leaving the metal d-electron count unchanged. For such complexes determining the true (spectroscopic) oxidation states of the metal (d-electron configuration) and/or the ligands (ligand electron configuration) requires a detailed spectroscopic study, and the thus obtained ‘spectroscopic oxidation state’ can differ markedly from the formal oxidation state. In such situations the term ‘ligand redox non-innocence’ is used to underline the oxidation state ambiguity.17

The involvement of ligand redox non-innocence in the mechanisms of several enzymes is meanwhile well established. Redox non-innocent ligands play an active and important role in many (bio)catalytic substrate activation mechanisms, often in cooperation with the transition metal to which they are coordinated.18 Galactose Oxidase (see section 1.1) provides a seminal example for the involvement of a cooperative and reactive redox non-innocent ligand playing an active role in the bio-catalytic turnover process of this enzyme (Figure 1).7

Many organometallic complexes bear ligands which undergo π-bonding with the metal (e.g. CO, olefins and carbenes) some of which even have an extended π-delocalized system (e.g. Cp-derivatives, multi-enes and multi-enyls) and are thus capable of easy charge and spin delocalization upon oxidation or reduction of the complex. Such ligands are by definition suspect, and can easily get oxidized or reduced instead of the metal to which they coordinate. Ligand redox non-innocence indeed plays an important role in the reactivity of open-shell organometallic compounds, although this is not yet so broadly recognized within the organometallic community.

The concept of ligand redox non-innocence in organometallic compounds can be easily understood by looking at the simplified orbital interaction schemes shown in Figure 4.

Chapter 1

6

metald-orbital

filled ligand orbital

σ/π

σ∗/π∗ empty ligand orbital

π

π∗

metal d-orbital

M-L σ or π bonding M-L π back bonding

metald-orbital

filled ligand orbital

σ/π

σ∗/π∗

empty ligand orbital

π

π∗

metal d-orbital

'oxidized ligand' 'reduced ligand'

metald-orbital

ligand orbital

σ/π

σ∗/π∗

'Ionic bonding' in (Werner-type) M−L σ-bonding and π-back bonding situations: Formal oxidation state = spectroscopic oxidation state

Covalent bonding situations:Oxidation state assignments impossible

'Ionic bonding' with oxidized or reduced redoxnon-innocent ligands: Spectroscopic oxidationstates are different from formal oxidation states.Spectroscopic oxidation states appropriate

Figure 4. Top: common ionic M−L binding; Middle: covalent M−L binding; Bottom: ionic M−L binding with ‘redox non-innocent’ ligands.

In common Werner-type coordination compounds, the metal d-orbitals are generally substantially higher in energy than the σ-donor type filled ligand orbitals, while their energy is commonly lower than that of the empty ligand π*-orbitals. Therefore the unpaired electrons are typically (primarily) localized at the metal. Radical reactions initiated by or occurring at such complexes are usually metal-centred. The essential assumption in formal oxidation state counting is that all σ-type metal-ligand interactions are mainly ionic with only small covalent contributions. This means that in using MO descriptions the (filled) valence ligand orbitals used for ligand-to-metal bonding (either σ-bonding or π-bonding) must always be substantially lower in energy than the metal d-orbitals (see Figure 4, top left). Most organometallic complexes also contain π-acceptor ligands. These ligand types contain relatively low-lying empty π*-orbitals capable of accepting electron density from (the appropriate π-type) metal d-orbitals. Textbook examples of such ligands are CO, NO, and olefinic ligands. The essential assumption made when regarding binding of π-acceptor ligands in formal oxidation state counting is again a large ionic character of the metal-ligand π-interactions, meaning that the ligand valence orbitals involved in the M−L π-back bonding interaction must always be substantially higher in energy than the metal d-orbitals (Figure 4, top right).

Formal oxidation states that go along with these ionic bonding interactions are quite appropriate for ‘classical’ Werner-type complexes, but do not hold for those compounds having more covalent character of the metal-carbon bonds. With strong mixing of more closely energetic metal and ligand orbitals (increasing covalency), formal oxidation


7

states become quite misleading, and no longer represent the true ‘d-electron’ configurations as the bonding and anti-bonding combinations can no longer be assigned being either of mainly ligand or mainly metal character (Figure 4, middle). The ‘real’ or spectroscopic oxidation state of such complexes is best described as in-between the two limiting metal radical and ligand radical ‘resonance structures’ (Figure 3b).

In certain complexes with redox non-innocent ligands the relative metal and ligand orbital energies can be reversed (Figure 4, bottom). For complexes containing oxidizable ligands, the metal d-orbitals can end-up lower in energy than the ligand orbitals, thus effectively leading to reduction of the metal and oxidation of the ligand. This situation typically involves (but is not restricted to) complexes in high formal oxidation states. In such cases we speak of ‘ligand redox non-innocent’ behavior. For S = ½ systems this has important consequences, as the ligand(s) now contain(s) unpaired spin-density (corresponding to x electrons), and the filling of the metal ‘d-orbitals’ (dn+x) is in reality higher than predicted by the formal oxidation state (dn). Obviously, ‘redox non-innocent’ behavior of ligands is also possible when π-acceptor type orbitals lie lower in energy than the metal d-orbitals. This situation typically involves (but is not restricted to) complexes in low formal oxidation states, and effectively leads to reduction of the ligands. Also in these cases, the ligand(s) end-up containing unpaired spin-density, and the actual filling of the metal ‘d-orbitals’ (dn-x) is now lower than predicted by the formal oxidation state of the metal.19

Ligand redox non-innocence (Figure 4, bottom) leading to formation of ligand radicals (Figure 3c) is an important aspect because in many cases the presence of discrete and substantial spin density at a ligand fragment leads to completely different reactivity patterns than commonly observed for these ligands in diamagnetic, closed-shell complexes. The ligand often becomes the most reactive site of the complex in such situations. Nonetheless, such reactive ligand radicals can undergo selective reactions, very different from the uncontrolled reactivity of free radicals. This is possible because the spin density is always shared between the metal and the ligand to some extent (Figure 4, bottom), which leads to higher stabilization of the ligand centered radicals. Moreover, besides their electronic influence, the transition metal and its ligand surroundings can have a profound steric influence on the radical-type reactivity of the substrate. Ligand redox non-innocence can thus be exploited in selective one-electron activation of substrates that usually undergo two-electron transformations, which significantly changes their reactivity patterns and gives access to fast, controlled and selective radical-type organometallic reactions.20 1.3.1 Open-shell reactivity of alkene and alkyne ligands

The π*-orbitals of alkenes coordinated to paramagnetic transition metals can accept (part of) the unpaired electron(s) from the transition metal, thus forming carbon centered radicals. Hence, the π-accepting alkene fragment is potentially ‘redox non-innocent’, and ‘metallo-radical’ and ‘alkene radical’ descriptions should both be taken into consideration for paramagnetic M(alkene) complexes (Figure 5). A similar behavior can be expected for M(alkyne) complexes. For paramagnetic RhII and IrII species this has become quite clear from their unusual ligand based reactivities.

Chapter 1

8

M Mn+ (n+1)+

Figure 5. Redox non-innocence of alkene ligands.

Recently, a quite detailed overview concerning the properties and ligand centered radical-type reactivity of open-shell paramagnetic Rh(alkene) and Ir(alkene) species has been presented.21 Here only the most illustrative examples will be summarized.

Rhodium(II)- and iridium(II)-porphyrin (por) complexes tend to react with alkenes and alkynes to form hydrocarbon bridged dinuclear MIII(por) species. Ogoshi et al. reported that Rh(OEP) (OEP = octaethylporphyrinato) generated in situ from the Rh(OEP) dimer reacts with acetylenes to form vinylidene bridged species22 (Figure 6). Several M(por) systems (M = Rh, Ir) react with a variety of different olefins to form (por)M−CH2−CHR−Μ(por) bridged species.23-27 Reacting Rh(por) species with olefins containing allylic hydrogens can also result in allylic radical-type H-atom transfer, pointing to radical-type disproportionation of the proposed β-alkyl radical intermediates (por)Rh−CH2CHR• (Figure 5).

With more bulky porphyrins, two of such β-alkyl radical intermediates (por)M−CH2CRH• can couple to form four-carbon −CH2−CHR−CHR−CH2− bridged species.27,28 Binuclear rhodium complexes with tethered porphyrins react faster with ethene forming the −CH2−CH2− bridged species due to lowering the unfavorable entropic term associated with these tri-molecular reactions. With 1,3-butadiene the unsaturated 2-buteno C4 bridged species are formed.29

Figure 6. Reactivity of open-shell RhII(Por) species with alkenes and alkynes.


9

The reactions of olefins with diamagnetic binuclear Rh2(OEP)2 complexes containing a (relatively weak) Rh−Rh bond likely proceed via radical chain processes (Figure 7), and the intermediacy of the proposed (OEP)Rh−CH2−CHR• ligand centered radicals was evidenced by radical trapping.23 More recent studies revealed that (por)Rh−CH2−CHR• ligand centered radicals are likely in equilibrium with metallo-radical (por)Rh(η2-alkene) species.27,28,30

Figure 7. Radical-chain mechanism operative in the reaction of Rh2(OEP)2 with olefins.

The chemistry of one-electron activated cot (cot = cyclooctatetraene) coordinated to Fe,31 Ru,32 Co33 or Rh34 has been systematically studied by the group of Connelly, revealing their ability for radical-type C−C bond forming reactions resulting in binuclear complexes with interesting structures (Figure 8). The η5-allyl structure of Co2(η5:η'5-C16H16)(Cp*)2 is a likely structure of an intermediate in the formation of the Fe, Ru and Rh dimeric complexes.20c

[M] [M] = Fe(CO)2(L)L = CO, P(OMe)3

-e

+[M]

[M]+

[M]+

+[M]

+[M]+[M]

M = Ru(CO)3M = Rh(Cp)

M = Co(Cp*)

-e

-e

Figure 8. Ligand-ligand radical-coupling of cot ligands upon one-electron oxidation of Fe(cot) complexes.

The above reactions clearly show that radical-type coupling reactions can proceed in a very selective manner within the coordination sphere of transition metals, and provide interesting synthetic opportunities. For most of these reactions the mechanistic details are not well understood, but more information about the involvement of ligand radicals in ligand-to-metal radical coupling reactions is available from recent studies of open-shell iridium-ethene complexes employing

Chapter 1

10

bulky tetradentate Mentpa nitrogen ligands (n = 3: Me3tpa = tris(6-methyl(2-picolyl))amine and n = 2: Me2tpa = bis(6-methyl(2-picolyl))(2-picolyl)amine). These [IrII(Mentpa)(ethene)]2+ complexes are relatively stable in weakly coordinating solvents, but undergo selective radical-type reactions at the coordinated ethene moiety in MeCN (Figure 9). This has been illustrated by selective formation of [IrIII(CH2CHO)(Mentpa)(MeCN)]2+ in reaction with dioxygen35 or C2-bridged [(Mentpa)(MeCN)IrIII−CH2CH2−IrIII(Mentpa)(MeCN)]4+ dinuclear species in pure MeCN.36 Kinetic studies, radical trapping experiments and supporting DFT calculations reveal that the reactions must proceed via M−CH2CH2• ligand radical intermediates formed upon MeCN coordination.

N

N NN Ir

N

N NN Ir

N

NN

NIr

NCMe

MeCN

N

N NN Ir

NCMe

N

N NN Ir

NCMe

N

N NN Ir

NCMe

O

2+

IIMeCN

III

III

4+

MeCN

-

2+

III

2+

II

2+

III

O2

-OH

Figure 9. Involvement of Ir−CH2−CH2• ligand radicals in the open-shell organometallic chemistry of IrII(ethene) complexes.

The analogous complex with propene does not undergo coupling reactions. Instead hydrogen atom transfer from the allylic position leads to formation of a 1:1 mixture of an iridium allyl and an iridium hydride complex.37 Similar activation of the allylic C−H bond on paramagnetic compounds was also reported for a series of cod complexes (cod = 1,5-cyclooctadiene). One-electron oxidation of the Rh(Cp-derivative)RhI(cod) complex shown in Figure 10, for example, leads to formation of an ene-η3-allyl complex.38

Figure 10. Radical-type activation of allylic C−H bonds of the cod ligand.

The mechanism of this type of transformations was studied further with a series of [(N3-ligand)MII(cod)]2+ complexes (M = Rh, Ir). These paramagnetic cod rhodium and iridium complexes, supported with bis-picolylamine type of ligands, easily transform into 1:1 mixtures of η3-allyl complexes and M−H species (Figure 11). For


11

iridium the latter hydride species are stable, but for rhodium the hydride undergoes reductive deprotonation by the ligand to form a [RhI(cod)(N3ligandH)]2+ species.37,39 Kinetic measurements showed a second order rate dependency on the concentration of the complex. The proposed mechanism involves hydrogen atom transfer from an allylic position of the olefin to the metal of another complex. The rate of the reaction decreases with increasing steric shielding of the metal, in the order bpa > pla > bla > Bn-bla (see Figure 11). Involvement of aminyl radicals formed by deprotonation of the NH moiety of the ligand could also play a role in this reaction.37

M

N

N

N2+

RhN

N

N

R2

R1

2+

R1

R2

R3

Ir

N

N

N2+

R1

R2

R3

H

M

N

N

N2+

R1

R2

R3

+R3

H

or

bpa: R1 = R2 = R3 = Hpla: R1 = R3 = H; R2 = Mebla: R1 = R2 = Me; R3 = HBn-bla: R1 = R2 = Me; R3 = Bn

Figure 11. Binuclear radical-type activation of allylic C−H bonds.

Formation of equimolar amounts of cod-allyl and hydride species were also reported for binuclear iridium-2,2-dimethylaziridine-cod radicals (Figure 12).40

Figure 12. Allylic activation of binuclear iridium-2,2-dimethylaziridine-cod species.

Relevant to the above examples, formation of the rhodium allyl complex RhI(cod)(η3-C8H13) by hydrogen atom transfer from the solvent or water to the electrochemically generated [Rh0(cod)2] radical has been postulated, but the exact mechanism of this process is not clear.41

Chapter 1

12

1.3.2 Open-shell reactivity of the carbonyl ligand

One-electron activation of carbon monoxide is a potentially attractive route for formation of new C−C bonds.42 Coordination of CO to rhodium(II) porphyrins (but not Co or Ir)43 results in one electron activation of the carbon atom of the carbonyl group (Figure 13). Similarly to one-electron activation of olefins, the unpaired electron is transferred from the metal into the π*-orbital of the ligand. Hence, the carbonyl ligand is clearly behaving as a redox non-innocent ligand.

Figure 13. Binding of CO to open-shell RhII(por) species to form a reactive formyl radical.

The follow-up reactivity of thus formed formyl radicals depends on the steric bulk of the porphyrinato ligand and the applied CO pressure. In case of reaction of Rh(OEP) (OEP = octaethylporphyrinato) with 1 atm CO a metalloketone (Rh−C(O)−Rh) is formed and at higher CO pressures a dimetal diketone complex (Rh−(O)C−C(O)−Rh).44 Slight increase of the steric bulk allows for selective reversible formation of the diketones for Rh(TXP)45 and Rh(TMP)26 (TXP = tetra-(3,5-dimethylphenyl)porphyrinato; TMP = tetra-(2,4,6-trimethylphenyl)-porphyrinato). For the very bulky Rh(TTiPP) the reductive C−C coupling cannot proceed, thus allowing the EPR detection of the formyl radical.46 Further evidence for ligand centered radical reactivity are the reactions of Rh(TMP)(CO) with the hydrogen atom donor HSn(CH3)3 to form a formyl complex Rh(TMP)(CHO). Reaction of the carbonyl complex with styrene results in formation of a C4 bridged species Rh(TMP)−C(O)−CH2−CH(Ph)−C(O)−Rh(TMP) (analogous to the one-electron activation of olefins, as described in section 3.1).26 Tethered dirhodium porphyrin complexes have more favorable thermodynamics for reductive coupling of CO and readily form the 1,2-ethanodione complex (Figure 14).29 Moreover the formyl radicals react with H2O, C2H5OH and H2. In case of reaction with H2, formyl groups on both rhodium centres are formed by reaction of the carbonyl radical with hydrogen atoms from homolytic cleavage of the dihydrogen molecules. When water or ethanol are the substrates the hydrogen atom of the ROH group is transferred to the formyl radical to form a formyl ligand at one of the metal centres whereas the second formyl radical couples with the thus generated RO• radical to form a formyl


13

ester (Rh−C(O)−OEt) or formyl acid (Rh−C(O)−OH). The formyl acid easily collapses to form the rhodium hydride with extrusion of CO2.

RhIICO

RhIII

OOXORhII O

RhIIIO

O

X

2COH-Y

RhIIIOXORhIII

H OH

Y = OH

RhIIIOXORhIII

OHEtO O

RhIIIOXORhIII

OHH O

Y = H

+ CO2

Y = OEt

Figure 14. Reactivity of tethered RhII(por) complexes with CO.

Ph2P PPh2

(OC)2(Cp)W W(Cp)(CO)2

COH

Ph2P PPh2

(OC)2(Cp)W W(Cp)(CO)2

CO

+ +

H

Ph2P PPh2

OC(Cp)Mo Mo(Cp)CO

+

OR

R=MeR=Ph

HPh2P PPh2

OC(Cp)Mo Mo(Cp)CO

+

R OH

Ph2P PPh2

OC(Cp)Mo Mo(Cp)CO

OR

-HPh2P PPh2

OC(Cp)Mo Mo(Cp)CO

R OH

+e-eH+

Figure 15. Oxygen-centered radical reactivity of bridging carbonyls.

One-electron activation of carbonyl ligands can also result in reactivity on its oxygen atom instead of its carbon in case of carbonyl bridged binuclear complexes (Figure 15). Radical binuclear tungsten complexes with a bridging carbonyl were

Chapter 1

14

found to react with water to form bridging hydroxyl-carbene species (Figure 15). The proposed mechanism involves hydrogen atom transfer from water to the oxygen atom of the bridging CO ligand, although the radical complex did not react with hydrogen atom donors such as R3SnH or diphenylsilane.47 A related hydrogen atom transfer reaction was also reported for the acyl dinuclear molybdenum complex.48 The resulting hydroxy complex can extrude a hydrogen atom when reduced by one electron (Figure 15). 1.3.3 Open-shell reactivity of carbene ligands

One-electron reduction of group 6 transition metal Fischer-type carbene complexes with external reducing agents has been reviewed recently by Sierra et al.49 The main conclusion of that paper is that Fischer-type carbene ligands are redox active, and can be easily reduced by external reducing agents. Although not mentioned as such in this review, it also implies that Fischer-type carbene ligands are potentially ‘redox non-innocent’, meaning that the metal to which they coordinate may function as an internal reducing agent to generate carbene radicals. Such ‘redox non-innocent’ behavior of Fischer-type carbenes has only very recently been disclosed in a couple of papers, and is so far restricted to the open-shell organometallic chemistry of oxidation state +II group 9 transition metals. Carbene formation at low-spin RhII and CoII species is clearly associated with radical reactivity. In this section we will shortly summarize some of the most illustrative examples dealing with the reactivity of one-electron reduced group 6 Fischer-type carbenes obtained by employing external reducing agents (for a more detailed overview we refer to the review of Sierra49), after which we focus on the typical ‘redox non-innocent’ behavior and reactivity of Fischer-type carbenes coordinated to (low-spin) RhII and CoII complexes. 1.3.3.1 Reactivity of group 6 transition metal ‘carbene radical’ complexes

The LUMO of Fischer-type carbene complexes is carbene carbon centered, which gives the ligands their typical electrophilic character. This property also makes it possible to form carbon-centered ‘carbene radicals’ upon one-electron reduction of these complexes.50 The redox activity of Fischer-type carbene ligands was first demonstrated by Casey et al., who showed that one-electron reduction of group 6 pentacarbonyl alkoxyaryl complexes with Na/K alloy leads to formation of carbon centered radical anions (Figure 16), which are persistent in solution at low temperatures (−50 °C and below), as evidenced by EPR spectrosopy.51

Figure 16. Formation of a carbon centered radical by one electron reduction of a group 6 transition metal carbene complexes (M= Cr, Mo, W).

These carbon-centered radicals could be employed in carbon-carbon bond forming reactions. Iwasawa and Fuchibe reported that tungsten aryl- or silylcarbene complexes dimerize (Figure 17) or undergo radical addition to electron poor olefins (Figure 18) after one-electron reductions with SmI2.

52 The nature of the para-substituent on the aryl group of the alkoxycarbene influences the outcome of the


15

reaction with ethyl acrylate. Electron-withdrawing groups lead to subsequent protonolysis, whereas electron-donating groups affect loss of MeO−, with formation of a tungsten-carbene species. Similar results were obtained with a tungsten diphenylmethylsilyl methoxycarbene complex.

Figure 17. Formation of trans-stilbene via dimerization of one-electron reduced Fischer-type tungsten carbenes.

(CO)5W

Ar

OMe

(CO)5W

ArMeO

CO2Et

+e+H+

(CO)5W

ArMeO

CO2EtMeO

Ar

CO2Et

CO2Et

(CO)5W

Ar

CO2Et

-MeO

CO2EtAr

H+

Figure 18. Reaction of one-electron reduced Fischer-type carbenes with ethyl acrylate.

The reactivity of one-electron reduced carbenes conjugated with olefinic bonds leads to radical-type ligand-ligand coupling reactions at more remote distances from the metal, presumably for steric reasons. One-electron reduction of chromium or tungsten α,β-unsaturated carbene complexes results in formation of dinuclear bis-carbene species by radical coupling between the γ-carbon atoms, followed by protonation (protic media) or quenching with another electrophile.52,53 Reduction of tungsten dienylcarbene complexes results in radical-type C−C coupling at the terminal olefinic carbon atoms most distant from the metal. Stabilization of the intermediate ligand radical density by the phenyl substituent at this carbon likely also plays a role in directing the reaction towards this remote ligand-coupling (Figure 19).52

Chapter 1

16

Figure 19. Remote reductive ligand-ligand coupling of unsaturated carbene complexes (M = Cr, W, R1 = Me, Ph; R2 = Me, H).

Somewhat related ligand radicals were obtained by deprotonation of Fischer-type carbene complexes bearing β-hydrogen containing substituents, followed by one-electron oxidation of the resulting anionic intermediates. Deprotonation of the chromium(0)-pentacarbonyl-methoxymethylcarbene complex, for example, led to an anionic carbene complex that underwent radical-type C–C coupling after one-electron oxidation (Figure 20).54,55 Related manganese carbene complexes show a similar reactivity.56 However, this type of reactivity is perhaps not characteristic for Fischer-type carbenes, as similar deprotonation/oxidation induced radical-type reactivity is also known for the methyl group of acetylferrocene.55

Figure 20. Deprotonation/oxidation induced radical-type reactivity of β-hydrogen containing carbene complexes. 1.3.3.2 Possible involvement of carbene radicals in RhII and CoII mediated cyclopropanation of olefins

Carbene ligands generated from diazo compounds at open-shell low-spin RhII and CoII porphyrin complexes seem to have a carbon-radical (carbene radical) character, even without one-electron reduction by an external reducing agent. These ‘carbene radical’ intermediates are capable of hydrogen atom abstraction from the reaction medium to form diamagnetic MIII-alkyl complexes, for example. This has been observed for CoII(TPP) (TPP = tetraphenylporphyrinato), which forms the diamagnetic alkyl species CoIII(TPP)(CH2COOEt) upon prolonged exposure to ethyl diazoacetate.57 A similar hydrogen atom abstraction reactivity was disclosed for RhII(TMP) (TMP = tetramesitylporphinato) (Figure 21).58


17

Figure 21. Formation of ‘carbene radicals’ upon reaction of diazo compounds with CoII(por) and RhII (por) complexes with subsequent hydrogen atom transfer to the carbon-centered radical.

Paramagnetic CoII salen59 and in particular CoII porphyrin complexes,57,60 are also effective catalysts for (electron poor) olefin cyclopropanation. The groups of Yamada61 and Zhang60a proposed that the active species may actually be a carbon-centered radical species. Paramagnetic mononuclear RhII cyclopropanation catalysts developed by the group of Bergman may also operate via comparable carbene radicals.62 So far, no direct evidence for the electronic structure of any of these active species has been presented. The intriguing activity of the cobalt porphyrin system, however, suggests a very different electronic structure of the active carbene transfer reagent in case of open-shell group 9 systems as compared to common Fischer-type carbene intermediates commonly observed with other cyclopropanation catalysts. Rationalization of the observed reactivity on the basis of the non-innocent character of the metal-carbene remains however an open issue. 1.4 Outline of the Thesis

As is evident from the above examples, group 9 transition metal (cobalt, rhodium and iridium) complexes with tri- or tetradentate nitrogen ligands (picolylamines and porphyrins) reveal unique redox non-innocent behavior of olefins and carbonyls. More detailed studies on the redox non-innocence of carbenes are however lacking. Considering their possible involvement in important catalytic transformations, we were triggered to investigate further the possible non-innocent behavior of CO, carbenes and olefins in group 9 complexes supported with nitrogen donor ligands. As a result, several aspects of one-electron activation of different ligands by organometallic complexes will be described in the following chapters.

In Chapter 2 we will describe the synthesis and redox properties of rhodium and iridium carbonyls supported by the Me3tpa ligand. Although this ligand allowed for studies of one-electron activation of ethene, we will see that the iridium carbonyl complex reveals selective metal-centered reactivity after one electron oxidation. In the investigated systems the CO moiety behaves as a redox innocent ligand. However, its two electron transformations occur easily with water or methanol for iridium complexes allowing the isolation of subsequent intermediates in the water gas shift reaction, whereas reaction of the rhodium carbonyl with dioxygen leads to selective formation of a carbonate complex.

In Chapter 3 we digress from radical reactivity to show how even tiny electronic changes can have an impact on the structure of rhodium carbonyls with N-donor ligands, and how can we selectively form terminal versus bridging carbonyl complexes.

We will return to the topic of ligand radicals in Chapter 4, in which we investigated the reactivity of open-shell rhodium and iridium complexes bearing an N–H moiety and

Chapter 1

18

suspected the involvement of aminyl radicals. Unexpected one-electron reduction is observed in the presence of bases and the possible reduction pathways are discussed.

Chapter 5 describes the reactivity of the paramagnetic Me3tpa iridium ethene complex with diazo compounds (used as carbene precursors). Here the ligand system proved to be appropriate to study the redox non-innocence of the carbene. Radical C–C and C–H bond forming reactions of the ‘carbene radical’ are presented and the electronic structure of one electron activated carbenes is discussed.

In Chapter 6 we show that the scope of reactions of ‘radical carbenes’ is expandable from Ir to Co and from stoichiometric (Chapter 5) to catalytic reactions. In a combined theoretical and experimental study we pin down the (radical) active species in cobalt porphyrin catalyzed cyclopropanation of alkenes and show the implications of the carbene redox non-innocence on the reactivity with alkenes having different electronic properties.

In Chapter 7 we investigate reversible hydrogen atom transfer between cobalt porphyrins and organic radicals. We show that HAT has pronounced consequences in the activity of Co(por) as mediators in metal-mediated radical polymerization of olefins.

The Thesis ends with a Perspective on possible applications of the ligand non-innocent complexes, and the resulting new (catalytic) reaction pathways disclosed in this Thesis, potentially leading to valuable new products via unprecedented catalytic pathways. 1.5 Notes and References

1 In a way this is a similar challenge as controlling the reactivity of highly reactive carbenes and nitrenes, which likewise requires transition metal control to achieve selectivity.

2 Buckel, W. Angew. Chem. Int. Ed. 2009, 48, 6779-6787. 3 Kaim, W.; Schwederski, B. Coord. Chem. Rev. 2010, 254, 1580-1588. 4 Meunier, B.; de Visser S. P.; Shaik, S. Chem. Rev. 2004, 104, 3947-3980. 5 Baik, M.-H.; Newcomb, M.; Friesner, R. A.; Lippard, S. J. Chem. Rev. 2003, 103, 2385-2419. 6 Stubbe, J.; Nocera, D. G.; Yee, C. S.; Chang, M. C. Y. Chem. Rev. 2003, 103, 2167-2201. 7 (a) Whittaker, M. M.; Whittaker, J. W. Biophys. J. 1993, 64, 762-772. (b) Wang, Y.; DuBois, J. L.;

Hedman, B.; Hodgson, K. O.; Stack, T. D. P. Science 1998, 279, 537-540. (c) Whittaker, J. W. Chem. Rev. 2003, 103, 2347-2363.

8 Whittaker, M. M.; Ballou, D. P.; Whittaker, J. W. Biochemistry 1998, 37, 8426-8436. 9 It is noteworthy that apart from [NiFe], [FeFe] and [Fe] hydrogenases, the family of cobalamines are the

only stable organometallic compounds known in nature. 10 Banerjee, R. Chem. Rev. 2003, 103, 2083-2094. 11 Toraya, T. Chem. Rev. 2003, 103, 2095-2127. 12 The bond dissociation energy of the Co–CH2Ado bond is approximately 30 kcal mol–1. (a) Finke, R. G.;

Hay, B. P. Inorg. Chem. 1984, 23, 3041-3043. (b) Halpern, J.; Kim, S.-H.; Leung, T. W. J. Am. Chem. Soc. 1984, 106, 8317-8319. (c) Hay, B. P.; Finke, R. G. J. Am. Chem. Soc. 1986, 108, 4820-4829.

13 (a) Chaudhuri, P.; Hess, M.; Flörke, U.; Wieghardt, K. Angew. Chem. Int. Ed. 1998, 37, 2217-2220. (b) Königsmann, M.; Donati, N.; Stein, D.; Schönberg, H.; Harmer, J.; Sreekanth, A.; Grützmacher, H. Angew. Chem. Int. Ed. 2007, 46, 3567-3570.

14 (a) Braunecker, W. A.; Matyjaszewski, K. Prog. Polym. Sci. 2007, 32, 93-146. (b) Li, S.; de Bruin, B.; Peng, C.; Fryd, M.; Wayland, B. B. J. Am. Chem. Soc. 2008, 130, 13373-13381. (c) see also Chapter 7 of this Thesis

15 Pintauer, T.; Matyjaszewski, K. Chem. Soc. Rev. 2008, 37, 1087-1097 16 (a) It should be noted that the cobalt catalyzed synthesis of terephtalic acid, which proceeds via a radical

pathway, is the largest scale operating industrial process based on homogeneous catalysis: van Leeuwen, P. W. N. M. Homogeneous Catalysis: Understanding the Art, Kluwer Academic Publishers: Dordrecht 2004. (b) ed. Bäckvall, J.-E. Modern oxidation methods, Wiley-VCH: Weinheim 2004.

17 Jørgensen, C. K. Oxidation numbers and oxidation states, Spinger: Berlin 1969.


19

18 Ligand non-innocence of a spectator ligand can be also beneficial. Such a ligand can serve as an electron reservoir allowing inter alia 2e reactions on a first row transition metal centre. For an overview see: (a) Chirik, P. J.; Wieghardt, K. Science 2010, 327, 794-795. (b) Dzik, W. I.; van der Vlugt, J. I.; Reek, J. N. H.; de Bruin, B. Angew. Chem. Int. Ed. in press.

19 Note that assigning ‘real’ (spectroscopic) oxidation states is a matter of deciding to which of the two extreme ionic formalisms the complex fits best; the ‘formal oxidation state’ extreme or the ‘reduced/oxidized ligand’ extreme. The real complex is almost always on a gliding scale somewhere in- between a purely covalent metal-ligand interaction and one of the two ionic extremes, as shown in Figure 4. In most cases the electronic structure of an S = ½ complex can be described reasonably well in terms of ‘being best described as’ a metal or a ligand radical (Figure 3a, Figure 3c) on the basis of spectroscopic measurements, X-ray diffraction and/or DFT calculations. Only in case of (nearly) purely covalent M−L bonding situations the electronic structure of a complex cannot be assigned to either one of these extremes, as it holds the exact middle between them (Figure 3b).

20 For reviews on the subject see: (a) de Bruin, B.; Hetterscheid, D. G. H.; Koekoek, A. J. J.; Grützmacher, H. Progr. Inorg. Chem. 2007, 55, 247-354. (b) Torraca, K. E.; McElwee-White, L. Coord. Chem. Rev. 2000, 206-207, 469-491. (c) Connelly, N. G. Chem. Soc. Rev. 1989, 18, 153-185. (d) Ustynyuk, N. A.; Gusev, O. V.; Novikova, L. N.; Peterleitner, M. G.; Denisovich, L. I.; Peganova, T. A.; Semeikin, O. V.; Valyaev, D. A. J. Solid State Electrochem. 2007, 11, 1621-1634.

21 de Bruin, B.; Hetterscheid, D.G. H. Eur. J. Inorg. Chem. 2007, 211-230. 22 Ogoshi, H.; Setsune, J.; Yoshida, Z. J. Am. Chem. Soc. 1977, 99, 3869-3870. 23 Paonessa, R. S.; Thomas, N. C.; Halpern, J. J. Am. Chem. Soc. 1985, 107, 4333-4334. 24 Wayland, B. B.; Feng, Y.; Ba, S. Organometallics 1989, 8, 1438-1441. 25 Del Rossi, K. J.; Wayland, B. B. J. Chem. Soc., Chem. Commun. 1986, 1653-1655. 26 Wayland, B. B.; Sherry, A. E.; Poszmik, G.; Bunn, A. G. J. Am. Chem. Soc. 1992, 114, 1673-1681. 27 Zhai, H.; Bunn, A.; Wayland, B. Chem. Commun. 2001, 1294-1295. 28 Bunn, A. G.; Wayland, B. B. J. Am. Chem. Soc. 1992, 114, 6917-6919. 29 Zhang, X. X.; Parks, G. F.; Wayland, B. B. J. Am. Chem. Soc. 1997, 119, 7938-7944. 30 Yeung, S. K.; Chan, K. S. Organometallics 2005, 24, 6426-6430. 31 Connelly, N. G.; Kelly, R. L.; Kitchen, M. D.; Mills, R. M.; Stansfield, R. F. D.; Whiteley, M. W.;

Whiting, S. M.; Woodward, P. J. Chem. Soc. Dalton Trans. 1981, 1317-1326. 32 Connelly, N. G.; Graham, P. G.; Sheridan, J. B. J. Chem. Soc. Dalton Trans. 1986, 1619-1621. 33 Aggarwal, R. P.; Connelly, N. G.; Dunne, B. J.; Gilbert, M.; Orpen, A. G. J. Chem. Soc. Dalton Trans.

1991, 1-9. 34 Brammer, L.; Connelly, N. G.; Edwin, J.; Geiger, W. E.; Orpen, A. G.; Sheridan, J. B. Organometallics

1988, 7, 1259-1265. 35 a) de Bruin, B.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Wilting, J. B. M.; de Gelder, R.; Smits, J.

M. M.; Gal, A. W. Angew. Chem. Int. Ed. 2002, 41, 2135-2138. (b) Hetterscheid, D. G. H.; Bens, M.; de Bruin, B. Dalton Trans. 2005, 979-984.

36 (a) de Bruin, B.; Thewissen, S.; Yuen, T. W.; Peters, T. P. J.; Smits, J. M. M.; Gal, A. W. Organometallics 2002, 21, 4312-4314. (b) Hetterscheid, D. G. H.; Kaiser, J.; Reijerse, E.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Smits, J. M. M.; de Gelder, R.; de Bruin, B. J. Am. Chem. Soc. 2005, 127, 1895-1905.

37 Hetterscheid, D. G. H.; Klop, M.; Kicken, R. J. N. A. M.; Smits, J. M. M.; Reijerse, E. J.; de Bruin, B. Chem. Eur. J. 2007, 13, 3386-3405.

38 Burrows, A. D.; Green, M.; Jeffery, J. C.; Lynam, J. M.; Mahon, M. F. Angew. Chem. Int. Ed. 1999, 38, 3043-3045.

39 Hetterscheid, D. G. H.; de Bruin, B.; Smits, J. M. M.; Gal, A. W. Organometallics 2003, 22, 3022-3024. 40 Tejel, C.; Ciriano, M. A.; Passarelli, V.; López, J. A.; de Bruin, B. Chem. Eur. J. 2008, 14, 10985-10998. 41 Orsini, J.; Geiger, W. E. J. Electroanal. Chem. 1995, 380, 83-90. 42 Wayland, B.; Fu, X. Science 2006, 311, 790-791. 43 Cui, W.; Li, S.; Wayland, B. B. J. Organomet. Chem. 2007, 692, 3198-3206. 44 Coffin, V. L.; Brennen, W.; Wayland, B. B. J. Am. Chem. Soc. 1988, 110, 6063-6069. 45 Wayland, B. B.; Sherry, A. E.; Coffin, V. L. J. Chem. Soc., Chem. Commun. 1989, 662-663. 46 Basickes, L.; Bunn, A. G.; Wayland, B. B. Can. J. Chem. 2001, 79, 854-856. 47 Alvarez, M. A.; Anaya, Y.; García, M. E.; Riera, V.; Ruiz, M. A.; Vaissermann, J. Organometallics 2003,

22, 456-463. 48 Alvarez, M. A.; Anaya, Y.; García, M. E.; Riera, V.; Ruiz, M. A.; Vaissermann, J. Organometallics 2005,

24, 2452-2465. 49 Sierra, M. A.; Gómez-Gallego, M.; Martínez-Álvarez, R. Chem. Eur. J. 2007, 13, 736-744.

Chapter 1

20

50 Block, T. F.; Fenske, R. F.; Casey, C. P. J. Am. Chem. Soc. 1976, 98, 441-443. 51 Krusic, P. J.; Klabunde, U.; Casey, C. P.; Block, T. F. J. Am. Chem. Soc. 1976, 98, 2015-2018. 52 Fuchibe, K.; Iwasawa, N. Chem. Eur. J. 2003, 9, 905-914. 53 Sierra, M. A.; Ramírez-López, P.; Gómez-Gallego, M.; Lejon, T.; Mancheño, M. J. Angew. Chem. Int.

Ed. 2002, 41, 3442-3445. 54 Lattuada, L.; Licandro, E.; Maiorana, S.; Molinari, H.; Papagni, A. Organometallics 1991, 10, 807-812. 55 Geisbauer, A.; Mihan, S.; Beck, W. J. Organomet. Chem. 1995, 501, 61-66. 56 Rabier, A.; Lugan, N.; Mathieu, R.; Geoffroy, G. L. Organometallics 1994, 13, 4676-4678. 57 Penoni, A.; Wanke, R.; Tollari, S.; Gallo, E.; Musella, D.; Ragaini, F.; Demartin, F.; Cenini, S. Eur. J.

Inorg. Chem. 2003, 1452-1460. 58 Zhang, L.; Chan, K. S. Organometallics 2007, 26, 679-684. 59 (a) Niimi, T.; Uchida, T.; Irie, R.; Katsuki, T. Tetrahedron Lett. 2000, 41, 3647-3651. (b) Niimi, T.;

Uchida, T.; Irie, R.; Katsuki, T. Adv. Synth. Catal. 2001, 33, 79-88. 60 (a) Huang, L.; Chen, Y.; Gao, G.-Y.; Zhang, X. P. J. Org. Chem. 2003, 68, 8179-8184. (b) Chen, Y.;

Fields, K. B.; Zhang, X. P. J. Am. Chem. Soc. 2004, 126, 14718-14719. (c) Chen, Y.; Zhang, X. P. J. Org. Chem. 2007, 72, 5931-5934. (d) Chen, Y.; Ruppel, J. V.; Zhang, X. P. J. Am. Chem. Soc. 2007, 129, 12074-12075. (e) Zhu, S.; Ruppel, J. V.; Lu, H.; Wojtas, L.; Zhang, X. P. J. Am. Chem. Soc. 2008, 130, 5042-5043. (f) Zhu, S.; Perman, J. A.; Zhang, X. P. Angew. Chem. Int. Ed. 2008, 47, 8460-8463 (g) Ruppel, J. V.; Gauthier, T. J.; Snyder, N. L.; Perman, J. A.; Zhang, X. P. Org. Lett. 2009, 11, 2273-2276.

61 Based on IR and DFT investigation Co(salen) carbene complexes were proposed to have radical character of the carbene moiety: Ikeno, T.; Iwakura, I.; Yamada, T. J. Am. Chem. Soc. 2002, 124, 15152-15153.

62 Krumper, J. R.; Gerisch, M.; Suh, J. M.; Bergman, R. G.; Tilley, T. D. J. Org. Chem. 2003, 68, 9705-9710.

Chapter 2

Activation of Carbon Monoxide by (Me3tpa)Rh and (Me3tpa)Ir*

N

NN

NIr

H

O OR

CO

N

N

N

NIr

CO

CO

N

N

N

NIr

HH

CO

O

N

N N

NRh

CO

N

N N

NRh

OO

O

N

N NN Ir

CO

H

4+ R = Me5+ R = H

+

III

+

I

+

III

2+ 6+

H-OR

Δ, R = Me

- O=C=

R = H

+

I

3+

+

III

8+

O2

III

2+

72+

Ag+

* Part of this work has been published: Reproduced in part with permission from [Dzik, W. I.; Smits, J. M. M.; Reek, J. N. H.; de Bruin, B. Organometallics 2009, 28, 1631-1643] Copyright [2009] American Chemical Society

Chapter 2

22

2.1 Introduction

Pyridyl-amine type of ligands1 are widely applied in coordination chemistry. Many synthetic models for active sites of metalloenzymes are based on these ligands, such as manganese catalases,2 heme-copper oxidases,3 Cu-based4 or non-heme Fe5 oxidases and oxygenases. Pyridyl-amine type ligands were also valuable in Ru catalyzed oxygenation of alkanes.6

Mentpa (tpa = tris(2-piridylmethyl)amine, n = 0-3) ligand supported olefin complexes of rhodium and iridium were extensively studied over the past few years and proved to be versatile synthetic models to study activation of ethene and other olefins towards oxygenation.7 Depending on the number of methyl groups (n) of the Mentpa ligand, various reaction pathways have been uncovered. Stable 2-metallaoxetanes obtained by oxygenation of MI(ethene) fragments showed also noteworthy follow-up reactivity (Scheme 1).7b,8

M

+

M

+

O

Rh

2+

HNO

Rh

2+

ONH

M

+

OO

Ir+

OO

Rh

+

Rh2+

NH4+

MeI

OMeI

H+

Rh+Cl−

OHCl

Δ

Rh

+

HCl

Cl

Ir

2+

O2

MeCN

-e-

O2

O2O2

RhO

O

+

OH

Scheme 1. Reactivity of [MI(Mentpa)(C2H4)]+ species.

Interestingly, the [(Mentpa)IrI(ethene)]+ (n = 2, 3) complexes could be easily oxidized by one electron to form the first reported stable paramagnetic IrII-ethene complexes. The [(Me3tpa)IrII(ethene)]2+ species (Me3tpa = N,N,N-tri(6-methyl-2-pyridylmethyl)amine) is a stable metal-centered radical, but coordination of a fifth σ-donor ligand (i.e. MeCN) triggers radical-type reactivity of the ethene carbon atoms. This was illustrated by formation of [IrIII(CH2CHO)(Me3tpa)(MeCN)]2+ in reaction with dioxygen;9 [(Me3tpa)(MeCN)IrIII–CH2CH2–IrIII(Me3tpa)(MeCN)]4+ dinuclear ethylene-bridged species in neat MeCN;10 or a C3 bridged dinuclear species in the presence of ethyl diazoacetate.11

The abovementioned remarkable reactivity of both diamagnetic and paramagnetic (Mentpa)M-olefin (M = Rh, Ir) complexes triggered us to explore the reactivity of the analogous carbonyl complexes. Herein we report the synthesis of (Me3tpa)M-carbonyl complexes (M = Rh, Ir), their dynamic behavior in solution and their reactivity towards nucleophiles and dioxygen as well as their redox properties.

Carbon monoxide is an important substrate in many industrial processes like hydroformylation, synthesis of methanol, acetic acid (Monsanto) and hydrocarbons

Activation of Carbon Monoxide by (Me3tpa)Rh and (Me3tpa)Ir

23

(Fischer-Tropsch), or production of hydrogen in the water gas shift reaction (WGSR).12 These reactions involve 2 electron transformations, and so far the redox non-innocent behavior of the carbonyl ligand is restricted to only several examples (see Chapter 1). The ability of Me3tpa ligand to stabilize the iridium radical complexes of ethene suggests that this platform could allow us to isolate carbonyl radical complexes of rhodium or iridium in order to study the redox non-innocent behavior of the carbonyl ligand.

In this Chapter we describe the results of our investigations using Me3tpa carbonyl complexes of rhodium and iridium wherein we expected to find one-electron activation pathways of the CO ligand. The outcome of the study revealed, however, that two-electron activation of the metal carbonyls proved easier for these systems. 2.2 Results and discussion

2.2.1 Synthesis and characterization of [(Me3tpa)Ir(cod)]PF6

For synthetic reasons we decided to first prepare a diolefin complex and exchange the olefinic ligand for CO. [(κ3-Me3tpa)Ir(η4-cod)]PF6 ([1]PF6) – the precursor of the iridium carbonyl complex – was synthesized by reaction of [Ir(η4-cod)(μ-Cl)]2 (cod = cis,cis-1,5-cyclooctadiene) with Me3tpa (Me3tpa = N,N,N-tris(6-methyl-2-pyridylmethyl)-amine) in methanol followed by exchange of the chloride with a hexafluorophosphate anion (Scheme 2).

N

N

N

NIrCl

N

N

NN

+

I

1+

Ir +KPF6

- KCl

Me3tpa

Scheme 2. Synthetic route toward [(κ3-Me3tpa)Ir(η4-cod)]+ (1+).

Crystals of [1]PF6, suitable for X-ray diffraction, were obtained by cooling of a saturated methanolic solution of the complex. The structure of 1+ is shown in Figure 1. The X-ray structure reveals a κ3-coordination mode of the Me3tpa ligand. The coordinated lutidyl fragments are not equivalent with significantly different Ir-Npy and C=C bond lengths. The τ parameter13 value of 0.5 indicates a structure that is exactly in between an idealized square pyramid and a trigonal bipyramid. This is partly caused by the geometrical constraints imposed by the restricted ligand bite-angles. Therefore, the geometry of 1+ is perhaps best described as a distorted trigonal bipyramid (tbpy), where lutidyl N2 is bound apically and lutidyl N1 and aminyl N4 equatorially. Stronger binding of olefins in the equatorial plane – typical for tbpy d8 metal complexes14 – results in a significantly longer C5–C6 bond (1.433(6) Å) compared to the C1–C2 bond (1.398(5) Å) in the apical position. The elongated C5–C6 distance indicates a substantial iridia(III)cyclopropane character of this bond caused by stronger π-back donation in the equatorial plane. Similar coordination modes were reported for other bis-(picolyl/lutidyl) amine ligated Rh and Ir cyclooctadiene complexes.15

Chapter 2

24

Figure 1. X-ray structure of [Ir(κ3-Me3tpa)(η4-cod)]+ (1+). Thermal ellipsoids are drawn with 50% probability

(hydrogen atoms and the PF6− counterion are omitted for clarity). Selected bond lengths [Å] and angles [°]:

Ir1–N1 2.257(2), Ir1–N2 2.120(2), Ir1–N4 2.337(2), Ir1–C1 2.162(3), Ir1–C2 2.140(3), Ir1–C5 2.075(3), Ir1–C6 2.092(3), C1–C2 1.398(5), C5–C6 1.433(6), N1–Ir1–N2 91.11(9), N1–Ir1–N4 74.98(9), N2–Ir1–N4 74.11(9).

The Me3tpa and cod ligands are fluxional in solution, causing very broad signals in the 1H NMR spectrum recorded at 20 °C. The broadening is caused by rapid exchange of the coordinated and non-coordinated lutidyl fragments. Upon cooling, the signals become sharp and correspond to a complex in which two of the lutidyl ligands are coordinated to the metal and the third remains unbound (Figure 2). The two coordinated lutidyl fragments are magnetically equivalent. Thus, in solution these fragments most likely rapidly exchange axial and equatorial positions, assuming that the lowest energy solution structure of 1+ corresponds to the tbpy geometry found in the X-ray structure (Figure 1).

ppm (t1)2.03.04.05.06.07.0

Figure 2. 1H NMR spectrum of [Ir(κ3-Me3tpa)(η4-cod)]+ (1+) recorded in CD3CN at 20 °C (top) showing a rapid exchange of the three lutidyl fragments at room temperature. The process is frozen at −30 °C (bottom) resulting in a spectrum showing a κ3 coordination mode of the ligand.


25

The barrier for cod rotation is generally low in solution, making the interchange between square pyramidal and tbpy coordination modes very fast. This likely causes the observed averaging of the signals of the coordinated lutidine fragments. 2.2.2 Synthesis and characterization of (Me3tpa)MI carbonyl complexes

Bis-carbonyl iridium complex [(κ3-Me3tpa)Ir(CO)2]PF6 ([2]PF6) was obtained by bubbling CO through a solution of cod precursor [1]PF6 in CH2Cl2 (Scheme 3).

N

NN

NIr

N

NN

NIr

CO

CO

+

I

CO

1+

+

I - cod

2+

Scheme 3. Synthesis of [(κ3-Me3tpa)Ir(CO)2]+ (2+).

Crystals suitable for X-ray diffraction were obtained by layering a solution of [2]PF6 in CH2Cl2 with hexanes.

Figure 3. X-ray structure of [Ir(κ3-Me3tpa)(CO)2]+ (2+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms and the PF6

− counterion are omitted for clarity). Selected bond lengths [Å] and angles [°]: Ir1–C1 1.823(3), Ir1–C2 1.846(3), Ir1–N1 2.147(2), Ir1–N3 2.200(2), Ir1–N4 2.364(2), C1–O1 1.147(3), C2–O2 1.136(4), C1–Ir1–C2 87.00(13), C1–Ir1–N1 91.60(11), C2–Ir1–N1 175.33(12), C1–Ir1–N3 160.42(12), C2–Ir1–N3 91.18(12), N1–Ir1–N3 88.65(9), C1–Ir1–N4 123.95(11), C2–Ir1–N4 109.14(11), N1–Ir1–N4 75.32(9), N3–Ir1–N4 74.98(8).

The structure of 2+ (Figure 3) is similar to that of 1+, but its τ value of 0.25 suggests a structure closer to a square pyramid. As expected for d8 sqpy complexes,14 the apical position is occupied by the weaker donating amine donor, while the π-accepting CO,

Chapter 2

26

and stronger σ-donating lutidyl groups are coordinated in the basal plane. The third lutidyl group is non-coordinating in the solid state, but in solution the molecule shows rapid exchange of the coordinated and non-coordinated lutidyl fragments on the NMR time scale. This results in a very simple pattern of the 1H NMR spectrum at room temperature revealing an averaged signal for all arms of the ligand (Figure 4). In contrast to 1+, the fluxionality in 2+ is so fast that separate peaks could not be observed even at −94 ºC (the freezing point of acetone). We interpret this fluxionality as a fast exchange of the coordinated and non-coordinated lutidyl fragments proceeding via a four-coordinated intermediate. The IR spectrum in dichloromethane shows only two bands at 2058 and 1975 cm−1 and thus does not reveal the presence of the proposed four-coordinated species in detectable amounts.

ppm (f1)2.03.04.05.06.07.08.0

Figure 4. 1H NMR spectrum of [Ir(κ3-Me3tpa)(CO)2]+ (2+) recorded in acetone-d6 at 20 °C showing averaged signals indicating a rapid exchange of the three lutidyl fragments.

Compared to ethene, the affinity of carbon monoxide towards iridium is much higher, since the dangling lutidiyl group is not capable of substituting the CO ligand to form a monocarbonyl complex (as was the case for [(κ3-Me3tpa)Ir(ethene)2]

+ where one of the ethene molecules is readily displaced by the dangling lutidyl fragment).7c

In marked contrast to iridium, similar synthetic procedures for rhodium led to a mono-carbonyl Me3tpa complex. Easier displacement of a CO ligand for Rh compared to Ir can be rationalized by the lower π-basicity of rhodium. Complex [Rh(κ3-Me3tpa)(CO)]PF6 ([3]PF6) was obtained in a one-pot reaction of [Rh(coe)2(μ-Cl)]2 (coe = cis-cyclooctene) with CO, Me3tpa and NH4PF6 in methanol (Scheme 4). Crystals suitable for X-ray diffraction were obtained by layering an acetone solution of [3]PF6 with methanol at −20 °C. The structure of 3+ is shown in Figure 5.

The tetradentate Me3tpa ligand coordinates in a κ3 rather than κ4 fashion (as confirmed by IR and NMR spectroscopies, and X-ray diffraction) resulting in a distorted square planar 16 VE complex. Higher tendency to form square planar complexes for rhodium compared to iridium has been reported for related pyridine-amine-pyrrole ligands.16 The distortion from the square planar coordination mode is a result of interactions of the methyl groups of the lutidine fragments with CO, causing a nearly 23° bending of the Rh-CO axis below the coordination plane passing through N1, N2, N4 and Rh (Figure 5).


27

N

N N

NRh

COCl

Rh

N

N

NN

+

I

3+

+CO, KPF6

- coe, - KCl

Me3tpa

Scheme 4. Synthesis of [Rh(κ3-Me3tpa)(CO)]+ (3+).

Figure 5. X-ray structure of [Rh(κ3-Me3tpa)(CO)]+ (3+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms and the PF6

− counterion are omitted for clarity). Selected bond lengths [Å] and angles [°]: Rh1–C1 1.812(4), Rh1–N2 2.056(3), Rh1–N1 2.064(3), Rh1–N4 2.108(3), C1–O1 1.152(5), C1–Rh1–N2 97.81(15), C1–Rh1–N1 99.36(15), N2–Rh1–N1 162.43(12), C1–Rh1–N4 157.39(18), N2–Rh1–N4 80.64(13), N1–Rh1–N4 82.08(12).

The Rh1−N3 distance (2.901(4) Å) is far too long to consider the Rh−N3 interaction as a bond; moreover, the lone pair of N3 does not point towards the metal. Also in solution 3+ remains four-coordinate. Both the IR frequency (1983 cm−1) and 13C NMR shift (190.7 ppm) of the CO ligand have values close to those reported for square planar [Rh(κ3-tpa)(CO)]+ (1991 cm−1 and 190.8 ppm; tpa = N,N,N-tris(2-pyridylmethyl)amine).17 The IR CO stretch frequency of 3+ is lower than for the related [Rh(κ3-tpa)(CO)]+, in spite of the fact that (for steric reasons) the lutidyl groups of Me3tpa are weaker donors than the picolyl groups of tpa. The low stretch frequency is a result of the slight distortion from the square planar geometry toward a sawhorse (butterfly) geometry. This distortion allows for some back bonding from the metal dz2 to a π*CO orbital, and an antibonding interaction between the σCO and dyz orbitals appears (Figure 6), while other bonding and antibonding interactions between the metal and CO are hardly affected.18 In this way the out of plane bending of CO molecule results in increased π-back donation, as compared with the square planar structure, giving a lower than expected IR CO stretch frequency.

Chapter 2

28

Figure 6. Orbital interactions between the metal and the carbonyl ligand in a tetracoordinate carbonyl complex with a sawhorse distortion.

Similar to 1+ and 2+, the coordinated and non-coordinated lutidyl groups exchange rapidly in solution causing very broad signals in the NMR spectrum at room temperature (Figure 7). The spectrum measured at −30 ºC is consistent with the structure obtained by X-ray diffraction: the signals of the protons of the dangling lutidine are shifted upfield compared to those of [Rh(κ4-Me3tpa)(ethene)]+,7c and their chemical shifts values are closer to the ones of the free ligand. The NMR spectroscopy data reveal only the square planar isomer, but solution IR measurements reveal an equilibrium between the κ3 and κ4 isomers19 (a broad peak at 1983 cm−1 with a small shoulder at 1966 cm−1; Figure 8), albeit in favor of the κ3 coordination.

ppm (t1)2.03.04.05.06.07.08.0

Figure 7. 1H NMR spectrum of [Rh(κ3-Me3tpa)(CO)]+ (3+) recorded in acetone-d6 at 20 °C (top) showing a rapid exchange of the three lutidyl fragments at higher temperatures. The process is frozen at −30 °C (bottom).

The different coordination modes for [Rh(κ3-Me3tpa)(CO)]+ and [Rh(κ4-

Me3tpa)(ethene)]+ complexes are remarkable. At first sight one would expect that CO, being a stronger π-acid and weaker σ-donor than ethene20 would cause rhodium to be less electron rich, thus making the Rh–N interactions stronger. However, in this case extensive π-back donation to the olefin apparently results in a substantial contribution of the metalla-cyclopropane resonance structure with the rhodium atom adopting a formal +III oxidation state.21 This resonance structure is stabilized by the κ4 coordination mode of Me3tpa leading to an 18 VE configuration. This situation is not attainable for the CO complex, and thus the +I oxidation state of the rhodium atom in case of the carbonyl complex leads to the preferred κ3 coordination mode of Me3tpa. In general there is a strong preference for N-ligand supported group VIII d8 metals to form square planar rather than penta-coordinate complexes. Hypodenticity22 of tpa type ligands has been


29

reported for d8 PdII,23 and PtII chloride complexes.23b Also those complexes showed fluxional behavior of the ligand.23b

35

45

55

65

75

85

95

170018001900200021002200Wavenumbers [cm-1]

Tra

nsm

itan

ce [

%]

Figure 8. Carbonyl region of the IR spectrum of [Rh(κ3-Me3tpa)(CO)]+ (3+) in dichloromethane. The signal at 1983 cm–1 corresponds to the κ3 isomer and the shoulder at 1966 cm–1 corresponds to the κ4 isomer.

Fluxionality of multidentate N-donor ligands in RhI/IrI metalloorganic complexes has also been observed for bis-carbonyl and diene complexes with tridentate pyrazolyl-type ligands.24,25 NMR and IR spectroscopic studies of those compounds reveal equilibria between square planar and square pyramidal coordination modes, involving a rapid exchange between κ2 to κ3 denticity of the anionic N3-ligand. Theoretical studies on this fluxionality were performed by Hall et al. for a broad range of rhodium complexes.26 Some picolyl-amine type ligands in Rh(cod) complexes were also reported to show high fluxionality.16,27

The fluxionality of the Me3tpa ligand in complexes 1+, 2+ and 3+ may be mainly caused by a fast equilibrium between 16 VE square planar and 18 VE pentacoordinate geometries. We propose that in the case of 1+ and 2+ the exchange of the lutidyl groups proceeds via a dissociative pathway in which dissociation of a coordinated lutidine leads to a 4-coordinate 16 VE [(κ2-Me3tpa)Ir(CO)2]

+ complex, which is followed by coordination of the dangling lutidine to regain a 18 VE κ3 complex with two lutidyl groups being exchanged (Scheme 5, top). In the case of 3+ the process may be associative, proceeding via coordination of the dangling lutidine to form an 18 VE κ4 intermediate, followed by dissociation of another coordinated lutidine to regain a 16 VE κ3 complex (Scheme 5, bottom).

IR spectroscopy reveals the presence of a major κ3 isomer and a minor κ4 isomer in solution at room temperature. With room-temperature NMR spectroscopy fast exchange of all lutidyl fragments is observed, but this process is frozen at −30 ºC. At this temperature, the signals correspond to the major κ3 isomer, whereas no signals of the minor κ4 isomer are detectable with NMR spectroscopy. These data can only be interpreted with the exchange processes depicted in Scheme 5 (bottom): A fast equilibrium between the major κ3 isomer and a minor κ4 isomer through lutidyl association/dissociation (still fast at −30º C on the NMR time scale, but observable with IR spectroscopy) and a much slower process involving the positional exchange of the

Chapter 2

30

coordinated lutidyl groups in the minor κ4 isomer (presumably via Berry pseudorotation), followed by a much faster lutidyl dissociation to re-form the 16 VE square planar complex.

Scheme 5. Dissociative (top) and associative (bottom) pathways proposed for the exchange of the Me3tpa lutidyl groups at the metal centre in complexes 1+, 2+ (top) and 3+ (bottom). Lu = lutidyl = 2,6-dimethylpiridyl. 2.2.3 Reactivity of complex 2+ with H2O and methanol

The electron density of the iridium atom of 2+ (ν(C≡O) = 2058, 1975 cm−1 in CH2Cl2) is comparable to that of the κ3-hydrotris(pyrazolyl)borate biscarbonyl iridium complex (ν(C≡O) = 2051, 1971 cm−1 in MeCN),24 which was reported to react with water and alcohols to form hydroxy- and alkoxy-carbonyl species.28 The similar electronic properties of 2+ compared to [κ3-TpIr(CO)2]

24 encouraged us to investigate its reactivity with water and methanol. Indeed, 2+ reacts with both methanol and water in acetone to give the hydrido-methoxycarbonyl and hydrido-hydroxycarbonyl complexes 4+ and 5+ respectively (Scheme 6).

N

NN

NIr

H

O OR

CO

N

N

N

NIr

CO

CO

+

III

+

I

H-OR

2+4+ R = Me5+ R = H

Scheme 6. Reaction of 2+ with methanol and water to hydrido-methoxycarbonyl and hydrido-alkoxycarbonyl species 4+ and 5+.

Crystals of [4]PF6, suitable for X-ray diffraction, were grown overnight from a concentrated solution of [2]PF6 in methanol. The X-ray structure of 4+ is shown in


31

Figure 9. The X-ray structure of the methoxycarbonyl complex 4+ reveals coordination of the hydride trans to the amine nitrogen. One of the lutidyl fragments remains noncoordinating and the two coordinated lutidyl groups bind trans to the carbonyl and methoxycarbonyl ligands respectively. The Ir–N3 bond, trans to the methoxycarbonyl moiety is substantially longer than the Ir–N2 bond, trans to CO, reflecting the stronger donor capacity of the methoxycarbonyl ligand. Because of the intrinsic instability of the hydroxycarbonyl complex 5+ in solution we were not able to grow crystals suitable for X-ray diffraction. However, judging from very similar 1H NMR chemical shifts of the hydrides (16.75 vs. 16.56 ppm) and CO stretching frequencies (2062 vs. 2064 cm–1) of the two compounds, we expect that 5+ has a structure similar to 4+.

Figure 9. X-ray structure of [Ir(κ3-Me3tpa)(CO)(H)(COOMe)]+ (4+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms and the PF6

− counterion are omitted for clarity). Selected bond lengths [Å] and angles [°]: Ir1–C1 1.843(4), Ir1–C2 2.027(4), Ir1–N1 2.130(3), Ir1–N3 2.169(3), Ir1–N4 2.200(3), C1–O1 1.144(5), C2–O2 1.193(5), C2–O3 1.363(5), C1–Ir1–C2 90.51(17), C1–Ir1–N1 177.81(14), C2–Ir1–N1 87.64(15), C1–Ir1–N3 93.41(15), C2–Ir1–N3 175.57(14), N1–Ir1–N3 88.48(12), C1–Ir1–N4 104.19(16), C2–Ir1–N4 96.95(14), N1–Ir1–N4 77.23(13), N3–Ir1–N4 80.09(12). The hydrogen atom bound to the iridium atom could not be located reliably in the difference Fourier map due to residual electron density surrounding the heavy atom.

Formation of alkoxycarbonyl iridium complexes from corresponding alcohols is usually assisted by a strong base that can deprotonate the alcohol, allowing subsequent nucleophilic attack of the alkoxide on the coordinated CO.29 However, introduction of strongly donating ligands can make the iridium centre sufficiently nucleophilic to serve as the proton acceptor.30 Oro and co-workers reported such reactivity for iridium tris(pyrazol-1-yl) complexes28,31 and a study on the reaction of the hydrotris(pyrazolyl)borate biscarbonyl iridium complex with water was performed by the group of Haynes.24 There is no consensus on the mechanism of CO activation towards water and alcohols by the neutral [TpIrI(CO)2]. Two mechanisms have been proposed: (I) direct nucleophilic attack of the ROH on the coordinated electron deficient CO followed by protonation of the metal centre,28 or (II) protonation of the metal to form cationic IrIII−H species followed by nucleophilic attack of the RO– group on CO24

Chapter 2

32

(Scheme 7). The cationic nature of 2+ makes it rather unlikely that protonation of the metal would be the first step in the sequence leading to 4+ and 5+, but protonation of the dangling lutidine remains a possibility for initial activation of ROH (pathway III in Scheme 7). Pathway I remains also a plausible mechanism for formation of 4+ and 5+. A concerted nucleophilic attack with simultaneous proton transfer to the metal could be a further alternative.

Ir

NN

N

CO

CO

N

H OR

Ir

NN

N

CO

C

N

H+

+

OR

Ir

NN

N

CO

CO

N H

OR

Ir

NN

N

CO

C

NH

Ir

NN

N

CO

C

N+

OR

+

OR

-H+

+H+

O

O O

Ir

NN

N

CO

CO

N

H OR

Ir

NN

N

CO

CO

N

Ir

NN

N

CO

C

N

H

+ +

OR

OH

2+

OR-

I

II

III

Scheme 7. Possible mechanisms of addition of water and methanol to 2+. R = H, Me.

The addition of methanol is reversible and heating compound 4+ to 90 oC in DMSO for 6 hours quantitatively regenerates 2+ via a reductive elimination process. This behavior is somewhat similar to thermolysis of [Cp*(PMe3)Ir(H)(C(O)CF3)] where CF3H and iridium carbonyl is formed.32

The hydroxycarboxyl iridium complex 5+ is unstable in solution and decomposes in less than one day to bis-hydride species 6+ and CO2 at room temperature (Scheme 8). The subsequent reactions of complex 2+ with water thus represent a synthetic model of one turnover in the water gas shift reaction (WGSR).33 Crystals of [6]PF6, suitable for X-ray diffraction, were grown overnight by layering an acetone solution of [6]PF6 with diethyl ether. The X-ray structure of 4+ is shown in Figure 10.

N

NN

NIr

H

O OH

CO

N

NN

NIr

CO

CO

N

NN

NIr

HH

CO

+

III

+

I

H2O/Acetone Acetone

+

III

2+ 5+ 6+

- CO2

Scheme 8. Reaction of 2+ with water to form bis-hydride species 6+ via a hydrido-hydroxycarbonyl species 5+.


33

The thus obtained complex 6+ is a classical dihydride with C1 symmetry. Two distinct hydride signals are observed in the 1H NMR spectrum, resonating at –18.14 and –18.25 ppm as an AB pattern with a coupling constant of 8.4 Hz. The X-ray structure of 6+ is very similar to 4+. The Ir–N1 (trans to CO) and Ir–N4 (trans to a hydride) bond distances of 4+ and 6+ are equal. Ir–N3 appears slightly longer in 6+, indicating a stronger trans influence of the hydride, but this difference is not significant within the 3σ criterion.

Structural similarities between 4+, 5+ and 6+ – those compounds differ only by the ligand X coordinated trans to one of the lutidyl groups – allow us to directly compare the donor capacities of the MeOOC–, HOOC– and H– ligands by measurement of the C≡O stretch frequency (Table 1). The π-back donation from the metal to the carbonyl ligand decreases in the order –H > –COOMe > –COOH as can be derived from increasing CO stretch frequencies. The hydride is clearly a stronger σ−donor ligand than the hydroxycarbonyl and methoxycarbonyl ligands. The difference between two latter ones is almost negligible, although not unexpectedly the methoxycarbonyl has slightly better donor properties.

Figure 10. X-ray structure of [Ir(κ3-Me3tpa)(CO)(H)2]+ (6+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms bound to the carbon atoms and the PF6

− counterion are omitted for clarity). Selected bond lengths [Å] and angles [°]: Ir1–C1 1.828(5), Ir1–N1 2.131(4), Ir1–N3 2.181(3), Ir1–N4 2.200(3), C1–O1 1.137(6), C1–Ir1–N1 172.39(16), C1–Ir1–N3 96.35(17), N1–Ir1–N3 90.69(13), C1–Ir1–N4 106.92(17), N1–Ir1–N4 77.17(13), N3–Ir1–N4 79.59(12). Table 1. Comparison of donor abilities of hydroxycarbonyl, methoxycarbonyl and hydride ligands in complexes of the type [Ir(κ3-Me3tpa)(CO)(H)(X)]+.

Complex X ν(C≡O) [cm–1] 5+ COOH 2064 4+ COOMe 2062 6+ H 2031

In contrast to bis-carbonyl iridium complex 2+, the mono-carbonyl rhodium complex

3+ does not react with water or methanol. The π-back donation is much stronger in case of the mono-carbonyl rhodium complex (ν(C≡O) = 1983 for 3+ vs. 2058 cm–1 for 2+ in CH2Cl2), thus making the CO ligand less electrophilic and thereby apparently unreactive toward these weak nucleophiles.

Chapter 2

34

2.2.4 Redox chemistry

The redox properties of 1+, 2+ and 3+ were investigated with cyclic voltammetry (CV) in dichloromethane and acetone (Table 2). The electrochemical behavior of cod complex 1+ is similar to related [Ir(N3-ligand)(cod)]+ complexes,15 and the CV data are indicative of a reversible oxidation of IrI to IrII at 0.20 V vs. Fc/Fc+. The large peak separation (186 mV) in the CV measurements is most probably a result of Ohmic resistance of the electrochemical cell,34 and/or slow electrode kinetics. The anodic peak potential of bis-carbonyl complex 2+ is more positive than that of cod complex 1+, likely because of the higher π-acidity of carbon monoxide compared to olefins, thus making the metal centre less electron rich. Both 2+ and 3+ show irreversible oxidation waves in dichloromethane, but in acetone, oxidation of 3+ is quasi-reversible. Table 2. Electrochemical data for complexes 1+ – 3+ (scan rate 20 mV/sec, potentials vs. Fc/Fc+).

Complex Solvent Ea (V) ΔE (mV)

E1/2 (V)

If/Ib

[Ir(κ3-Me3tpa)(η4-cod)]PF6 ([1]PF6) CH2Cl2 0.29 186 0.20 1.0 CH2Cl2 0.45 - - - [Ir(κ3-Me3tpa)(CO)2]PF6 ([2]PF6) Acetone 0.33 - - - CH2Cl2 0.10 - - - [Rh(κ3-Me3tpa)(CO)]PF6 ([3]PF6) Acetone 0.01 192 -0.05 0.8

Ea = anodic peak potential, ΔE = peak separation E1/2 = half-wave potential, If/Ib = anodic peak current/cathodic peak current.

Since the redox potentials of 1+, 2+ and 3+ are all more positive than Fc+, we decided to use the more potent Ag+ oxidant in further investigations. Compounds 1+, 2+ and 3+ were oxidized in situ with AgPF6 in acetone and investigated by EPR spectroscopy at a temperature of 20 K. Formation of the one-electron oxidized IrII complex 12+ was clearly confirmed by its characteristic X-band EPR spectrum (Figure 11). The EPR spectrum is very similar to that of the related complex [(Bn-dla)IrII(cod)]2+ (Bn-dla = N-benzyl-N,N-bis(6-methyl-2-pirydylmethyl)amine),15 with only slightly different g-values. The resolved hyperfine pattern at g33 = 1.93 stems from hyperfine coupling with Ir (I = 3/2; AIr ~ 120 MHz) and one nitrogen atom (I = 1; AN ~ 60 MHz). Hyperfine couplings are not resolved along g22 = 2.38 and g11 = 2.51. Two low intensity signals along the flanks of the g11 and g22 signals are ΔMI = 2 “forbidden transitions”, which become weakly allowed due to strong coupling with the large iridium quadrupole moment. These features are all identical to those recently reported for [(Bn-dla)IrII(cod)]2+.15

Only a low intensity signal for the product obtained by Ag+ oxidation of 3+ was obtained (Figure 12). Most likely the spectrum corresponds to 32+, but we cannot exclude a different (follow-up) product being responsible for the observed signal. The spectrum reveals the expected rhombic g-tensor (g11 = 2.22, g22 = 2.15 and g33 = 1.99) for a metal centered RhII radical. The g33 signal reveals hyperfine couplings, and the pseudoquartet pattern is indicative of hyperfine coupling with Rh and a single N atom having nearly equal hyperfine coupling constants (~ 70 MHz).

No detectable signal for 22+ could be revealed, even upon rapid freeze-quenching in liquid N2 directly after oxidation of 2+ with Ag+. This is suggestive of a very fast follow up reaction of the oxidized species, in agreement with the irreversible oxidation of 2+ to 22+ in the CV measurements.


35

2400 2600 2800 3000 3200 3400 3600

g

dX

''/d

B

2.8 2.6 2.4 2.2 2 1.8

Figure 11. X-band EPR spectrum of [Ir(Me3tpa)(cod)]2+ (12+) recorded at 20 K in frozen acetone. Frequency = 9.380793 GHz, modulation amplitude = 4 Gauss, attenuation 30 dB. The salt [NBu4N]PF6 (~0.1 M) was added to the solution to obtain a better glass.

2400 2600 2800 3000 3200 3400 3600

g

dX

''/d

B

2.8 2.6 2.4 2.2 2 1.8

*

Figure 12. X-band EPR spectrum of in situ oxidized [Rh(Me3tpa)(CO)]+ 3+ recorded at 20 K in frozen acetone. Frequency = 9.382252 GHz, modulation amplitude = 4 Gauss, attenuation 30 dB. The salt [Bu4N]PF6 (~0.1 M) was added to the solution to obtain a better glass. The sharp signal marked with * is an artifact caused by a paramagnetic impurity present in the cryostat of the EPR spectrometer.

2.2.5 Chemical oxidation of 2+

The main motivation for the preparation of the iridium carbonyl complex 2+ was our interest in the expected different, but unknown radical-type reactivity of the one-electron oxidized species 22+ based on the thoroughly studied [IrII(κ4-Me3tpa)(ethene)](PF6)2.

9,10 It has been shown by Wayland and co-workers that mononuclear RhII porphyrins can form ethylene- (just like the IrII(Me3tpa) system) or

Chapter 2

36

butylene-bridged species by reaction with ethene, and interestingly RhIIIC(O)C(O)RhIII (1,2-ethanodione bridged) species are produced by reductive coupling of two CO molecules between two (por)RhII metallo-radicals.35 Similar reactions for (por)IrII complexes led to intramolecular electron transfer from the porphyrin ligand to Ir, with formation of porphyrin radical cation species (por·+)IrI(CO)2, which prevented radical reactivity of the carbonyl ligand.35d

The analogy between (por)RhII and (Me3tpa)IrII systems prompted us to explore the possibility of reductive coupling of CO by the (Me3tpa)IrII metallo-radical or other, new reactivity of the one-electron activated CO. The advantage of Me3tpa over the porphyrin system would be that the trispyridyl amine ligand allows for subsequent cis-type reactivity pathways. Since we obtained no EPR signal after oxidation of 2+ with Ag+ (vide supra), we wondered what follow-up reaction(s) had caused the transformation of 22+ into (an) EPR-silent species. We further investigated this with NMR spectroscopy and mass spectrometry.

Oxidation of 2+ with 1 eq. of Ag+ did not lead to the expected Ir–CO–CO–Ir bridged species. Instead, quantitative formation of the diamagnetic dicationic iridium(III) mono-carbonyl hydride complex 72+ was observed (Scheme 9).36 This unexpected reaction raises the question of the origin of the hydride. Since the experiment was performed in deuterated solvent, only two sources of hydrogen are available: the Me3tpa ligand and water introduced with the silver salt. High resolution mass spectrometry did not show any substantial enrichment of the ligand in deuterium. Furthermore NMR spectra recorded with an internal standard (toluene) revealed that the oxidation of 2+ produces complex 72+ in >80% yield, along with another, unidentified species. Hence, adventitious water from AgPF6 is the most probable source of the hydrogen atom.37 Addition of few drops of water into the reaction mixture also results in formation of the complex 72+, but with lower selectivity, which makes it difficult to obtain further information about the origin of the H atom.

N

N

N

NIr

CO

CO

N

N NN Ir

CO

H

+

I

+

III

2+

Ag+

Acetone

2+ 72+

H2O (traces)

Scheme 9. Oxidation of 2+ with Ag+ in acetone to form dicationic hydride species 72+, presumably by reaction of an IrII-carbonyl intermediate with Ag+ bound H2O.

Milstein et al.38 recently reported somewhat similar reactivity of RhII(PNP) complexes. The paramagnetic [(PNP)RhII(Cl)]+ reacted with triphenylphosphine in the presence of traces of water to yield a diamagnetic [(PNP)RhIII(H)(Cl)]+ complex and triphenylphosphine oxide. Under the same conditions reaction of the dicationic [(PNP)RhII]2+ with CO gave a [(PNP)RhI(CO)]+ complex and CO2. Other reports on reactivity of mononuclear RhII species with CO showed reduction of the metal to RhI or disproportionation.39 Although we are not aware of any reports on such reactivity of paramagnetic iridium species, we propose that formation of 72+ could have taken place via disproportionation of a short-living paramagnetic complex as depicted in Scheme 10.


37

-CO[IrI] CO

CO

-e[IrII] CO

[IrI] CO

H+

[IrIII] CO

H

-CO2+CO

2+ A2+B+ 72+

[IrIII] CO+H2O

[IrIII]O

OH

B3+ C2+

Scheme 10. Proposed mechanism of formation of dicationic hydride species 72+.

Instantaneous evolution of gas from the reaction mixture suggests that CO is lost upon

one-electron oxidation of 2+ to 22+. Due to a weaker π-back donation, a decreased Ir-CO bond strength is expected for higher valent species. This would lead to formation of the paramagnetic monocarbonyl species A2+, which likely disproportionates to B+ and B3+. The electron deficient carbonyl group of B3+ would readily react with water to form hydroxycarboxyl iridium complex C2+ (somewhat similar to 5+) and a proton. Reaction of this proton with B+ then yields 72+. Complex 72+ is likely also produced by CO2 loss from the hydroxycarbonyl fragment of C2+ (similar formation of 6+ from 5+), followed by capture of a CO molecule (Scheme 10). 2.2.5 Reactivity of 3+ with O2

Although reaction of 3+ with dioxygen in pure acetone yielded a mixture of compounds with only a minor amount of the carbonato complex [Rh(κ4-Me3tpa)(CO3)]PF6 ([8]PF6), we found that addition of some water to the reaction mixture markedly increases selectivity leading to exclusive formation of 8+ (Scheme 11). We are aware of only one claim of a stable rhodium carbonate complex obtained via oxidation of CO to CO3

2- by an arsine-supported rhodium dioxygen complex.40 Such reactivity was also described for some iridium complexes,41 and in all cases an iridium dioxygen carbonyl complex was proposed as the crucial intermediate. Different mechanisms were proposed to explain the subsequent intramolecular rearrangement transforming the peroxo and carbonyl fragments into a carbonato ligand, including outer sphere attack of a coordinated peroxo fragment by external CO. In case of [Ir(C7H4NS2)(O2)(CO)(PPh3)2] (C7H4NS2 = benzothiazole-2-thiolate) Ciriano et al. used 18O labeling studies to prove that two water molecules serve as reactants in rearrangement of the iridium dioxygen carbonyl complex to a carbonate.41d Most likely a similar mechanism is responsible for oxidation of 3+ to 8+, because H2O is crucial for a selective reaction.

N

N N

NRh

CO

N

N N

NRh

OO

O

+

I

3+

Acetone/H2O

+

III

8+

O2

Scheme 11. Reaction of 3+ with dioxygen in acetone to form a rhodium carbonate complex 8+.

Chapter 2

38

Crystals of [8]PF6 suitable for X-ray diffraction were obtained upon crystallization of [3]PF6 from methanol-water solution in –20 °C without taking precautions against air. The structure is shown in Figure 13.

Figure 13. X-ray structure of [Rh(κ4-Me3tpa)(CO3)]+ (8+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms and the PF6

− counterion are omitted for clarity). The crystal had two independent cations in the asymmetric unit. Selected bond lengths [Å] and angles [°]: Rh(1A)–O(2A) 2.0016(18), Rh(1A)–N(4A) 2.022(2), Rh(1A)–N(2A) 2.052(2), Rh(1A)–O(1A) 2.0634(17), Rh(1A)–N(3A) 2.069(2), Rh(1A)–N(1A) 2.096(2), Rh(1A)–C(1A) 2.464(3), C(1A)–O(3A) 1.222(3), C(1A)–O(1A) 1.314(3), C(1A)–O(2A) 1.322(3), O(2A)–Rh(1A)–N(4A) 99.36(8), O(2A)–Rh(1A)–N(2A) 89.37(8), N(4A)–Rh(1A)–N(2A) 80.78(9), O(2A)–Rh(1A)–O(1A) 64.60(7), N(4A)–Rh(1A)–O(1A) 163.58(8), N(2A)–Rh(1A)–O(1A) 94.91(8), O(2A)–Rh(1A)–N(3A) 176.84(8), N(4A)–Rh(1A)–N(3A) 82.91(8), N(2A)–Rh(1A)–N(3A) 93.18(8), O(1A)–Rh(1A)–N(3A) 113.26(8), O(2A)–Rh(1A)–N(1A) 93.53(8), N(4A)–Rh(1A)–N(1A) 84.59(9), N(2A)–Rh(1A)–N(1A) 165.36(9), O(1A)–Rh(1A)–N(1A) 99.30(8), N(3A)–Rh(1A)–N(1A) 84.47(8). Rh(1B)–O(2B) 2.0053(19), Rh(1B)–N(4B) 2.020(2), Rh(1B)–N(1B) 2.052(2), Rh(1B)–O(1B) 2.0622(18), Rh(1B)–N(3B) 2.083(2), Rh(1B)–N(2B) 2.093(2), Rh(1B)–C(1B) 2.469(3), C(1B)–O(3B) 1.224(3), C(1B)–O(1B) 1.318(3), C(1B)–O(2B) 1.327(3), O(2B)–Rh(1B)–N(4B) 99.72(8), O(2B)–Rh(1B)–N(1B) 88.25(8), N(4B)–Rh(1B)–N(1B) 80.48(9), O(2B)–Rh(1B)–O(1B) 64.67(7), N(4B)–Rh(1B)–O(1B) 163.84(9), N(1B)–Rh(1B)–O(1B) 94.24(8), O(2B)–Rh(1B)–N(3B) 176.73(8), N(4B)–Rh(1B)–N(3B) 82.74(9), N(1B)–Rh(1B)–N(3B) 94.30(9), O(1B)–Rh(1B)–N(3B) 113.02(8), O(2B)–Rh(1B)–N(2B) 93.04(9), N(4B)–Rh(1B)–N(2B) 84.51(9), N(1B)–Rh(1B)–N(2B) 164.93(9), O(1B)–Rh(1B)–N(2B) 99.88(8), N(3B)–Rh(1B)–N(2B) 85.04(9).

The structure of 8+ is similar to the structure of recently reported [Co(Me3tpa)(CO3)]

+,42 but with considerably longer metal-ligand bonds (both M−N and M−O bonds). The metal is coordinated by all Me3tpa nitrogen donors and two oxygen atoms of the carbonato ligand. The Rh1–O1 bond trans to the N4 amine is longer than the Rh1–O2 bond trans to lutidyl N2, which results from steric interactions between the methyl group of the trans lutidine with the carbonato ligand.42 Consequently, weaker interactions of O1 with rhodium result in the C1–O1 bond being shorter than the C1–O2 bond.


39

2.3 Conclusions

Both the iridium and rhodium carbonyl complexes with the Me3tpa ligand have remarkably different coordination behavior as compared to their ethene analogues. CO binding to iridium is much stronger than in case of ethene, resulting in formation of a stable biscarbonyl complex, while the bis-ethene iridium complex easily converts to a mono-ethene complex by substitution of ethene by a lutidyl group of the Me3tpa ligand. Contrary to [(κ4-Me3tpa)Rh(ethene)]+ the rhodium carbonyl complex remains 4-coordinate in the solid state. Stabilization of the pentacoordinate geometry of the ethene complex is likely due to stabilization of the rhodo(III)cyclopropane resonance structure, which is not accessible for the carbonyl analogue.

Coordination of Me3tpa ligand in a κ3 fashion results in fast exchange of the lutidyl groups on the NMR time scale. We propose that the fluxionality of the complexes is a result of rapid exchange between 4- and 5-coordinate species.

Bis-carbonyl iridium complex [(κ3-Me3tpa)Ir(CO)2]+ (2+) has electronic properties

similar to the previously reported 1-pyrazolyl complexes, and shows reactivity towards weak nucleophiles like water and methanol at room temperature. Addition of water leads to an unstable hydroxycarbonyl complex, which evolves carbon dioxide to form a bis-hydride complex. Hence, the products obtained in reaction of 2+ with water represent synthetic models for the water gas shift reaction in which all subsequent intermediates of one catalytic WGSR turnover sequence were isolated.

One-electron oxidation of complex 2+ leads to a short-living radical species 22+ that rapidly reacts with adventitious water to yield a dicationic mono-carbonyl hydride complex [(κ4-Me3tpa)Ir(CO)(H)]2+. Thus, the reactivity of the intermediate IrII-carbonyl species towards water is markedly different from the reactivity of its IrI-carbonyl precursor, and also very different from the reactivity of the analogous IrII-ethene species. While [(Me3tpa)Ir(ethene)]2+ tends to react at the ethene ligand via IrIII–CH2CH2• ligand radical intermediates, the putative [(Me3tpa)Ir(CO)]2+ complex apparently behaves as a metallo-radical, without any indications for the expected redox non-innocent behavior of the CO ligand.

Reaction of the rhodium mono-carbonyl complex 3+ with oxygen in the presence of water selectively yields the carbonato complex 8+, which is the first example of an unambiguously characterized rhodium carbonate derived from oxygenation of a carbonyl species. 2.4 Experimental

General procedures:

All manipulations were performed in an argon atmosphere by standard Schlenk techniques or in a glovebox. Methanol and dichloromethane were distilled under nitrogen from CaH2. Hexanes were distilled under nitrogen from Na wire. Acetone and water were deoxygenated using freeze-pump-thaw method. NMR experiments were carried out on a Bruker DRX300 (300 and 75 MHz for 1H and 13C respectively). Solvent shift reference: CD3CN δ = 1.94 and δ = 1.24 for 1H and 13C, acetone-d6 δ = 2.05 and δ = 29.84 for 1H and 13C, CDCl3 δ = 7.26 and δ = 77.16 for 1H and 13C respectively. Abbreviations used are s = singlet, d = doublet, t = triplet, m = multiplet, br = broad. IR measurements in solution were carried out on a Bruker Vertex 70 FTIR spectrometer and solid state measurements were performed on a Shimadzu FTIR 8400S spectrometer equipped with a Specac Golden Gate Single Reflection ATR system. Elemental analyses (CHN) were carried out by H. Kolbe Mikroanalytisches Laboratorium (Germany). X-band EPR spectroscopy measurements were performed with a Bruker EMX Plus spectrometer. Cyclic voltammograms of

Chapter 2

40

~2 mM parent compounds in 10-1 M Bu4NPF6 electrolyte solution were recorded in a gastight single-compartment three-electrode cell equipped with platinum working electrode (apparent surface of 0.42 mm2), coiled platinum wire auxiliary, and silver wire pseudoreference electrodes. The cell was connected to a computer-controlled PAR Model 283 potentiostat. All redox potentials are reported against the ferrocene/ferrocenium (Fc/Fc+) redox couple. Cobaltocenium was used as internal standard. Fast Atom Bombardment (FAB) mass spectrometry was carried out using a JEOL JMS SX/SX 102A four-sector mass spectrometer, coupled to a JEOL MS-MP9021D/UPD system program. Samples were loaded in a matrix solution (3-nitrobenzyl alcohol) onto a stainless steel probe and bombarded with xenon atoms with energy of 3 KeV. During the high resolution FAB-MS measurements a resolving power of 10,000 (10% valley definition) was used. AgPF6 was purchased from Strem. [Ir(cod)(μ-Cl)]2

43 and [Rh(coe)2(μ-Cl)]2 44 have been prepared

according to published procedures. Me3tpa45 has been prepared from bis((6-methyl-2-pyridyl)methyl)amine.46 X-ray diffraction

The structures are shown in Figures 1, 3, 5, 8, 9 and 12,47 which include selected bond distances and angles. The crystal data are shown in Table 3. Crystals were mounted on glass needles. The intensity data were collected at –65 °C on a Nonius Kappa CCD single-crystal diffractometer, using Mo Kα radiation and applying φ and ω scan modes. The intensity data were corrected for Lorentz and polarization effects. A semi-empirical multiscan absorption correction was applied (SADABS).48 The structures were solved by the PATTY option49 of the DIRDIF program system.50 All nonhydrogen atoms were refined with anisotropic temperature factors. The hydrogen atoms were placed at calculated positions, and refined isotropically in riding mode. [1]PF6: The assignment of carbon or nitrogen to C33 and N3 is based on the refinement of the occupancy factors and the anisotropic thermal displacement parameters and on the clear presence of a hydrogen atom near C33 and the absence of any such hydrogen atom near N3 in the difference Fourier map. Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. The calculations revealed no higher symmetry and no (further) solvent accessible areas. [2]PF6: The assignment of carbon or oxygen to C1 and O1 is based on the refinement of the occupancy factors and the anisotropic thermal displacement parameters and on chemical evidence. The assignment of carbon or nitrogen to C22 and N2 is based on the refinement of the occupancy factors and the anisotropic thermal displacement parameters and on the clear presence of a hydrogen atom near C22 and the absence of any such hydrogen atom near N2 in the difference Fourier map. There is some slight disorder in the PF6 moiety but no satisfactory parameterization could be found to describe this disorder. Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. The calculations revealed no higher symmetry and no (further) solvent accessible areas. [3]PF6: The assignment of carbon or oxygen to C1 and O1 is based on the refinement of the occupancy factors and the anisotropic thermal displacement parameters and on chemical evidence. Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. The calculations revealed no higher symmetry and no (further) solvent accessible areas. [4]PF6: Chemical evidence suggests the presence of an additional hydrogen atom bound to the iridium atom, but no such atom could be located reliably in the difference Fourier map due to residual electron density surrounding the heavy atom. All calculated physical properties assume the presence of this hydrogen atom. The assignment of carbon or oxygen to C1 and O1 is based on the refinement of the occupancy factors and the anisotropic thermal displacement parameters and on chemical evidence. There is quite some disorder in the PF6 moiety but no satisfactory parameterization could be found to describe this disorder. Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. The calculations revealed no higher symmetry and no solvent accessible areas. [6]PF6: Most hydrogen atoms were placed at calculated positions, and refined isotropically in riding mode. Hydrogen atoms H1 and H2, bonded to Ir1, were found from the difference fourier map and were chosen from among the residual peaks around Ir1 based on their bond distances and angles. They could not be refined and were kept at fixed positions, assuming that Ir1 would not shift significantly at the end of the refinement. There is some disorder in the PF6 moiety which, unfortunately, could not be parameterized satisfactorily. The highest residual peak in the difference fourier map, 1.58 e·A–3, is close to Ir1, all other significant residual peaks are located within the disordered PF6 moiety. Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. The calculations revealed no higher symmetry and no (further) solvent accessible areas. [8]PF6: Geometrical calculations51 revealed neither unusual geometric features, nor unusual short intermolecular contacts. PLATON, however, reports a void of 100 Å3 containing 3 electrons. No further action was taken to take this electron density into account because no reasonable chemical or physical explanation could be given for this electron density.


41

Table 3. Crystallographic data for [1]PF6, [2]PF6, [3]PF6, [4]PF6 and [8]PF6

[1]PF6 [2]PF6 [3]PF6 [4]PF6 [6]PF6 [8]PF6 Crystal color translucent

light yellow translucent light yellow

translucent orange-red

translucent yellow-brown

translucent colorless

translucent light yellow-brown

Crystal shape rather regular fragment

regular fragment

rather regular fragment

rough fragment

rough fragment

rough rod

Crystal size [mm]

0.21 x 0.14 x 0.06

0.23 x 0.18 x 0.09

0.29 x 0.23 x 0.23

0.26 x 0.17 x 0.13

0.28 x 0.23 x 0.12 mm

0.39 x 0.19 x 0.13

Empirical formula

C29H36F6IrN4P C23H24F6IrN4

O2P C22H24F6N4OPRh

C24H28F6IrN4O3P

C22H26F6IrN4OP

C22H24F6N4O3P Rh

Formula weight

777.79 725.63 608.33 757.67 699.64 640.33

Temperature [K]

208(2) 208(2) 208(2) 208(2) 208(2) 208(2)

Radiation MoKα (graphite mon.) Wavelength [Å]

0.71073

Crystal system Monoclinic Monoclinic Monoclinic Triclinic Triclinic Triclinic Space group P21/c P21/n P21 P-1 P-1 P-1 a [Å] 8.5161(5) 13.8704(14) 7.6326(5) 9.15020(10) 10.9728(7) 8.2853(7) b [Å] 18.9926(16) 13.7945(10) 10.2472(4) 11.1785(6) 11.6245(10) 13.8246(12) c [Å] 18.2925(17) 14.7492(10) 15.0234(4) 13.8428(5) 11.7703(10) 22.4016(14) α [°] 90 90 90 77.101(6) 109.290(6) 79.382(6) β [°] 100.487(8) 115.699(6) 91.011(4) 84.419(3) 110.719(5) 81.380(6) γ [°] 90 90 90 82.292(2) 101.645(5) 79.582(6) Volume [Å3] 2909.3(4) 2542.9(4) 1174.84(9) 1364.39(9) 1236.02(17) 2462.3(3) Z 4 4 2 2 2 4 Density [Mg m-3]

1.776 1.895 1.720 1.844 1.880 1.727

Absorption coefficient [mm-1]

4.710 5.386 0.866 5.026 5.534 0.837

Diffractometer Nonius KappaCCD with area detector Scan φ and ω scan F(000) 1536 1408 612 740 680 1288 θ range [°] 2.14 to 27.50 2.13 to 27.49 2.41 to 27.50 2.25 to 27.50 2.13 to 27.50 2.21 to 27.50 Index ranges -11 ≤ h ≤ 11

-24 ≤ k ≤ 24 -23 ≤ l ≤ 23

-18 ≤ h ≤ 17 -17 ≤ k ≤ 17 -19 ≤ l ≤ 19

-9 ≤ h ≤ 9 -13 ≤ k ≤ 13 -19 ≤ l ≤ 19

-11 ≤ h ≤11 -14 ≤ k ≤ 14 -17 ≤ l ≤ 17

-13 ≤ h ≤14 -15 ≤ k ≤ 15 -15 ≤ l ≤ 15

-10 ≤ h ≤ 10 -17 ≤ k ≤ 17 -29 ≤ l ≤ 29

Reflections collected / unique

50743 / 6645 38847 / 5824 36513 / 5382 38274 / 6229 29328 / 5594 55629 / 11163

R(int) 0.0309 0.0308 0.0275 0.0242 0.0241 0.0189 Reflections observed [Io > 2σ(Io)]

5584 4814 5197 5634 5205 9637

Data / restraints / parameters

6645 / 0 / 373 5824 / 0 / 337 5382 / 1 / 319 6229 / 0 / 356 5594 / 0 / 319 11163 / 0 / 673

Goodness-of-fit on F2

1.139 1.131 1.255 1.188 1.181 1.100

SHELXL-97 weight parameters

0.0105, 4.6873

0.0121, 2.3611

0.0263, 1.6604 0.0429, 1.7367 0.0265, 1.3394 0.0216, 4.6866

Final R indices [I > 2σ(I)]

R1 0.0264 0.0210 0.0266 0.0255 0.0228 0.0336 wR2 0.0441 0.0382 0.0792 0.0710 0.0560 0.0704 R indices (all data)

R1 0.0386 0.0345 0.0281 0.0322 0.0269 0.0429 wR2 0.0472 0.0424 0.0799 0.0757 0.0578 0.0743 Largest diff. peak and hole [e·Å-3]

0.646 / -0.896 0.728 / -0.796 0.976 / -0.604 1.702 / -0.858 1.582 / -0.551 1.082 / -0.937

Chapter 2

42

Syntheses:

[Ir(κ3-Me3tpa)(η4-cod)]PF6 ([1]PF6): 356 mg of [Ir(cod)(μ-Cl)]2 (1.06 mmol) and 333 mg Me3tpa (1.00 mmol) were dissolved in 25 ml MeOH. After the solution became clear, 193 mg KPF6 (1.05 mmol) were added and the solution was stirred for 1 hour. Subsequently, 5 ml H2O were added, and partial evaporation of the solvent caused precipitation of yellow powder that was filtered off and washed with cold MeOH. Yield: 392 mg (0.50 mmol, 50%). A second crop was obtained by leaving the filtrate at –20 oC for one week (187 mg – 24 %). Total yield: 579 mg (0.74 mmol, 74 %). 1H NMR (300 MHz, CD3CN, –20 oC): δ = 7.81 (t, 3J(H,H) = 7.8 Hz, 1H; PyD); 7.67 (t, 3J(H,H) = 7.8 Hz, 2H; PyC); 7.73 (d, 3J(H,H) = 7.5 Hz, 1H; PyD); 7.36 (d, 3J(H,H) = 7.5 Hz, 1H; PyC); 7.36 (d, 3J(H,H) = 7.8 Hz, 1H; PyC); 7.31 (d, 3J(H,H) = 7.5 Hz, 1H; PyD); 7.18 (d, 3J(H,H) = 7.5 Hz, 2H; PyC); 4.99 (d[AB], 2J(H,H) = 15.6 Hz, 2H; N-CH2-PyC); 4.77 (s, 2H; N-CH2-PyD); 3.67 (d[AB], 3J(H,H) = 15.9 Hz, 2H; N-CH2-PyC); 3.28 (m, 2H; CH); 3.26 (s, 6H, PyC-CH3); 2.58 (s, 3H; PyD-CH3); 2.54 (m, 4H; CH, CH2); 2.24 (m, 2H; CH2); 1.53 (m, 2H; CH2); 1.28 (m, 2H; CH2). 13C NMR (75 MHz, CD3CN, –20 oC): δ = 161.6 (Py); 161.3 (Py); 159.2 (PyD); 153.4 (PyD); 138.6 (Py-C4); 138.5 (PyD-C4); 126.7 (Py); 124.2 (PyD); 123.5 (PyD); 121.6 (Py); 63.1 (PyD-CH2-N); 62.8 (CH); 60.5 (Py-CH2-N); 52.8 (CH); 33.8 (CH2); 30.2 (CH2); 29.4 (Py-CH3); 24.3 (PyD-CH3). Elemental analysis: Calcd (C29H36F6IrN4P): C: 44.78, H: 4.67, N: 7.20, found C: 44.76, H: 4.76, N: 7.30. Crystals suitable for X-ray diffraction were obtained from saturated solution of [1]PF6 in wet MeOH at −20 °C under argon. [Ir(κ3-Me3tpa)(CO)2]PF6 ([2]PF6): 251 mg of [1]PF6 (0.317 mmol) was dissolved in 40 ml DCM. CO was bubbled through the solution for 20 minutes, and the mixture was stirred under a CO atmosphere for 30 minutes after which 200 ml Hexanes were added to cause formation of an opaque solution. The vessel was sonicated which caused precipitation of a cotton-like greenish-yellow solid that was filtered and washed with hexanes. Yield: 164 mg (0.226 mmol – 71%). 1H NMR (300 MHz, CDCl3): δ = 7.72 (t, 3J(H,H) = 7.8 Hz, 3H; Py); 7.31 (d, 3J(H,H) = 7.8 Hz, 3H; Py); 7.24 (d, 3J(H,H) = 7.8 Hz, 3H; Py); 4.61 (s, 6H; N-CH2-Py); 2.84 (s, 9H, Py-CH3). 13C NMR (75 MHz, CDCl3): δ = 172.7 (CO); 159.7 (Py-C6); 159.5 (Py-C2); 139.4 (Py-C4); 124.7 (Py-C5); 122.3 (Py-C3); 64.1 (Py-CH2-N); 28.3 (Py-CH3). IR (CH2Cl2: ν(C≡O) = 1975, ν(C≡O) = 2058 cm–1) (solid: ν(C≡O) = 1954, ν(C≡O) = 2036 cm–1). Elemental analysis: Calcd (C23H24F6IrN4O2P): C: 38.07, H: 3.33, N: 7.72, found C: 37.96, H: 3.40, N: 7.67. Crystals suitable for X-ray diffraction were obtained by layering a CH2Cl2 solution of ([2]PF6) with hexanes under argon. [Rh(κ3-Me3tpa)(CO)]PF6 ([3]PF6): 184 mg [Rh(coe)2(μ-Cl)]2 (0.256 mmol) and 173 mg Me3tpa (0.517 mmol) were mixed with 10 ml MeOH. CO was bubbled through until all the solid dissolved. Subsequently 132 mg of KPF6 (0.717 mmol) was added and the solvent was condensed to 5 ml under vacuum causing precipitation of orange powder that was filtered and washed with methanol. Yield 166 mg (0.273 mmol – 53%). Product was recrystallized by layering 0.5 ml acetone solution with 2.5 ml methanol in –20 °C to yield 132 mg (0.217 mmol – 42 %) of the pure product. 1H NMR (300 MHz, acetone-d6, –30 oC, all peaks broad): δ = 7.85 (t, 3J(H,H) = 7.8 Hz, 2H; PyC); 7.38 (m, 6H, PyC/D); 6.95 (d, 3J(H,H) = 7.2 Hz, 1H; PyD); 5.49 (d[AB], 3J(H,H) = 15.3 Hz, 2H; N-CH2-PyC); 4.70 (d[AB], 3J(H,H) = 15.6 Hz, 2H, N-CH2-PyC); 4.11 (s, 2H; N-CH2-PyD); 2.75 (6H, s, PyC-Me); 2.61 (s, 3H; PyD-CH3). 13C NMR (75 MHz, acetone-d6, –30 oC): δ = 190.7 (d, 1J(Rh,C) = 80.9 Hz, CO); 165.4 (PyC); 162.8 (PyD); 159.7 (PyD); 154.8 (PyC); 140.5 (PyC); 138.5 (PyD); 126.3 (PyC); 124.6 (PyC); 124.2 (PyD); 122.2 (PyD); 69.2 (PyC-CH2-N); 66.2 (PyD-CH2-N); 29.9 (PyC-CH3); 26.1 (PyD-CH3).


43

IR (CH2Cl2: ν(C≡O) = 1983, shoulder 1966 cm–1) (solid: ν(C≡O) = 1939 (large), 1896 (very small) cm–1). Elemental Analysis Calcd (C22H24F6N4OPRh): C: 43.44, H: 3.98, N: 9.21, found C: 42.19, H: 3.87, N: 8.97 (Due to the rather high oxygen sensitivity of 3+ the compound seems to have partly (~ 50%) reacted with O2 by the time it was received by the external analytical laboratory). Crystals suitable for X-ray diffraction were obtained by layering acetone solution of ([3]PF6) with MeOH under argon. [Ir(κ3-Me3tpa)(CO)(H)(COOMe)]PF6 ([4]PF6): 57.5 mg of [2]PF6 (0.078 mmol) were dissolved in 3 ml MeOH and stirred overnight. The solution was condensed to approx 1 ml and the solvent was removed with a pipette. The thus obtained off-white powder was dried in vacuo. Yield: 25.8 mg (0.033 mmol – 42.3%). 1H NMR (300 MHz, CDCl3): δ = 7.81 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.72 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.63 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.53 (d br, 3J(H,H) = 7.5 Hz, 1H, Py); 7.33 (d br, 3J(H,H) = 7.8 Hz, 1H; Py); 7.31 (d, 3J(H,H) = 7.8 Hz, 2H; Py); 7.21 (d br, 3J(H,H) = 7.8 Hz, 1H; Py); 7.13 (d br, 3J(H,H) = 7.8 Hz, 1H; Py); 5.39 (d[AB], 2J(H,H) = 17.4 Hz, 1H; N-CH2-Py); 5.10 (d[AB], 2J(H,H) = 15.6 Hz, 1H; N-CH2-Py); 4.84 (d[AB], 2J(H,H) = 15.6 Hz, 1H; N-CH2-Py); 4.77 (d[AB], 2J(H,H) = 13.5 Hz, 1H; N-CH2-Py); 4.69 (d[AB], 2J(H,H) = 13.5 Hz, 1H; N-CH2-Py); 4.62 (d[AB], 2J(H,H) = 17.4 Hz, 1H; N-CH2-Py); 3.48 (s, 3H; OMe); 2.89 (s, 3H; Py-CH3); 2.80 (s, 3H; Py-CH3); 2.29 (s, 3H, Py-CH3); -16.75 (s, 1H, Ir-H). 13C NMR (75 MHz, CDCl3): δ = 163.2 (Py); 163.0 (COO); 161.3 (Py); 160.6 (CO); 160.1 (Py), 159.9 (Py); 158.5 (Py); 152.6 (Py); 140.8 (Py-C4); 140.2 (Py-C4); 137.3 (Py-C4); 125.6 (Py); 124.6 (Py); 123.5 (Py); 122.7 (Py); 121.6 (Py); 120.5 (Py); 67.7 (br, NCH2Py); 51.4 (OMe); 30.6 (2x PyCH3); 23.8 (PyCH3). IR (CH2Cl2) (ν(C–O) = 1093, ν(C≡O) = 2062 cm–1), C=O signal is obscured by lutidine ring signals. Elemental Analysis: Calcd (C24H28F6N4Ir): C: 38.04; H: 3.72; N: 7.39, found: C: 37.88; H: 3.67; N: 7.30. Crystals suitable for X-ray diffraction grew overnight from saturated solution of ([2]PF6) in wet MeOH under argon. [Ir(κ3-Me3tpa)(CO)(H)(COOH)]PF6 ([5]PF6): 50 mg of [2]PF6 (0.069 mmol) was dissolved in a mixture of 0.8 ml acetone and 0.8 ml water and stirred for 5 hours. 2.5 ml of water was added and the solution was condensed in vacuo to approx. 2 ml causing precipitation of an off-white solid. Yield: 32 mg (0.043 mmol – 62%). 1H NMR (300 MHz, acetone-d6): δ = 8.01 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.92 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.78 (t, 3J(H,H) = 7.8 Hz, 1H; Py); 7.64 (d, 3J(H,H) = 7.8 Hz, 1H; Py); 7.57 (m, 2H, Py); 7.45 (m, 2H, Py); 7.29 (d, 3J(H,H) = 7.8 Hz, 1H; Py); 5.57 (d[AB], 2J(H,H) = 17.4 Hz, 1H; N-CH2-Py); 5.38 (d[AB], 2J(H,H) = 15.9 Hz, 1H; N-CH2-Py); 5.09 (d[AB], 2J(H,H) = 13.8 Hz, 1H; N-CH2-Py); 4.97 (d[AB], 2J(H,H) = 13.8 Hz, 1H; N-CH2-Py); 4.94 (d[AB], 2J(H,H) = 15.6 Hz, 1H; N-CH2-Py); 4.80 (d[AB], 2J(H,H) = 17.4 Hz, 1H; N-CH2-Py); 3.00 (s, 3H; Py-CH3); 2.93 (s, 3H; Py-CH3); 2.34 (s, 3H; Py-CH3); -16.56 (s, 1H; Ir-H). Decomposition of the sample in acetone solution prevented recording a 13C NMR spectrum. IR (CH2Cl2: ν(C≡O) = 2064 ν(C=O) = 1612 cm–1) IR (solid: ν(C≡O) = 2050 cm–1), C=O signal is obscured by lutidine ring signals. FAB+-MS: m/z = 599.2[M]+, 555.2 [M - CO2]

+. Elemental analysis: Calcd (C23H26F6IrN4O3P): C: 37.15, H: 3.52, N: 7.53, found C: 36.94, H: 3.47, N: 7.44. [Ir(κ3-Me3tpa)(CO)(H)2]PF6 ([6]PF6): 30 mg of [4]PF6 in was dissolved in 2 ml acetone and stirred for 20 hrs. The reaction was 93% selective in formation of [6]+ according to 1H NMR indicating presence of 7% of a very similar symmetric bishydride species with hydride signal at 18.04 ppm. The compound was isolated as an off-white foam by evaporation of the solvent.

Chapter 2

44

1H NMR (300 MHz, acetone-d6): δ = 8.00 (t, 3J(H,H) = 7.8 Hz, 1H; PyA); 7.95 (t, 3J(H,H) = 7.8 Hz, 1H; PyB); 7.76 (t, 3J(H,H) = 7.8 Hz, 1H; PyD); 7.58 (m, 3H; 2xPyA,PyB); 7.48 (m, 2H; PyB,PyD); 7.26 (d, 3J(H,H) = 8.1 Hz, 1H; PyD); 5.51 (d[AB], 2J(H,H) = 16.5 Hz, 1H; N-CH2-Py); 5.34 (d[AB], 3J(H,H) = 15.9 Hz, 1H; N-CH2-Py); 4.96 (s, 2H, N-CH2-PyD); 4.93 (d[AB], 2J(H,H) = 15.9 Hz, 1H; N-CH2-Py); 4.77 (d[AB], 2J(H,H) = 16.5 Hz, 1H; N-CH2-Py); 2.93 (s, 3H; PyA-CH3); 2.86 (s, 3H; PyB-CH3); 2.37 (s, 3H; PyD-CH3); -18.14 (d[AB], 2J(H,H) = 8.4 Hz, 1H; Ir-H); -18.25 (d[AB], 2J(H,H) = 8.4 Hz, 1H; Ir-H). 13C NMR (75 MHz, acetone-d6): δ = 164.9, 164.4, 163.1, 162.4, 159.9, 155.9, 152.0, (Py and CO); 142.3 (Py-C4); 141.7 (Py-C4); 139.2 (Py-C4); 127.3 (Py); 126.5 (Py); 125.0 (Py); 124.9 (Py); 122.7 (Py); 122.2 (Py); 71.0 (NCH2Py); 69.7 (NCH2Py); 68.7 (NCH2Py); 32.5 (2x PyCH3); 25.2 (PyCH3). IR (CH2Cl2: ν(C≡O) = 2031 cm–1) IR (solid: ν(C≡O) = 2018 cm–1). Elemental analysis: Calcd (C22H26F6IrN4O2P): C: 37.77, H: 3.75, N: 8.01, found C: 38.05, H: 3.70, N: 7.80. Crystals suitable for X-ray diffraction were obtained by layering acetone solution of [6]PF6 with diethyl ether under argon. [Ir(κ4-Me3tpa)(CO)(H)](PF6)2 ([7](PF6)2): In a glovebox 121 mg (0.167 mmol) [2]PF6 and 42.1 mg (0.166 mmol) AgPF6 were put in a flame-dried schlenktube. 2 ml of acetone-d6 (H2O + D2O < 0.02%) was added and little gas evolution could be observed. The mixture was left for three days. Exchange of the Ir-H and deuterium was observed leading to a low intensity of the hydride signal which totally disappears in one week time. This batch gave a very clean NMR spectrum and the product was isolated as yellow-green foam by evaporation of solvent. 1H NMR (300 MHz, acetone): δ = 8.10 (t, 3J(H,H) = 7.8 Hz, 2H; PyE); 7.91 (t, 3J(H,H) = 7.8 Hz, 1H; PyA); 7.72 (d br, 3J(H,H) = 7.8 Hz, 2H; PyE); 7.67 (d br, 3J(H,H) = 7.8 Hz, 2H; PyE); 7.62 (d br, 3J(H,H) = 7.8 Hz, 1H; PyA); 7.44 (d br, 3J(H,H) = 7.8 Hz, 1H; PyA); 5.88 (d[AB], 2J(H,H) = 15.9 Hz, 2H; N-CH2-PyE); 5.72 (d[AB], 2J(H,H) = 15.9 Hz, 2H; N-CH2-PyE); 5.44 (s, 2H; N-CH2-PyA); 3.47 (s, 3H, PyA-Me); 3.13 (s, 6H, PyE-CH3); -18.32 (s, 1H, Ir-H). 13C NMR (75 MHz, Acetone): δ=166.2 (CO); 165.7 (PyE-C6); 165.5 (PyA-C6); 162.0 (PyA-C2); 161.8 (PyE-C2); 143.7 (PyE-C4); 142.5 (PyA-C4); 129.4 (PyE); 128.3 (PyA); 124.3 (PyE); 122.4 (PyA); 74.1 (PyE-CH2-N); 72.1 (PyA-CH2-N); 29.6 (PyA-CH3); 29.5 (PyE-CH3). IR (CH2Cl2) (ν(C≡O) = 2069 cm–1). Elemental analysis: Calcd (C22H25F12IrN4OP2): C: 31.32, H: 2.99, N: 6.64, found C: 31.37, H: 3.08, N: 6.55. FAB+-MS: m/z = 699.2 {[M]PF6}

+, 553.2 [M (minus hydride)]+. [Rh(κ4-Me3tpa)(CO3)]PF6 ([8]PF6): 33.0 mg (0.0542 mmol) of [3]PF6 was dissolved in a mixture of 0.5 ml of acetone-d6 and 0.2 ml D2O and the solution was left under air atmosphere overnight leading to a nearly quantitative conversion to the product. Solution was condensed under vacuum to approx 0.4 ml and left overnight at –20 °C. Yield: 25.3 mg (0.0395 mmol – 73%) 1H NMR (300 MHz, acetone-d6 : D2O (5:2)): δ = 7.90 (t, 3J(H,H) = 7.8 Hz, 2H; PyE); 7.74 (t, 3J(H,H) = 7.8 Hz, 1H; PyA); 7.54 (d, 3J(H,H) = 7.2 Hz, 2H; PyE); 7.45 (d, 3J(H,H) = 8.4 Hz, 1H; PyA); 7.41 (d, 3J(H,H) = 7.8 Hz, 2H; PyE); 7.25 (d, 3J(H,H) = 7.8 Hz, 1H; PyA); 5.53 (d[AB], 2J(H,H) = 15.9 Hz, 2H; N-CH2-PyE); 5.21 (d[AB], 2J(H,H) = 15.9 Hz, 2H; N-CH2-PyE); 5.07 (s, 2H; N-CH2-PyA); 3.37 (s, 3H; PyA-Me); 2.97 (s, 6H; PyE-CH3). 13C NMR (75 MHz, acetone-d6 : D2O (5:2)): 168.1 (CO3

2- ligand), 166.3 (PyE); 165.4 (PyA); 163.8 (PyE); 163.1 (PyA); 142.3 (PyE-C4); 141.3 (PyA-C4); 129.6 (PyE); 128.6 (PyA); 122.8 (PyE); 121.1 (PyA); 71.7 (PyE-CH2-N); 69.7 (PyA-CH2-N); 27.0 (PyA-CH3); 24.4 (PyD-CH3). IR (MeCN: ν(C=O) = 1681 (br) cm–1). Elemental Analysis Calcd (C22H24F6N4O3PRh · ½H2O): C: 40.69, H: 3.88, N: 8.63, found C: 40.52, H: 3.72, N: 8.49.


45

Crystals suitable for X-ray diffraction were obtained from saturated solution of [3]PF6 in methanol-water at –20 °C without taking precautions against air.

2.5 Acknowledgements

We thank Jan Smits for the X-ray structure determinations, Jan Meine Ernsting for his assistance with the low temperature NMR experiments and Han Peeters for measuring FAB+-MS mass spectra. 2.6 Notes and References

1 (a) Blackman, A. G. Polyhedron, 2005, 24, 1-39. (b) Blackman, A. G. Eur. J. Inorg. Chem. 2008, 2633-2647.

2 Wu, A. J.; Penner-Hahn, J. E.; Pecoraro, V. L. Chem. Rev. 2004, 104, 903-938. 3 Kim, E.; Chufán, E. E.; Kamaraj, K.; Karlin, K. D. Chem. Rev. 2004, 104, 1077-1134 4 (a) Mirica, L. M.; Ottenwaelder, X.; Stack, T. D. P. Chem. Rev. 2004, 104, 1013-1046. (b) Lewis, E. A.;

Tolman, W. B. Chem. Rev. 2004, 104, 1047-1076. 5 (a) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que Jr., L. Chem. Rev. 2004, 104, 939-986. (b) Tshuva, E. Y.;

Lippard, S. J. Chem. Rev. 2004, 104, 987-1012. 6 Yamaguchi, M.; Kousaka, H.; Izawa, S.; Ichii, Y.; Kumano, T.; Masui, D.; Yamagishi, T. Inorg. Chem.

2006, 45, 8342-8354. 7 (a) de Bruin, B.; Boerakker, M. J.; Donners, J. J. J. M.; Christiaans, B. E. C.; Schlebos, P. P. J.; de Gelder,

R.; Smits, J. M. M.; Spek, A. L.; Gal. A. W. Angew. Chem. Int. Ed. 1997, 36, 2064-2067. (b) de Bruin, B.; Boerakker, M. J.; Verhagen, J. A. W.; de Gelder, R.; Smits, J. M. M.; Gal, A. W. Chem. Eur. J. 2000, 6, 298-312. (c) de Bruin, B.; Peters, T. P. J.; Wilting, J. B. M.; Thewissen, S.; Smits, J. M. M.; Gal, A. W. Eur. J. Inorg. Chem. 2002, 2671-2680. (d) Budzelaar, P. H. M.; Blok A. N. J. Eur. J. Inorg. Chem. 2004, 2385-2391. (e) For a review on models for Rh-mediated olefin oxygenation catalysis see: de Bruin B.; Budzelaar, P. H. M.; Gal, A. W. Angew. Chem. Int. Ed. 2004, 43, 4142-4157.

8 de Bruin, B.; Boerakker, M. J.; de Gelder, R.; Smits, J. M. M.; Gal, A. W. Angew. Chem. Int. Ed. 1999, 38, 219-222.

9 (a) de Bruin, B.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Wilting, J. B. M.; de Gelder, R.; Smits, J. M. M.; Gal, A. W. Angew. Chem. Int. Ed. 2002, 41, 2135-2138. (b) Hetterscheid, D. G. H.; Bens, M.; de Bruin, B. Dalton Trans. 2005, 979-984.

10 (a) de Bruin, B.; Thewissen, S.; Yuen, T. W.; Peters, T. P. J.; Smits, J. M. M.; Gal, A. W. Organometallics 2002, 21, 4312-4314. (b) Hetterscheid, D. G. H.; Kaiser, J.; Reijerse, E.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Smits, J. M. M.; de Gelder, R.; de Bruin, B. J. Am. Chem. Soc. 2005, 127, 1895-1905.

11 (a) See Chapter 5 of this Thesis. (b) Dzik, W. I.; Reek, J. N. H.; de Bruin, B. Chem. Eur. J. 2008, 14, 7594-7599.

12 (a) van Santen, R. A.; van Leeuwen, P. W. N. M.; Moulijn, J. A.; Averill, B. A. (Eds) Catalysis: An Integrated Approach, Second, Revised and Enlarged Edition, Elsevier, Amsterdam 1999. b) van Leeuwen, P. W. N. M. Homogeneous Catalysis: Understanding the Art, Kluwer Academic Publishers, Dordrecht, Boston, London 2004.

13 Addison, A. W.; Rao, T. N.; Reedijk, J.; van Rijn, J.; Verschoor, G. C. J. Chem. Soc., Dalton Trans. 1984, 1349-1356.

14 Rossi, A. R.; Hoffmann, R. Inorg. Chem. 1975, 14, 365-374. 15 Hetterscheid, D. G. H.; Klop, M.; Kicken, R. J. N. A. M.; Smits, J. M. M.; Reijerse, E. J.; de Bruin, B.

Chem. Eur. J. 2007, 13, 3386-3405. 16 de Bruin, B.; Kicken, R. J. N. A. M.; Suos, N. F. A.; Donners, M. P. J.; den Reijer, C. J.; Sandee, A. J.; de

Gelder, R.; Smits, J. M. M.; Gal, A. W.; Spek, A. L. Eur. J. Inorg. Chem. 1999, 1581-1592. 17 (a) see Chapter 3 of this Thesis. (b) Dzik, W. I.; Creusen, C.; de Gelder, R.; Peters, T. P. J.; Smits, J. M.

M.; de Bruin, B. Organometallics 2010, 29, 1629-1641. 18 Elian, M.; Hoffmann, R. Inorg. Chem. 1975, 14, 1058-1076 19 A similar equilibrium was found for [Cu(tpa)(CO)]+ complexes, revealing a shoulder in the IR spectrum

and a single NMR signal corresponding to the coordinated CO molecule: (a) Kretzer, R. M.; Ghalidi, R. A.; Lebeau, E. L.; Liang, H. -C.; Karlin, K. D. Inorg. Chem. 2003, 42, 3016-3025. (b) Fry, H. C.; Lucas, H. R.; Narducci Sarjeant, A. A.; Karlin, K. D.; Meyer, G. J. Inorg. Chem. 2008, 47, 241-256.

20 Mitoraj, M.; Michalak, A. Organometallics 2007, 26, 6576-6580.

Chapter 2

46

21 DFT (BP86, def-TZVP basis set) optimized structure of the [Rh(Me3tpa)(ethene)]+ complex has the ethene C-C bond distance equal to 1.44 (vs. 1.34 and 1.54 for ethene and ethane respectively) which suggests a significant rhodo(III)cyclopropane contribution to the Rh-ethene interactions.

22 (a) Constable, E. C. Prog. Inorg. Chem. 1994, 42, 67-138. (b) Blackman, A. G. C. R. Chim. 2005, 8, 107-119.

23 (a) Zhang, Z. H.; Bu, X. H.; Zhu, Z. A.; Chen, Y. T. Polyhedron 1996, 15, 2787-2792. (b) Lonnon, D. G.; Craig, B. C.; Colbran, S. B. Dalton Trans. 2006, 3785-3797.

24 Elliott, P. I. P.; Haslam, C. E.; Spey, S. E.; Haynes, A. Inorg. Chem. 2006, 45, 6269-6275. 25 (a) Fernandez, M. J.; Rodriguez, M. J.; Oro, L. A. Polyhedron 1991, 10, 1595-1598. (b) Mathieu, R.;

Esquius, G.; Lugan, N.; Pons, J.; Ros, J. Eur. J. Inorg. Chem. 2001, 2683-2688. (c) Aullon, G.; Esquius, G.; Lledos, A.; Maseras, F.; Pons, J.; Ros, J. Organometallics 2004, 23, 5530-5539. (d) Adams, C. J., Anderson, K. M.; Charmant, J. P. H.; Connely, N. G.; Field, B. A.; Halett, A. J.; Horne, M. Dalton Trans. 2008, 2680-2692.

26 (a) Webster, C. E.; Hall, M. B. Inorg. Chim. Acta. 2002, 330, 268-282, 336, 168. (b) Blake, A. J.; George, M. W.; Hall, M. B.; McMaster, J.; Portius, P.; Sun, X. Z.; Towrie, M.; Webster, C. E.; Wilson, C.; Zarić, S. D. Organometallics 2008, 27, 189-201.

27 de Bruin, B.; Brands, J. A.; Donners, J. J. J. M.; Donners, M. P. J.; de Gelder, R.; Smits, J. M. M.; Gal, A. W.; Spek, A. L. Chem. Eur. J. 1999, 5, 2921-2936.

28 Fernandez, M. J.; Rodriguez, M. J.; Oro, L. A. J. Organomet. Chem. 1992, 438, 337-342. 29 (a) Malatesta, L.; Caglio, G.; Angoletta, M. J. Chem. Soc. 1965, 6974-83. (b) Clark, H. C.; von Werner,

K. Syn. React. Inorg. Metal-org. Chem., 1974, 4, 355-364. 30 (a) Ibekwe, S. D.; Taylor, K. A. J. Chem. Soc. A 1970, 1-3. (b) Del Zotto, A.; Mezzetti, A.; Dolcetti, G.;

Rigo, P.; Pahor, N. B. J. Chem. Soc., Dalton Trans. 1989, 607-613. 31 Esteruelas, M. A.; Oro, L. A.; Clarmunt, R. M.; López, C.; Lavandera, J. L.; Elguero, J. J. Organomet.

Chem. 1989, 366, 245-255. 32 Cordaro, J. G.; Bergman, R. G. J. Am .Chem. Soc. 2004, 126, 16912-16929. 33 For a general overview see: (a) Ford, P. C. Acc. Chem. Res. 1981, 14, 31. (b) King, R. B. J. Organomet.

Chem. 1999, 586, 2. (c) Laine, R. M.; Crawford, E. J. J. Mol. Catal. 1988, 44, 357. 34 The peak separation of cobaltocene used as internal standard in the measurement was 124 mV. In this

case we would expect a fully reversible wave with a peak separation equal to 59 mV. 35 (a) Zhang, X. X.; Parks, G. F.; Wayland, B. B. J. Am. Chem. Soc. 1997, 119, 7938-7944. (b) Bunn, A.G.;

Wayland, B. B. J. Am. Chem. Soc. 1992, 114, 6917-6919. (c) Wayland, B. B.; Sherry, A. E.; Coffin, V. L. J. Chem. Soc., Chem. Commun. 1989, 662-663. (d) Cui, W.; Li, S.; Wayland, B. B. J. Organomet. Chem. 2007, 692, 3198-3206.

36 Although no X-ray diffraction data was available for 72+, we assume that the hydride is coordinated trans to the weaker donating amine, and CO is coordinated trans to the stronger donating lutidyl moiety. Also for steric reasons, coordination of hydride trans to the amine seems more probable, since this position is strongly shielded by three methyl groups of the coordinated lutidine fragments.

37 Formation of a N–H bond in deuterated THF was observed for [Ir(trop2N•)(phen)]+ aminyl radical and traces of water were considered as the hydrogen atom source: Donati, N.; Stein, D.; Büttner, T.; Schönberg, H.; Harmer, J.; Anadaram, S.; Grützmacher, H. Eur. J. Inorg. Chem. 2008, 4691-4703.

38 Feller, M.; Ben-Ari, E.; Gupta, T.; Shimon, L. J. W.; Leitus, G.; Diskin-Posner, Y.; Weiner, L.; Milstein, D. Inorg. Chem. 2007, 46, 10479-10490.

39 (a) Dixon, F. M.; Masar III, M. S.; Doan, P. E.; Farrell, J. R.; Arnold Jr, F. P.; Mirkin, C. A.; Incarvito, C. D.; Zakharov, L. N.; Rheingold, A. L. Inorg. Chem. 2003, 42, 3245-3255. (b) Garcia, M. P.; Jimenez, M. V.; Cuesta, A.; Siurana, C.; Oro, L. A.; Lahoz, F. J.; Lopez, J. A.; Catalan, M. P.; Tiripicchio, A.; Lanfranchi, M. Organometallics 1997, 16, 1026-1036. (c) Haefner, S. C.; Dunbar, K. R.; Bender, C. J. Am. Chem. Soc. 1991, 113, 9540-9553.

40 Booth, B. L.; McAluliffe, C. A.; Stanley, G. L. J. Organomet. Chem. 1982, 226, 191-198. 41 (a) Siegl, W. O.; Lapporte, S. J.; Collman, J. Inorg. Chem. 1971, 10, 2158-2165. (b) Lawson, H. J.;

Atwood, J. D. J. Am. Chem. Soc. 1988, 110, 3680-3682. (c) Lawson, H. J.; Atwood, J. D. J. Am. Chem. Soc. 1989, 111, 6223-6227. (d) Ciriano, M. A., López, J. A.; Oro, L. A.; Pérez-Torrente, J. J.; Lanfranchi, M.; Tripicchio, A.; Tripicchio Camellini, M. Organometallics 1995, 14, 4764-4775.

42 Cheyne, S. E.; McClintock, L. F.; Blackman, A. G. Inorg. Chem. 2006, 45, 2610-2618. 43 Winkhaus, G.; Singer, H. Chem. Ber. 1966, 99, 3610-3618. 44 van der Ent, A.; Onderdenlinden, A. L. Inorg. Synth. 1990, 28, 90-91. 45 Nagao, H.; Komeda, N.; Mukaida, M.; Suzuki, M.; Tanaka, K. Inorg. Chem. 1996, 35, 6809-6815. 46 Komatsu, K.; Kikuchi, K.; Kojima, H.; Urano, Y.; Nagano, T. J. Am. Chem. Soc. 2005, 127, 10197-

10204.


47

47 (a) Ortep-3 for Windows: Farrugia, L. J. J. Appl. Cryst. 1997, 30, 565. (b) Rendering was made with POV-Ray 3.6, Persistence of Vision Pty. Ltd. (2004), Persistence of Vision Raytracer (Version 3.6), Retrieved from http://www.povray.org/download/

48 Sheldrick, G. M. SADABS; University of Göttingen: Germany, 1996. 49 Beurskens, P. T.; Beurskens, G.; Strumpel, M.; Nordman, C.E. in Patterson and Pattersons, Clarendon,

Oxford, 1987, p.356. 50 Beurskens, P. T.; Beurskens, G.; Bosman, W. P.; de Gelder, R.; García-Granda, S.; Gould, R. O.; Israël,

R.; Smits, J. M. M. University of Nijmegen, The Netherlands, 1996. 51 Spek, A. L. PLATON, A Multipurpose Crystallographic Tool. Utrecht University, Utrecht, The

Netherlands, 2003.

Chapter 3

Carbonyl Complexes of Rhodium with N-donor Ligands: Factors Determining the Formation of

Terminal versus Bridging Carbonyls*

R

N N

NRh

CO

N

N N

NRh

N

NN

NRh

O

O

N

N

NRh

N

N

NRh

O

O

O

R

R

+

I

2+

I

I

I

I

− CO

2+

R = H, Alkyl R = 2-picolyl * Part of this work has been published: Reproduced in part with permission from [Dzik, W. I.; Creusen, C.; de Gelder, R.; Peters, T. P. J.; Smits, J. M. M.; de Bruin, B. Organometallics 2010, 29, 1629-1641] Copyright [2010] American Chemical Society

Chapter 3

50

3.1 Introduction

Carbonyl complexes of rhodium with the tridentate nitrogen-donor ‘scorpionato’ trispyrazolyl (Tp) type of ligands have received quite some attention over the past two decades.1 It was shown that complexes of the type Rh(Tp*)(CO)2 (Tp* = tris(3,5-dimethylpyrazolyl)borate) photochemically activate C-H bonds of alkanes,2 show a high degree of fluxionality in solution3 and reveal interesting one-electron redox properties in the presence of a supporting phosphorous ligand.4

The arrangement of the pyrazolyl moieties of the Tp-type ligands imposes a fac- coordination mode.1 Furthermore, RhI complexes with Tp-type ligands reveal a hemilabile coordination mode through the κ2-κ3 equilibrium, producing diverse structures spanning from mononuclear square planar and trigonal bypiramidal or square pyramidal bis-carbonyls,5 to octahedral dinuclear triscarbonyl bridged6 species (Scheme 1). Similar structures (although κ2-κ3 isomerism was not reported) were found for cyclic fac- coordinating tri-7 and hexa-amines,8 with the latter ones forming tris-carbonyl bridged tetranuclear assemblies.

Scheme 1. Different coordination behavior of trispyrazolyl ligands X = BH, CH, GaMe.

While these strictly fac-coordinating N3-donor ligands have been thoroughly studied, surprisingly little attention has been given to rhodium carbonyl complexes with more flexible podal N3-donor ligands that can adopt both fac- and mer- coordination modes.

Mathieu and Ros reported that bis[(3,5-dimethyl-1-pyrazolyl)methyl]ethylamine can bind in both fashions to a cationic rhodium carbonyl centre resulting in formation of either mono-carbonyl square planar or bis-carbonyl square pyramidal complexes (Scheme 2).9 The factors that determine the coordination mode of the N3-donor ligand and as a result the number of carbonyl ligands per metal atom, were not investigated.

Scheme 2. Mer- and fac- coordination of a N3-ligand towards rhodium carbonyl.

The above structural diversity of N3-donor ligand Rh-carbonyl complexes is intriguing and could well result in different reactivities. For this reason we became interested in the factors that determine the coordination mode of both the carbonyl ligands and the N-donor ligands in cationic [{Rh(CO)x(N-donor ligand)}y]

y+ complexes with flexible podal N-donor ligands.

Given the rich chemistry of rhodium olefin complexes with bispicolylamine (bpa) type ligands10,11 and dual fac- and mer- coordination behavior of bpa, we were interested in the coordination modes of the flexible ligands shown in Scheme 3, and in

Carbonyl Complexes of Rhodium with N-Donor Ligands

51

understanding the factors that drive the formation of mononuclear complexes with terminal carbonyl ligands versus binuclear carbonyl bridged complexes.

N

N

N

N

N

N

H N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

N

Bu-bpabpa tpa

tpbntppntpen

Me3tpaMe-bpa

Scheme 3. Ligands used in this study.

Bridging carbonyl complexes are frequently ‘dormant state’ or ‘dead end’ species in several catalytic carbonylation reactions (e.g. hydroformylation)12,13 and therefore a better understanding of the factors that determine their formation might well be of synthetic relevance. 3.2 Results and discussion

For the tridentate bpa type of ligands we compared the Me-bpa and Bu-bpa ligands having a methyl or butyl substituent on the central amine donor, respectively, with the non-functionalized bpa ligand to investigate the influence of the electronic effects (stronger bpa donor vs. weaker Me-bpa or Bu-bpa)14 on the formation of mononuclear terminal vs. bridging carbonyl complexes (Scheme 3). We also investigated ditopic alkyl-bpa bridged complexes with an alkyl tether between two bpa units and their potential more favourable entropic factors associated with the formation of bridged carbonyl complexes. For the tetradentate tpa type ligands we compared the stronger non-functionalized tpa donor with the Me3tpa donor bearing methyl substituents on the pyridine-6 positions (Scheme 3). These substituents have some steric influence on the metal coordination which results in the pyridine units being weaker donors to the metal compared with non-substituted pyridines.11 3.2.1 Rhodium carbonyl complexes with bis(picolyl)amine, N-methyl(bis(picolyl)amine) and N-butyl(bis(picolyl)amine)

Reaction of [Rh(κ3-bpa)(η4-cod)]PF6 ([1]PF6) (cod = cis,cis-1,5-cyclooctadiene) with CO at a pressure of 1 bar in dichloromethane results in formation of the tris-carbonyl bridged binuclear complex [Rh2(κ3-bpa)2(μ-CO)3](PF6)2 ([2](PF6)2) (Scheme 4). The complex has C2 symmetry and is not fluxional in solution on the NMR timescale, which results in separate NMR signals of the inequivalent pyridine and methylene groups of

Chapter 3

52

the bpa ligand. The inequivalence is a result of the coordination mode in which two CO ligands are trans to picolyl and amine moieties, whereas the third CO is trans to the two picolyl groups of the different bpa ligands. This coordination mode is similar to that previously reported for [Rh2(μ-CO)3Cl2(Py)2]

15 and was confirmed by single crystal X-ray diffraction. The crystal was grown by layering an acetone solution of [2](PF6)2 with hexanes at −20 °C. Although the quality of the crystal data is hampered by severe PF6 disorder, the molecular structure of the dication 22+ is unambiguously revealed by the X-ray data (Figure 1).

H

N

N

PyN

O

O Py

HNRh

O

RhN Rh

HN

N

N

N

H Rh COI II

+

Py

+

1 bar CO I

3+

2+

22+1+

− CO

+ CO

Scheme 4. Synthesis of [Rh2(κ3-bpa)2(μ-CO)3]2+ (22+) and [Rh(κ3-bpa)(CO)]+ (3+)

Figure 1. X-ray structure of [Rh2(κ3-bpa)2(μ-CO)3]2+ (22+). Thermal ellipsoids are drawn with 50% probability. Hydrogen atoms bound to the carbon atoms, the PF6

– counter ions and acetone molecules are omitted for clarity. Selected bond lengths [Å] and angles [°]: Rh(1)–C(1) 2.004(14); Rh(1)–C(1i) 2.019(12); Rh(1)–C(2) 2.019(18); Rh(1)–N(1) 2.165(11); Rh(1)–N(2) 2.176(12); Rh(1)–N(3) 2.184(12); Rh(1)–Rh(1i) 2.5710(18); C(1)–O(1) 1.140(16); C(2)–O(2) 1.14(2); C(1)–Rh(1)–C(1i) 82.1(6); C(1)–Rh(1)–C(2) 84.6(5); C(1i)–Rh(1)–C(2) 84.2(5); C(1)–Rh(1)–N(1) 177.4(5); C(1i)–Rh(1)–N(1) 100.5(5); C(2)–Rh(1)–N(1) 96.0(4); C(1)–Rh(1)–N(2) 95.2(5); C(1i)–Rh(1)–N(2) 99.3(5); C(2)–Rh(1)–N(2) 176.5(5); N(1)–Rh(1)–N(2) 84.1(4); C(1)–Rh(1)–N(3) 98.6(5); C(1i)–Rh(1)–N(3) 177.5(5); C(2)–Rh(1)–N(3) 98.3(5); N(1)–Rh(1)–N(3) 78.9(4); N(2)–Rh(1)–N(3) 78.2(5); Rh(1)–C(1)–Rh(1i) 79.5(4); Rh(1i)–C(2)–Rh(1) 79.1(8).

The IR spectrum of the carbonyl region reveals two overlapping bands at 1835 and 1828 cm–1 and 13C NMR spectroscopy reveals two triplets of intensity 2:1 at 215.8 and 210.4 ppm, with Rh–C coupling constants of 28.4 and 27.9 Hz, respectively. This complex is stable in the solid state for at least 2 months. In solution it slowly converts to the mono-carbonyl complex [Rh(κ3-bpa)(η1-CO)]PF6 ([3]PF6) (approx. 30% conversion in acetone after 10 days, heating under reflux in acetone for 2 h leads to quantitative


53

conversion) (Scheme 4). Complex 3+ reveals a single CO stretch band at 1989 cm–1 and a 13C NMR resonance at 190.3 ppm with a Rh–C coupling constant of 77.5 Hz in MeCN.

In marked contrast to its bpa analog, reaction of the [(κ3-Me-bpa)Rh(η4-cod)]PF6 complex ([4]PF6) with 1 bar CO does not lead to a binuclear, tris-carbonyl bridged species analogous to 22+. Instead, treatment of [4]PF6 with 1 bar CO in CH2Cl2 results directly in a clean and facile substitution of cod by CO with formation of only the square planar carbonyl complex [(κ3-Me-bpa)Rh(CO)]PF6 ([5]PF6) (Scheme 5). In the conversion of the η4-cod to the monocarbonyl complexes the coordination mode of the Me-bpa ligand changes from fac to mer as evidenced by X-ray diffraction measurements (Figure 2).

N Rh

N

N

N

N

Rh COI

+

Py

+

1 bar CO I

5+4+

Scheme 5. Synthesis of [Rh(κ3-Me-bpa)(CO)]+ ([5]+).

Crystals of [5]PF6 suitable for X-ray diffraction were grown by layering MeCN solution of [5]PF6 with diethyl ether. The structure of 5+ is shown in Figure 2.

Figure 2. Coordination geometry of the rhodium atom in the complex [(κ3-Me-bpa)Rh(CO)]PF6 ([5]PF6).

Hydrogen atoms and the PF6– counter ion are omitted for clarity. Thermal ellipsoids are drawn with 50%

probability. Selected bond lengths [Å] and angles [°]: Rh1–C1 1.834(3), Rh1–N2 2.034(2), Rh1–N1 2.039(2), Rh1–N3 2.079(3), C1–O1 1.144(4), C1–Rh1–N2 98.29(11), C1–Rh1–N1 98.82(11), N2–Rh1–N1 162.85(10), C1–Rh1–N3 177.53(13), N2–Rh1–N3 81.80(10), N1–Rh1–N3 81.16(10).

The substantial different stabilities of the Rh(μ-CO)3Rh bridged complexes with bpa versus Me-bpa is remarkable, considering the structural resemblance of these ligands. Formation of a square planar monocarbonyl rhodium complex was also observed for the Bu-bpa ligand.16 Interestingly, the formation of a Rh(μ-CO)3Rh complex with Bu-bpa could be enforced at higher CO pressures.

Chapter 3

54

Treatment of [Rh(κ3-Bu-bpa)(CO)2]PF6 [6]PF6 with 50 bar CO for approx. 7 days results in partial conversion (approx. 15%) to a mixture of the dinuclear, dicationic complex [Rh2(κ3-Bu-bpa)2(μ-CO)3](PF6)2 ([7](PF6)2) and mononuclear, monocationic complex [Rh(κ3-Bu-bpa)(CO)2]PF6 ([8](PF6)) containing two terminal CO ligands as observed by mass spectrometry and IR spectroscopy (Scheme 6).16,17

N

N

N

Rh CO

N

N

PyRh

N

NRh

Py

O

O O N Rh CO

CO

N

+

6+

50 bar CO

CH3CN

2+

72+

II I

I

+

Py

+

8+

Scheme 6. Reaction of [Rh(κ3-Bu-bpa)(CO)]+ (6+) with 50 bar CO to form [Rh2(κ3-Bu-bpa)2(μ-CO)3]2+ (72+) and [Rh(κ3-Bu-bpa)(CO)2]+ (8+).16

FAB-MS (m/z = 945, 917) and IR spectra (ν(CO)= 1832 cm–1 (KBr)) of 72+ are in accordance with the tris-μ-carbonyl bridged structure while m/z 772 and ν(CO)= 1992 and 2042 cm–1 confirm the presence of the bis-carbonyl complex 8+. Attempts to prepare [7](PF6)2 in high yield were unsuccessful, which might indicate an equilibrium between 72+ and 6+ in the presence of CO. In accordance with this, [7](PF6)2 is not stable in the solid state at room temperature in the absence of CO, and converts to [6]PF6 within days. Heating of the isolated mixture of [6]PF6 and [7](PF6)2 in acetonitrile to 50 ºC for 2 hours results in full conversion of the contained [7](PF6)2 to [6]PF6.

16 Table 1. IR and 13C NMR spectroscopy data of the carbonyl ligands.

Complex ν(C≡O) solution [cm–1]

ν(C≡O) solid state [cm–1]

1J(Rh,C) [Hz]

[Rh2(bpa)2(μ-CO)3]2+ (22+) 1835 (sh. 1828) 1836, 1815 28.4, 27.9

[Rh(bpa)(CO)]+ (3+) 1989 77.5 [Rh(Me-bpa)(CO)]+ (5+) 1996 (DCM) 78.0 [Rh(Bu-bpa)(CO)]+ (6+) 1994 79.1 [Rh2(Bu-bpa)2(μ-CO)3]

2+ (72+) 1832 [Rh(Bu-bpa)(CO)2]

+ (8+) 1992, 2042 [Rh2(tpen)(CO)2]

2+ (122+) 1999 67.5 [Rh2(tppn)(CO)2]

2+ (132+) 1994 79.0 [Rh2(tppn)(μ-CO)3]2

4+ (144+) 1838 (sh. 1828) 29.1, 29.1 [Rh2(tpbn)(CO)2]

2+ 1992 [Rh2(tpbn)(CO)2]n

2+ 1843 [Rh(tpa)(CO)]+ (15+) 1991 1985 79.5 [Rh(tpa)(μ-CO)]2

2+ (162+) 1749 1740 19.6

It is difficult to imagine that substitution of a proton for an alkyl group will have any significant steric influence on the formation of the Rh(μ-CO)3Rh bridge (tethered alkyl substituted bpa complexes do form carbonyl bridged species, vide infra). Therefore, the markedly different stability of the dirhodium triscarbonyl species 22+ and 72+ is most likely caused by the decrease in donor capacity of the ligand on going from bpa to Me-


55

and Bu-bpa. The more electron rich rhodium atom of bpa complex 22+ is expected to bind CO more strongly than the Bu-bpa ligated metal in the complex 72+ which should result in relative stabilization of the (μ-CO)3 bridge in 22+ compared to 72+.

The CO stretch frequencies of the square planar complexes 3+ and 6+ (1989 cm–1 vs. 1994 cm–1 and 1996 cm–1 for bpa, Bu-bpa, and Me-bpa respectively) confirm that Bu-bpa is indeed a weaker donor than bpa (Table 1). Although the difference in stretch frequencies (5 cm–1) does not seem to be large, it is significant and might explain the different relative stability of the bridged carbonyl complexes 22+ and 72+. 3.2.2 Structure of cyclooctadiene and carbonyl rhodium complexes with tethered bpa ligands

Concluding that the Bu-bpa complex 72+ is thermodynamically unstable, we decided to investigate the stabilization of tris-carbonyl bridged species by chelate cooperativity18 of binuclear rhodium bpa species tethered with an alkyl chain. The reduction of the entropy of binding by tethering of the rhodium carbonyl complexes could lead to structures similar to the reported hexamine macrocylic triscarbonyl bridged rhodium complexes.8 For that reason we took a closer look at the corresponding cod and carbonyl complexes with ligands that have bpa moieties tethered by a chain of 2, 3 and 4 carbon atoms, and their reactivity with CO.16

Analysis of the X-ray structures of [Rh2((μ-(bis-κ3)tpen)2(η4-cod)2]2+

(92+), [Rh2((μ-

(bis-κ3)tpbn)(η4-cod)2]2+ 11 (112+) and the DFT optimized structure of [Rh2((μ-(bis-

κ3)tppn)(η4-cod)2]2+ 19

(102+) (tpen = N1,N1,N2,N2-tetrakis(pyridin-2-ylmethyl)ethane-1,2-diamine, tppn = N1,N1,N2,N2-tetrakis(pyridin-2-ylmethyl)propane-1,2-diamine, tpbn = N1,N1,N2,N2-tetrakis(pyridin-2-ylmethyl)butane-1,2-diamine) indicates that the length of the methylene tether connecting bis(picolyl)amine moieties in tpbn, tppn and tpen influences the solid state structure of the binuclear complex. In complexes with 2 or 4 carbon atoms in the tether, the two metal centers are found to be on the opposite sides of the molecule, in a ‘one hand up, one hand down’ configuration. In complex 102+ having a C3 linker the metal centers are on the same side, in a ‘hands up’ configuration (see Figure 3).

Figure 3. Molecular structures of [Rh2((μ-(bis-κ3)tpen)2(η4-cod)2]2+

(92+), [Rh2((μ-(bis-κ3)tppn)(η4-cod)2]2+

(102+) and the previously reported [Rh2((μ-(bis-κ3)tpbn)(η4-cod)2]2+ (112+).

We expect that the same structures should be also favored in solution since the zigzag conformation of the alkyl chain results in minimal steric interactions of the moieties coordinated to the rhodium centre. Above structural features of the tethered bpa-alkyl ligands proved to have an impact on the structure of corresponding rhodium carbonyl complexes (see below).

In contrast to the reaction of [6]PF6, the reaction of [Rh2((μ-(bis-κ3)tppn)(CO)2](PF6)2, ([13](PF6)2) in CH3CN with 50 bar CO is nearly quantitative within 4 days. The

Chapter 3

56

tetranuclear complex [Rh4((μ-CO)3)2((μ-(bis-κ3)tppn)2]4+ [14]4+ is formed, with two tris-

μ-carbonyl bridges (Scheme 7). The ν(CO) bands at 1838 and 1828 cm–1 (CH3CN) and 13C-NMR triplets at δ = 215 (1JC-Rh= 29.1 Hz) and δ = 213 (1JC-Rh= 29.1 Hz) ppm in approx. intensity ratio of 2:1, are indicative for the tris-μ-carbonyl bridges (Table 1). 1H NMR and 13C-NMR signals for the tppn ligand show that [14]4+ has effective D2h symmetry in solution (8 equivalent pyridyl-fragments). The tppn ligand in [14]4+ is fac-coordinated, as indicated by 1H NOESY NMR spectroscopy.16 The DFT optimized structure of [14]4+ is presented in (Figure 4).

NN Py

NN

Py

Rh Rh

NN

Rh

Py

OO

O

NN

Rh

Py

OOON

N

N

Rh CON

N

N

RhOC

4+

50 bar CO

CH3CN

2+

132+ 144+

I I

II

II

Scheme 7. Formation of a tetranuclear rhodium carbonyl complex [Rh4((μ-CO)3)2(tppn)2]4+ (144+).16

Figure 4. DFT optimized structure of 144+.

In contrast to 72+, 144+ is stable at room temperature both in the solid state and in solution. Heating 144+ in CH3CN to 80 ºC results in only approx. 20% conversion to 132+ in 2 hours.16

The stability of the tetranuclear complex 144+ is most likely a result of the cooperative binding of the four rhodium atoms through carbonyl bridges. Formation of the first Rh(μ-CO)3Rh bridge (that is not thermodynamically stable for the binuclear analog of 144+ – the Bu-bpa complex 72+) enhances the effective local concentration of rhodium and entropically favors the formation of the second Rh(μ-CO)3Rh bridge. This effect is similar to the well-known chelate effect that results in stronger binding of multidentate versus monodentate ligands. Consequently, multivalent binding18 (involving the interaction of four Rh atoms) stabilizes 144+ from fragmentation into terminal Rh-CO complexes and results in higher stability of 144+ compared to 72+.


57

Under similar carbonylation conditions the tpen carbonyl complex [Rh2((μ-(bis-κ3)tpen)(CO)2]

2+ (122+) (nor the tpen cod complex 9+) did not form any bridged carbonyl species and only 122+ could be observed in the solution. Although formation of an intramolecular CO bridge should be disfavored because of the preferred ‘one hand up, one hand down’ conformation of the molecule, one could expect that intermolecular CO bridges could be formed, leading to polymeric structures. Formation of the CO bridged species from the Bu-bpa complex 6+ (for which the partial conversion to the bridged species 72+ was observed) and tpen complex 132+ would have approximately the same unfavorable entropy contribution, but the enthalpy of formation of the bridged species is likely even lower for tpen because of electronic effects. The tpen ligand imposes a lower electron density on the metal (ν(CO) = 1999 cm–1, 1JC-Rh= 67.5 Hz) compared to the Bu-bpa or tppn ligands (ν(CO) = 1994 cm–1, 1JC-Rh= 79 Hz) (Table 1), which most probably does not allow for coordination of any extra CO molecule. The lower donor capacity of tpen ligand compared to Bu-bpa can be rationalized by the fact that the amine nitrogens of tpen are rivaling for the electrons from a very short C2 alkyl bridge, which effectively acts as an electron poor “CH2• radical” substituent of the bpa moiety.

Treatment of the [Rh2((μ-(bis-κ3)tpbn)(η4-cod)2]2+ complex with 10 bar of CO at 50

ºC for 10 hours in acetonitrile results in formation of a yellow precipitate. IR measurements (KBr) showed the presence of bridging carbonyls (ν(CO) = 1843 cm–1) and a minor amount of terminal carbonyls (ν(CO) = 1993 cm–1) in the precipitate, while the filtrate after evaporation of the solvent gave signals at 1838 and 1992 cm–1. This could indicate formation of polymeric Rh(μ-CO)3Rh bridged species, which should be driven by precipitation. 3.2.3. Rhodium carbonyl complexes with tris(picolyl)amine

The above results suggest that the formation of the Rh(μ-CO)3Rh bridged species is driven by use of stronger σ-donating N3-ligands and can be further stabilized by entropic factors in case of homoditopic ligands. To further investigate the effect of stronger σ-donors we studied the tetradentate tpa (tpa = tris(picolyl)amine) ligand that can be regarded as a bpa ligand functionalized with an additional picolyl group. This ligand can be compared with our previously reported Me3tpa complexes in its coordinating behavior towards Rh-carbonyl species.20

Reaction of the in situ generated [Rh(μ-Cl)(CO)2]2 with tpa (tpa = tris(picolyl)amine) in methanol yields [Rh(κ3-tpa)(CO)]PF6 ([15]PF6) as a yellow powder after precipitation with KPF6. Solution IR in low concentration shows only one CO absorption band at 1991 cm–1, indicating that the complex has a 16VE square-planar geometry with the tpa ligand being in a κ3-coordination mode. In polar solvents like acetone or acetonitrile, the mononuclear species 15+ is in equilibrium with the dinuclear bis-μ-CO bridged species [Rh(κ4-tpa)(μ-CO)]2

2+ (162+) (Scheme 8) and at higher concentrations a weak IR absorption band at 1749 cm–1 reveals the formation of bridging ketonic carbonyls. After evaporation of the solvent a dark purple solid is obtained, showing an IR (solid state) absorption band of the bridging carbonyls of the dinuclear complex 162+ at νCO = 1740 cm–1 and a terminal carbonyl band at 1985 cm–1 (Table 1). Formation of 162+ is reversible and dissolution of the solid purple mixture of 15+ and 162+ leads to disappearance of the bridging carbonyl band. Such a monomer-dimer equilibrium is remarkable and has not been observed for the related [Rh(κ3-Me3tpa)(CO)]+ complex.20

Chapter 3

58

This different behavior should be ascribed to the stronger donor capacity of tpa versus Me3tpa.

N

N N

NRh

CO

N

NN

NRh

N

N N

NRh

O

O

+

I

2+

I

15+

I

162+

2

Scheme 8. Dynamic equilibrium between the monouclear [Rh(κ3-tpa)(CO)]+ 15+ and dinuclear [Rh(κ4-tpa)(μ-CO)]2

2+ 162+.

The equilibrium between the mononuclear and binuclear complexes could be studied by NMR spectroscopy in polar solvents like acetonitrile or acetone. Both the monomer and the dimer are fluxional on the NMR time scale, involving exchange of the axial and equatorial picolyl moieties. VT-NMR measurements further allowed us to calculate the thermodynamics for dimerization of 15+ in acetone through a van ’t Hoff plot in the range from 283 K to 225 K (Figure 5).

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0.0035 0.0037 0.0039 0.0041 0.0043 0.0045

1/T [K-1]

lnK

Figure 5. Van ’t Hoff plot of the dimerization of 15+ to 162+: ΔH = –28.4 ± 1.7 kJ·mol-1; ΔS = –134 ± 7 J·mol–1·K–1; (r2 = 0.9985). The equilibrium constant is defined as: K = [162+]/[15+]2 .

Formation of binuclear complex 162+ from 15+ is enthalpically favorable by –28 kJ·mol–1, but entropically disfavored. The large negative entropy factor –134 J·mol·K–1 agrees well with a dimerization process and dominates at room temperature. The overall process at 298 K is slightly endergonic (ΔG298 = +11.5 kJ·mol–1, K298 ≈ 9.5 · 10–3).

The 1H NMR spectra recorded in the temperature range from 330 to 218 K are presented in Figure 6. At 330 K only the mononuclear form 15+ is visible (K330 ≈ 3.2 · 10–3). All the picolyl arms of the ligand are equivalent, which indicates that at this temperature the molecule is fluxional on the NMR time scale. This process is frozen at 263 K where the protons of the coordinated and dangling picolyl groups show different signals. We expect that the mechanism of fluxionality of 15+ is the same as for the recently reported [Rh(Me3tpa)CO]+, i.e. the dangling picolyl arm coordinates to the


59

metal forming a transient 18 VE κ4 complex followed by dissociation of another picolyl moiety to reform the 16 VE square planar κ3 complex. 20

Figure 6. Variable temperature 1H NMR spectrum showing dimerization of 15+ and the fluxional behavior of 15+ and 162+ in acetone-d6. Legend: ♦ = 15+; ♠ = 162+

.

The signals of the dinuclear species 162+ start to appear upon cooling the solution

close to room temperature. Both the signals of mononuclear species 15+ and dinuclear species 162+ are substantially broadened at 308 K. Sharpening of the signals of both compounds appears in the same temperature range. This behavior indicates that the broadening is due to coalescence associated with the reversible dimerization of 15+ to 162+. The signals of all three picolyl arms of the dinuclear compound 162+ are equivalent down to 218 K. Clearly the dinuclear complex remains fluxional in the entire measured temperature range. At temperatures below 253 K the signals of the dimer are becoming broader and the signal of Py-H6 completely disappears at 218 K while approaching coalescence.21 Since the fluxional behavior of 162+ down to 218 K is not caused by the dimer-monomer equilibrium, it has to originate from exchange of the picolyl moieties on the octahedral 18VE rhodium centre that is fast on the NMR timescale. Noticeably, the octahedral 18VE complex 22+ does not show a similar fluxional behavior in solution.

♦♦ ♦ ♦ ♦

♠♠♠♠

♦ ♦ ♦ ♦ ♦ ♦♦♦ ♠♠♠♠♦♦♦♠

♦ ♦ ♦ ♦ ♦ ♦♦♦ ♠♠♠♠♦♦♦♠

Chapter 3

60

An important difference between 22+ and 162+ is the presence of a Rh–Rh bond in the former, absent in the latter (see below). We therefore propose that the exchange process of 162+ follows the sequence shown in Figure 7. Decoordination of one of the picolyl arms (PyC trans to PyA) leads to a transient 16VE square pyramidal species that can rearrange the coordinated picolyl groups via a Berry-preudorotation mechanism (via a trigonal bypiramidal structure). Recoordination of the free picolyl group leads to reformation of the octahedral species with a different order of the picolyl groups. DFT calculations suggest that the detachment of the picolyl moiety of 162+ is facilitated by formation of a weak σ-type metal-metal interaction in the ‘unsaturated’ tbpy intermediate. 22+ already has a σ-type Rh–Rh bond and cannot easily increase its bond order and hence the pentacoordinate intermediate required for exchange should be less easily accessible.

Figure 7. Proposed mechanism of exchange of the pyridyl groups of 162+ leading to observed fluxionality of the complex.

Formation of the dinuclear species 162+ was confirmed by single-crystal X-ray diffraction.22 The molecular structure of 162+ is shown in Figure 8. The Rh–Rh distance of 3.0585(7) indicates little or no bonding interaction between the metal atoms,23 in agreement with the saturated 18 VE configuration of the two metals in the absence of a Rh–Rh bond. The Rh–C–Rh angle (100.9(2)°) is considerably larger than the usual 80–90° bond of the ‘classical’ bridging rhodium carbonyls.24 Although the angle is somewhat more acute than most of the reported M−C−M angles for ketonic carbonyls (having more sp2 character of the carbon atom) which are in the range of 107–120°,25 the CO stretching frequency of 1749 cm–1 (MeCN) is in full agreement with a ketonic character of the carbonyl group.25d,e To our best knowledge, complex 162+ is the first example of a binuclear bis-CO bridged rhodium species not supported by any other ancillary bridging ligands, nor a Rh–Rh bond.26 This seems to be also the first example of a dynamic equilibrium between a mononuclear terminal carbonyl RhI species and binuclear Rh(μ-CO)xRh bridged species measured in solution.


61

Figure 8. X-ray structure of [Rh(κ4-tpa)(μ-CO)]22+ 162+ Selected bond lengths [Å] and angles [°]: Rh1–C1

1.972(4), Rh1–C1a 1.993(4), Rh1–N2 2.088(3), Rh1–N1 2.111(3), Rh1–N4 2.201(3), Rh1–N3 2.242(3), C1–O1 1.205(5), N1–Rh1–N4 80.14(12), N1–Rh1–N3 81.07(12), N2–Rh1–N4 79.13(12), N2–Rh1–N3 93.62(12), C1–Rh1–C1a 79.05(18). 3.3 Conclusions

Synthesis of rhodium carbonyl complexes ligated with a series of different bis(2-picolyl)amine derivatives allowed us to study the factors that determine the relative stabilities of terminal mono-carbonyl and bridging tris-carbonyl bridged. Whereas for the relatively strong electron donating bpa ligand the formation of the Rh(μ-CO)3Rh bridge is thermodynamically favorable at 1 bar of CO, weaker donors such as Me-bpa, Bu-bpa or tpen do not allow formation of such species under such conditions. The unfavorable entropy of formation of the carbonyl bridges can be reduced by tethering the bpa moieties with a propylene linker. This allows for cooperative binding of four rhodium centers to assemble a stable tetranuclear compound with two tris-carbonyl bridges.

Similarly, the stronger tris(2-picolyl)amine (tpa) N4-donor allows formation of ketonic bis-carbonyl bridged species 162+ that exist in dynamic equilibrium with the mononuclear mono-carbonyl species 15+ in solution, whereas Rh-carbonyl species with the weaker N4-donor Me3tpa (tris(2,6-lutidyl)amine) exist only in the mononuclear terminal carbonyl form. We thus showed that subtle changes in the ligand structure have a major impact on the stability of the carbonyl bridged compounds compared to their terminal mono-carbonyl analogs. The presented results clearly show that the stability of the bis- and tris-carbonyl bridged structures depends on a delicate balance between favorable enthalpic factors (enhanced with stronger σ-donor ligands) and unfavorable entropic factors (that can be reduced by multinuclear binding using ditopic ligands). 3.4. Experimental Section

General methods

All procedures were performed under N2 using standard Schlenk techniques. Solvents (p.a.) were deoxygenated by bubbling through a stream of N2 or by the freeze-pump-thaw method. The temperature indication r.t. corresponds to ca. 20 ºC. [Rh(κ3-bpa)(η4-cod)]PF6 ([1]PF6), [Rh(κ3-Bu-bpa)(η4-cod)]PF6 ([4]PF6) and [Rh2((μ-(bis-κ3)tpbn)(η4-cod)2](PF6)2 ([11](PF6)2)

11 were

O1

C1

C1a

Rh1

N1

N4N2

N3

Chapter 3

62

prepared according to literature procedures. Synthetic procedures for [Rh(κ3-Bu-bpa)(CO)]PF6

([6]PF6), [Rh2((μ-(bis-κ3)tpen)(cod)2](PF6)2 ([9](PF6)2), [Rh2((μ-(bis-κ3)tppn)(cod)2](PF6)2

([10](PF6)2), [Rh2((μ-(bis-κ3)tpen)(CO)2](PF6)2 ([12](PF6)2), [Rh2((μ-(bis-κ3)tppn)(CO)2](PF6)2 ([13]PF6)2), [Rh2((μ-(bis-κ3)tppn)]{(μ-CO)3)2}(PF6)4 [14](PF6)4 are reported in reference 16. All other chemicals are commercially available and were used without further purification. NMR experiments were carried out on a Bruker DRX300 (300 MHz and 75 MHz for 1H and 13C respectively) and Varian Inova500 (500 MHz and 125 MHz for 1H and 13C respectively). Solvent shift reference for 1H NMR: acetone-d6 δH = 2.05, CD3CN δH = 1.98. For 13C-NMR: acetone-d6 δC

= 29.50, CD3CN δC = 1.28. Abbreviations used are s = singlet, d = doublet, dd = doublet of doublets, t = triplet, p = pentet, m = multiplet and br = broad. The couplings between the protons in the pyridine ring are not fully resolved and hence we use a simplified assignment of the multiplicities of the signals as doublets, triplets and double triplets. Elemental analyses (CHN) were carried out by H. Kolbe Mikroanalytisches Laboratorium (Germany). Solution on a Bruker Vertex 70 FTIR spectrometer. Solid state IR measurements were performed on a Shimadzu FTIR 8400S spectrometer equipped with a Specac MKII Golden Gate Single Reflection ATR system. X-ray diffraction

The structures are shown in Figures 1 and 2 27 which include selected bond distances and angles. The crystal data are shown in Table 2. Crystals were mounted on glass needles. The intensity data of [2](PF6)2 was collected at on a Nonius Kappa CCD single-crystal diffractometer, using Mo Kα radiation and applying φ and ω scan modes. The intensity data were corrected for Lorentz and polarization effects. A semi-empirical multiscan absorption correction was applied (SADABS)28 on [2]. The structures were solved by the PATTY option29 of the DIRDIF program system.30 All nonhydrogen atoms were refined with anisotropic temperature factors. The hydrogen atoms were placed at calculated positions, and refined isotropically in riding mode. [2](PF6)2: The crystal structure analysis was hampered by PF6 disorder and the poor quality of the data. Because of the limited reliability of the higher order data all data above 25° θ were not used in the refinement. The least squares refinement showed a moderate racemic twinning (BASF = 0.113). The difference Fourier map showed a substantial residual density which could not be parameterized, and therefore the SQUEEZE procedure31 was used to account for this electron density. Nevertheless the final difference Fourier map showed rather large residual density, especially close to the rhodium atom. To a large extend this is probably due to absorption effects. The parameter set suggests a ratio of 1 dinuclear Rh-complex : 1.5 PF6 moieties : 2 acetone moieties. Obviously, the correct charge balance of the dinuclear Rh-complex (as confirmed by the analytical and spectroscopic data of [2](PF6)2, see synthesis section below) requires the presence of 2 PF6 moieties per complex, therefore we have to assume that the residual electron density stands for 0.5 PF6 (4 PF6 moieties in the whole unit cell). The SQUEEZE procedure shows 4 voids of 229 Å3 each, containing 52 electrons in the whole unit cell. These voids are centred around positions with y = 0.25 and y = 0.75, so around special positions with a multiplicity of 4, which is in accord with the number of missing PF6 moieties. The number of electrons in these voids reported by the SQUEEZE procedure (52/void) is rather low as a PF6 moiety has 69 electrons. However, the voids are rather large and therefore merge to form channels through the structure running parallel to the c-axis. Measurements on crystals grown from acetone:diethyl ether did not result in improved data, indeed they were worse. However, these data showed indications of the missing PF6 moieties on a general position in these channels, in which case an occupancy factor of 0.5 has to be assumed, together with the presence of some more, unidentified solvent moiety (acetone or diethylether). Despite the problems encountered with this structure we do believe that the structure and geometry of the dicationic dinuclear Rh-complex are correct and reliable. The physical properties given here are based on the presence of 2 PF6 and 2 acetone moieties in the structure.


63

Table 2. Crystallographic data for [2](PF6)2 and [5](PF6)2. [Rh2(κ3-bpa)2(μ-CO)3](PF6)2·2C3H6O [Rh(κ3-bpa-Me)(CO)]PF6

Empirical formula C33H38F9N6O5 P1.50Rh2 C14H15F6N3OPRh Formula weight 1021.97 489.17 Temperature [K] 208(2) 110 Radiation MoKα (graphite mon.) Wavelength [Å] 0.71073 0.71073 Crystal system Tetragonal Monoclinic Space group I -4 2 d P 2 1/c a [Å] 30.368(8) 11.0276(4) b [Å] 30.368(8) 12.3702(4) c [Å] 9.2873(4) 15.6878(4) β [°] 90 126.6770(10) Volume [Å] 8565(3) 1716.34(10) Z 8 4 Density [Mg m-3] 1.585 1.8931(1) R1 0.0877 0.03

DFT calculations

Geometry optimizations were carried out with the Turbomole program package32 coupled to the PQS Baker optimizer33 at the ri-DFT34 level using the BP8635 functional and SV(P) basis set.36 Measurements of the equilibrium between 15+ and 162+.

Measurements were performed on a 0.036 M solution of 15+ in acetone-d6 on a Bruker DRX300 NMR spectrometer. The sample temperature was calibrated by measuring the chemical-shift separation between the OH and CH3 resonances of methanol. Relative concentrations of 15+ and 162+ were estimated by integration of the -CH2- proton signals of the tpa ligand. Syntheses:

[Rh2(κ3-bpa)2(μ-CO)3](PF6)2 ([2](PF6)2): 290 mg of [Rh(bpa)(cod)]PF6 (0.522 mmol) were placed in a 100 ml schlenk flask and dissolved in 15 ml CH2Cl2. CO was bubbled through the solution for 25 minutes during which yellow precipitate has formed. The solution was stirred for additional 30 minutes under CO atmosphere and the yellow solid was filtered and dried in vacuo. Yield: 167 mg (0.170 mmol, 65.4%) 1H NMR (500 MHz, CD3CN): δ = 8.62 (d, 3J(H,H) = 4.5 Hz, 1H; Py-H6); 8.52 (d, 3J(H,H) = 4.5 Hz, 1H; Py-H6’); 7.88 (dt, 4J(H,H) = 1.5 Hz, 3J(H,H) = 8.0 Hz, 2H; Py); 7.46 (m, 3H; Py); 7.38 (t, 3J(H,H) = 6.0 Hz, 1H; Py); 5.72 (t, 3J(H,H) = 4.0 Hz, 1H; NH); 4.87 (dd[AB], 2J(H,H) = 17.0 Hz, 3J(H,H) = 7.0 Hz, 2H; N-CH2-Py); 4.78 (dd[AB], 2J(H,H) = 17.0 Hz, 3J(H,H) = 7.0 Hz, 2H; N-CH2-Py); 4.33 (m, 4H; N-CH2-Py). 13C NMR (125 MHz, CD3CN): δ = 215.8 (t, 1J(C,Rh) = 28.4 Hz, 2xμ2-CO), 210.4 (t, 1J(C,Rh) = 27.9 Hz, μ2-CO); 161.1 (Py-C1); 160.9 (Py-C1’); 149.5 (Py-C5); 149.3 (Py-C5’); 140.8 (Py-C3); 140.7 (Py-C3’); 125.8 (Py); 124.8 (Py); 60.0 (N-CH2-Py); 59.9 (N-CH2-Py’) IR (MeCN): ν(C≡O) = 1835 with a shoulder at 1828 cm–1, solid state: ν(C≡O) = 1836, 1815 cm–1 Elemental Analysis: Calcd. (C27H26F12N6O3P2Rh2·0.5CH2Cl2): C: 32.36; H: 2.67; N: 8.23); Found: C: 32.34; H: 2.97; N: 8.43. [Rh(κ3-bpa)(CO)]PF6 ([3]PF6): 153 mg of [2](PF6)2 (0.156 mmol) were dissolved in 10 ml MeOH and kept under reflux for 3 hrs allowing the evolved CO to escape from the vessel. The unreacted [2](PF6)2 was filtered off and the filtrate was condensed to approx 5 ml causing precipitation of the product that was filtered off, and dried in vacuo. Yield: 59 mg (0.124 mmol, 39.6%) 1H NMR (300 MHz, CD3CN): δ = 8.39 (d, 3J(H,H) = 5.4 Hz, 2H; Py); 7.92 (dt, 4J(H,H) = 1.5 Hz, 3J(H,H) = 7.8 Hz, 2H; Py); 7.34 (d, 3J(H,H) = 8.1 Hz, 2H; Py); 7.30 (t, 3J(H,H) = 6.9 Hz, 2H; Py); 5.57 (s, br, 1H, N-H); 4.60 (d[AB], 2J(H,H) = 15.9 Hz, 3J(H,H) = 9.6 Hz, 2H; N-CH2-Py); 4.46 (d[AB], 2J(H,H) = 15.9 Hz, 3J(H,H) = 6.0 Hz, 2H; N-CH2-Py).

Chapter 3

64

13C NMR (75 MHz, CD3CN): δ=190.3 (d, 1J(Rh,C) = 77.5 Hz; (CO)); 165.2 (Py); 154.6 (Py); 139.8 (Py); 125.7 (Py); 123.4 (Py); 57.8 (N-CH2-Py). IR (MeCN): ν(C≡O)=1989 cm–1 Elemental Analysis: Calcd (C13H13F6N3OPRh): C, 32.86; H, 2.76; N, 8.84; Found: C: 32.93; H: 3.00; N: 8.72. [Rh(κ3-bpa-Me)(η4-cod)]PF6 ([4]PF6): 206 mg of [Rh(cod)(μ-Cl)]2 (0.836 mmol Rh) was suspended in 10 ml MeOH and 179 mg (0.839 mmol) of bpa-Me was added and the solution was stirred until the solution became clear. Next, 210 mg (1.14 mmol) of KPF6 and 2 ml of water were added and the solvent partially evaporated, causing precipitation of a bright yellow solid, which was filtered and washed with 2 ml MeOH. Yield: 222 mg (0.38 mmol, 46.6 %), a second crop of 63 mg was obtained by leaving the filtrate at –20 ºC for 3 days giving combined yield of 285 mg (0.501 mmol, 60%) 1H (300 MHz, acetone-d6): δ = 9.26 (d, 3J(H,H) = 4.8 Hz, 2H; Py); 7.73 (dd, 4J(H,H) = 1.5 Hz, 3J(H,H) = 7.8 Hz, 2H; Py); 7.33 (d, 3J(H,H) = 5.7 Hz, 2H; Py); 7.28 (d, 3J(H,H) = 8.1 Hz, 2H; Py); 4.65 (d[AB] br, 2J(H,H) = 16.2 Hz, 2H; N-CH2-Py); 4.16 (d[AB], 2J(H,H) = 16.2 Hz, 2H; N-CH2-Py); 3.85 (s, br, 4H; CH=CH); 3.66 (s, 3H; N-CH3); 2.67 (m, br, 4H; CH2); 1.87 (m, br, 4H; CH2). 13C (75 MHz, acetone-d6): δ = 161.1 (Py-C2); 153.2 (Py); 140.3 (Py); 126.2 (Py); 124.8 (Py); 76.9 (d, 1J(Rh,C) = 13.6 Hz; (CH=CH)); 67.9 (N-CH2-Py); 53.0 (N-CH3); 32.8 (CH2). Elemental Analysis: Calcd. C: 44.30; H: 4.78; N: 7.38 Found: C: 43.99; H: 4.91; N: 7.14. [Rh(κ3-bpa-Me)(CO)]PF6 ([5]PF6): 173 mg of [Rh(bpaMe)(cod)]PF6 (0.304 mmol) were placed in a 100 ml schlenk and dissolved in 15 ml CH2Cl2. CO was bubbled through the solution for 20 minutes and the solution was stirred for additional 30 minutes under CO atmosphere. 30 ml of hexanes were added causing precipitation of yellow solid that was filtered, washed with hexanes and dried in vacuo. Yield: 137 mg (0.260 mmol, 85.5%) 1H (500 MHz, CD3CN): δ = 8.42 (d, 3J(H,H) = 5.5 Hz, 2H; Py); 7.96 (dt, 4J(H,H) = 1.5 Hz, 3J(H,H) = 8.0 Hz, 2H; Py); 7.51 (d, 3J(H,H) = 7.5 Hz, 2H; Py); 7.35 (t, 3J(H,H) = 7.5 Hz, 2H; Py); 4.87 (d[AB], 2J(H,H) = 15.5 Hz, 2H; N-CH2-Py); 4.19 (d[AB], 2J(H,H) = 15.5 Hz, 2H; N-CH2-Py); 2.69 (s, 3H; N-CH3). 13C (125 MHz, CD3CN): δ = 190.3 (d, 1J(Rh,C) = 78 Hz; (CO));, 163.5 (Py); 155.0 (Py); 140.1 (Py); 126.1 (Py); 124.8 (Py); 66.9 (N-CH2-Py); 46.5 (N-CH3). IR (CH2Cl2) (ν(C≡O)=1996 cm–1). Elemental Analysis: Calcd. C: 34.38; H: 3.09; N: 8.59; Found: C: 34.24; H: 3.26; N: 8.30. Crystals of 4+ suitable for X-ray diffraction were obtained by layering MeCN solution of [4]PF6 with diethyl ether. [Rh(κ3-tpa)(CO)]PF6 ([15]PF6): 197 mg of [Rh(μ-Cl)(coe)2]2 (0.275 mmol) was suspended in 10 ml of MeOH and 167 mg of tpa (0.575 mmol) was added. CO was bubbled until al the solid dissolved and the solution turned yellow-brown. The solution has been stirred for additional 20 minutes under CO atmosphere after which 143 mg of KPF6 (0.781 mmol) was added causing precipitation of yellow solid that was filtered off and washed twice with 2 ml of MeOH. Yield: 213 mg (0.376 mmol). Compound was recrystallized from 10 ml of hot MeOH to yield 140 mg (0.247 mmol, 45.0 %) of analytically pure product. After dissolution of 15+ in acetone, the compound 15+ exists in equilibrium with 162+. Signals of 162+ were omitted for clarity. PyC = coordinated picolyl moiety, PyD = dangling picolyl moiety. 1H NMR (500 MHz, acetone-d6, –5 °C): δ = 8.44 (d, 3J(H,H) = 5.0 Hz, 2H; PyC-H6); 8.35 (d, 3J(H,H) = 3.5 Hz, 1H; PyD-H6); 8.02 (t, 3J(H,H) = 7.5 Hz, 2H; PyC-H4); 7.98 (d, 3J(H,H) = 7.5 Hz, 1H; PyD-H3); 7.65 (t, 3J(H,H) = 7.5 Hz, 1H; PyD-H4); 7.61 (d, 3J(H,H) = 8.0 Hz, 2H; PyC-H3); 7.40 (t, 3J(H,H) = 7.0 Hz, 2H; PyC-H5); 7.14 (t, 3J(H,H) = 6.0 Hz, 1H; PyD-H5); 5.23


65

(d[AB], 2J(H,H) = 15.5 Hz, 2H; N-CH2-PyC); 4.84 (d[AB], 2J(H,H) = 15.5 Hz, 2H; N-CH2-PyC); 4.38 (s, 2H; N-CH2-PyD). 13C NMR (125 MHz, acetone, –27 °C): δ = 190.8 (d, 1J(Rh,C) = 79.5 Hz; (CO)); 165.1 (PyC); 155.3 (PyC); 155.2 (PyD); 154.2 (Py); 151.1 (PyD); 150.9 (PyD); 140.7 (PyC); 138.0 (PyD), 129.5 (Py); 129.3 (py); 66.4 (N-CH2-Py); 65.9 (N-CH2-Py). IR (MeCN): ν(C≡O)=1991 cm–1. Elemental Analysis: Calcd. (C19H18F6N4OPRh) C: 40.30; H: 3.20; N: 9.89; Found: C: 40.63; H: 3.57; N: 9.83. [Rh(κ4-tpa)(CO)]2(PF6)2 ([16](PF6)2): After dissolution of 15+ in acetone, the compound 162+ exists in equilibrium with 15+. Signals of 15+ were omitted for clarity. 1H (500 MHz, acetone-d6, –5 °C): δ = 8.52 (d, 3J(H,H) = 5.0 Hz, 3H; Py-H6); 7.82 (d, 3J(H,H) = 7.5 Hz 4J(H,H) = 1.0 Hz, 3H; Py-H4); 7.46 (d, 3J(H,H) = 7.5 Hz, 3H; Py-H3); 7.25 (t, 3J(H,H) = 6.5 Hz, 3H; Py-H5); 5.08 (s, 6H; N-CH2-Py). 13C (125 MHz, acetone, –60 °C, all peaks broad): δ = 208.3 (t, 1J(Rh,C) = 19.6 Hz; (CO)); 162.5 (Py); 153.2 (Py); 140.4 (Py); 140.1 (Py); 126.4 (Py); 126.1 (Py); 125.2 (Py); 65.7 (N-CH2-Py) IR (MeCN): ν(C≡O)=1749 cm–1. 3.5 Acknowledgements

We thank Jan Meine Ernsting for his assistance with the low temperature NMR experiments. Jan Smits and Maxime Siegler are acknowledged for the X-ray structure determination of complexes 2 and 5 respectively. Ferry van Nisselroij and Caroline Schouten are acknowledged for the synthesis and characterization of compounds 6-14. Charlotte Creusen is acknowledged for the initial studies on complex 15. 3.6 Notes and References

1 For a review on trispyrazolyl borate rhodium complexes see: Slugovc, C.; Padilla-Martínez, I.; Sirol, S.; Carmona, E. Coord. Chem. Rev. 2001, 213, 129–157.

2 (a) Ghosh, C. K.; Graham, W. A. G. J. Am. Chem. Soc. 1987, 109, 4726-4727. (b) Bromberg, S. E.; Yang, H.; Asplund, M. C.; Lian, T.; McNamara, B. K.; Kotz, K. T.; Yeston, J. S.; Wilkens, M.; Frei, H.; Bergman, R. G.; Harris, C. B. Science 1997, 278, 260-263.

3 Webster, C. E.; Hall, M. B. Inorg. Chim. Acta 2002, 330, 268-282 and references therein. 4 For examples see: (a) Connelly, N. G.; Emslie, D. J. H.; Geiger, W. E.; Hayward, O. D.; Linehan, E. B.;

Orpen, A. G.; Quayle, M. J.; Rieger, P. H. J. Chem. Soc., Dalton. Trans. 2001, 670-683. (b) Geiger, W. E.; Camire Ohrenberg, N.; Yeomans, B.; Connelly, N. G.; Emslie, D. J. H. J. Am. Chem. Soc. 2003, 125, 8680-8688.

5 Compounds characterized by X-Ray diffraction: (a) Rheingold, A. L.; Liable-Sands, L. M.; Incarvito, C. L.; Trofimenko, S. J. Chem. Soc., Dalton. Trans. 2002, 2297-2301. (b) Adams, C. J.; Connelly, N. G. ; Emslie, D. J.; Hayward, O. D.; Manson, T.; Orpen, A. G.; Rieger, P. H. Dalton. Trans. 2003, 2835-2845. (c) Rheingold, A. L.; Liable-Sands, L. M.; Golan, J. A.; Trofimenko, S. Eur. J. Inorg. Chem. 2003, 2767-2773. (d) Blake, A. J.; George, M. W.; Hall, M. B.; McMaster, J.; Portius, P.; Sun, X. Z.; Towrie, M.; Webster, C. E.; Wilson, C.; Zarić, S. D. Organometallics 2008, 27, 189-201.

6 Compound characterized by X-Ray diffraction: (a) methyl(trispyrazol-1-yl)gallate: Louie, B. M.; Rettig, S. J.; Storr, A.; Trotter, J. Can. J. Chem. 1984, 62, 633-637. Other examples: (b) tris-(2-pyridyl)-phosphineoxide ligand: Cesares J. A.; Espinet, P.; Martín-Alvarez, J. M.; Espino, G.; Pérez-Manrique, M.; Vattier, F. Eur. J. Inorg. Chem. 2001, 289-296. (c) tris(pyrazol-1-yl)methane: Estruelas, M. A.; Oro, L. A.; Claramunt, R. M.; López, C.; Lavandera, J. L.; Elguero, J. J. Organomet. Chem. 1989, 366, 245-255. (d) tris(pyrazol-1-yl)borate: O’Sullivan, D. J.; Lalor, F. J. J. Organomet. Chem. 1974, 65, C47-49. e) Cocivera, M.; Desmond, T. J.; Ferguson, G.; Kaitner, B.; Lalor, F. J.; O’Sullivan, D. J. Organometallics 1982, 1, 1125. (f) tris(pyrazol-1-yl)methanesulfonate: Kläui, W.; Schramm, D.; Peters, W.; Rheinwald, G.; Lang H. Eur. J. Inorg. Chem. 2001, 1415-1424.

7 Compound characterized by X-ray diffraction: de Bruin, B.; Donners, J. J. J. M.; de Gelder, R. ; Smits, J. M. M.; Gal, A. W. Eur. J. Inorg. Chem. 1998, 401-406.

Chapter 3

66

8 Compounds characterized by X-ray diffraction: (a) macrocyclic hexamines: Lecomte, J.-P.; Lehn, J.-M.; Parker, D.; Guilheim, J.; Pascard, C. J. Chem. Soc., Chem. Comm. 1983, 296-298. (b) Johnson, J. M.; Bulkowski, J. E.; Rheingold, A. L.; Gates, B. C. Inorg. Chem. 1987, 26, 2644-2646. Other representative example: (c) Parker, D. J. Chem. Soc., Chem. Comm. 1985, 1129-1131.

9 Mathieu, R.; Esquius, G.; Lugan, N.; Pons, J.; Ros, J. Eur. J. Inorg. Chem. 2001, 2683-2688. 10 (a) Tejel, C.; Ciriano, M. A.; del Río, M. P.; van den Bruele, F. J.; Hetterscheid, D. G. H.; Tsichlis i

Spithas, N.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 5844–5845. (b) Hetterscheid, D. G. H.; Klop, M.; Kicken, R. J. N. A. M.; Smits, J. M. M.; Reijerse, E. J.; de Bruin, B. Chem. Eur. J. 2007, 13, 3386-3405. (c) de Bruin B.; Verhagen J. A. W.; Schouten C. H. J.; Gal A. W.; Feichtinger D.; Plattner, D.A. Chem. Eur. J. 2001, 7, 416-422.


12 van Leeuwen, P. W. N. M.; Claver C. (Eds.): Rhodium Catalyzed Hydroformylation, Kluwer Academic Publishers: Dordrecht, 2000

13 Formation of μ2-carbonyl bridged rhodium dimers with terminal carbonyls were reported for some hydroformylation catalysts: (a) Wilkinson’s catalyst: Evans, D.; Yagupsky, G.; Wilkinson, G. J. Chem. Soc. A 1968, 2660-2665. (b) Chan, A. S. C.; Shieh, H. S.; Hill, J. R. J. Chem. Soc., Chem. Commun. 1983, 688-689. (c) diphosphines: James, B. R.; Mahajan, S. J. Rettig, G. M. Williams, Organometallics 1983, 2, 1452-1458. (d) Castellanos-Páez, A.; Castillón, S.; Claver, C.; van Leeuwen, P. W. N. M.; de Lange, W. G. J.; Organometallics 1998, 17, 2543-2552. (e) and diphosphites: Buisman, G. J. H.; van der Veen, L. A.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics, 1997, 16, 5681-5687.

14 The weaker donor strength of alkyl-bpa derivatives is caused by steric effects. See ref. 11 15 Heaton, B. T.; Jacob, C.; Sampanthar, J. T. J. Chem. Soc. Dalton Trans. 1998, 1403-1410. 16 (a) de Bruin, B. PhD Thesis, University of Nijmegen, 1999. (b) Dzik, W. I.; Creusen, C.; de Gelder, R.;

Peters, T. P. J.; Smits, J. M. M.; de Bruin, B. Organometallics 2010, 29, 1629-1641 17 Due to the low yield of 62+ and 7+ it was not possible to distinguish between their NMR signals that were

generally overlapping with the signals of 5+. For that reason the structure of 62+ with two CO ligands being trans to two picolyl donors and one trans to two amine donors is tentative and it cannot be ruled out that the actual structure is similar to the structure of 22+ with one CO trans to two picolyl groups and two CO trans to one amine and one picolyl moiety.

18 We use the term chelate cooperativity (as defined by Hunter and Anderson) exchangeably with the multivalency definition by Mulder, Huskens and Reinhoudt. Essentially, both the chelate cooperativity and multivalency describe the same effect of increased local concentration and entropic contributions associated with binding of the tethered receptor to the binding site (a) Mulder, A.; Huskens, J.; Reinhoudt, D. N. Org. Biomol. Chem. 2004, 2, 3409-3424. (b) Hunter, C. A.; Anderson, H. L. Angew. Chem. Int. Ed. 2009, 48, 7488-7499.

19 The X-ray structure of the [Rh2(κ3-tppn)(η4-cod)2](PF6)2 has shown such connectivity, however very large disorder made the structure unsuitable for publication.

20 (a) see Chapter 2 of this thesis. (b) Dzik, W. I.; Smits, J. M. M.; Reek, J. N. H.; de Bruin, B. Organometallics 2009, 28, 1631-1643.

21 Unfortunately, we were not able to record spectra below 218K because of the NMR spectrometer restrictions.

22 Creusen, C. Master’s Thesis, Radboud University Nijmegen, 2005. 23 The longest reported unsupported Rh–Rh bonds do not exceed 2.93-2.94 Å: Chifotides, H. T. Dunbar, K.

R. Multiple Bonds Between Metal Atoms third edition, Springer: New York 2005, 465-589; ed Cotton, F. A.; Murillo, C. A.; Walton, R. A. and references therein.

24 Colton, R.; McCormick, M. J. Coord. Chem. Rev. 1980, 31, 1-52. 25 For examples see: (a) Cowie, M.; Southern, T. G. Inorg. Chem. 1982, 21, 246-253. (b) Cowie, M.;

Vasapollo, G.; Sutherland, B. R.; Ennett, J. P. Inorg. Chem. 1986, 25, 2648-2653. (c) Carmona, D.; Ferrer, J.; Lahoz, F. J.; Oro, L. A.; Reyes, J.; Esteban, M. J. Chem. Soc., Dalton Trans. 1991, 2811-2820. (d) Connelly, N. G.; Einig, T.; Herbosa, G. G.; Hopkins, P. M.; Mealli, C.; Orpen, A. G.; Rosair, G. M.; Viguri, F. J. Chem. Soc. Dalton Trans. 1994, 2025-2039. (e) Tejel, C.; Bordonaba, M.; Ciriano, M. A.; Edwards, A. J.; Clegg, W.; Lahoz, F. J.; Oro, L. A. Inorg. Chem. 1999, 38, 1108-1117.

26 Crystal structures of bis-μ-CO bridged rhodium species supported by Rh-Rh bond (a) Allevi, C.; Golding, M.; Heaton, B.; Ghilardi, C. A.; Midollini, S.; Orlandini, A. J. Organomet. Chem. 1987, 326, C19-C22. (b) Enders, M.; Kohl, G.; Pritzkow, H. J. Organomet. Chem. 2004, 689, 3024-3030. Similar tetranuclear species were also reported: (c) Lahoz, F. J.; Martin, A.; Estruelas, M. A.; Sola, E.; Serrano, J. L.; Oro, L. A. Organometallics 1991, 10, 1794-1799. (d) Cotton, F. A.; Dikarev, E. V.; Petrukhina, M. A. J. Chem. Soc., Dalton Trans. 2000, 4241-4243.


67


28 Sheldrick, G. M. SADABS; University of Göttingen: Germany, 1996. 29 Beurskens, P. T.; Beurskens, G.; Strumpel, M.; Nordman, C.E.; in Patterson and Pattersons, Clarendon,

Oxford, 1987, p. 356. 30 Beurskens, P. T.; Beurskens, G.; Bosman, W. P.; de Gelder, R.; García-Granda, S.; Gould, R. O.; Israël,

R.; Smits, J. M. M.; University of Nijmegen, The Netherlands, 1996. 31 Spek, A.L. PLATON. A Multipurpose Crystallographic Tool. Utrecht University, Utrecht, The

Netherlands, 2003. 32 Ahlrichs, R.; Turbomole Version 5, 2002, Theoretical Chemistry Group, University of Karlsruhe. 33 PQS version 2.4, 2001, Parallel Quantum Solutions, Fayettevile, Arkansas (USA); the Baker optimizer is

available separately from PQS upon request; Baker, J. J. Comput. Chem. 1986, 7, 385-395. 34 Sierka, M.; Hogekamp, A.; Ahlrichs, R. J. Chem. Phys. 2003, 118, 9136-9148. 35 (a) Becke, A. D. Phys. Rev. A, 1988, 38, 3098-3100. (b) Perdew, J. P. Phys. Rev. B 1986, 33, 8822-8824. 36 Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571-2577.

Chapter 4

Base Assisted Selective Reduction of Organometallic RhII and IrII Radicals

Chapter 4

70

4.1 Introduction

Hydrogen atom transfer (HAT) plays an important role in inorganic chemistry and is of fundamental importance in many biochemical transformations and some industrially relevant synthetic reactions. Transfer of a hydrogen atom from the substrate to a metal coordinated oxygen atom is one of the key steps in biotransformations catalyzed by galactose oxidase1, lipoxygenase,2 cytochrome P450,

3 methane monooxygenases4 and class I ribonucleotide reductase.5 HAT reactions are essential for chain transfer during metal-mediated radical polymerization of olefins,6 but can also lead to catalyst deactivation in cyclopropanation reactions.7

Open-shell second and third row transition metal complexes are excellent models to study HAT reactions, because the products are generally well defined NMR-characterizable diamagnetic compounds. This has been exploited by a number of groups over the past years, providing information on the fundamental reactivity of transition metal radical complexes towards Y-H bonds (Y = C, H, Si, S). The pioneering work of the group of Wayland on rhodium and iridium complexes has demonstrated that C–H8 and H–H8c bonds can be broken homolytically between two metalloradicals. Ligand radicals generated on these metals are also capable of abstracting hydrogen atoms from the reaction medium.9 Recently, the group of Grützmacher elegantly presented a galactose oxidase mimic based on an iridium aminyl radical complex.10 The study of hydrogen atom transfer at open-shell second and third row transition metal complexes clearly deserves much more attention as we are only beginning to understand such reactivity.

Recently, our group investigated the reactivity of paramagnetic rhodium and iridium compounds with the general formula [MII(N3-ligand)(cod)]2+ (cod = cis,cis-1,5-cyclooctadiene).

Scheme 1. Hydrogen atom transfer of allylic hydrogen in [MII(N3-ligand)(cod)]2+ complexes (R1, R2 = H or Me, R3 = H or Bn).

These complexes undergo interesting bimolecular HAT reactions, and the rate of the HAT pathway proved to be markedly dependent on the nature of the used bis-picolylamine (bpa) type N3-ligand. Irrespective of the R groups introduced in the bpa moiety (R1, R2 = H or Me, R3 = H or Bn) the paramagnetic complexes spontaneously transform into a 1:1 mixture of [MIII(allyl)(N3-ligand)]2+ and protonated M(cod)


71

complexes (in the form of [MI(olefin)(N3-ligand+H)]2+ for M = Rh or [MIII(H)(olefin)(N3-ligand)]2+ for M = Ir, Scheme 1).11 The rate of decomposition depends strongly on the nature of the substituents and only the complexes with R1, R2 = Me, R3 = Bn were stable enough to be isolated and fully characterized. The low stability of complexes with secondary amine bpa-type ligands (R3 = H) raises the question if the decomposition to the allyl and hydride complexes could proceed via the involvement of aminyl radicals (Scheme 2).

Scheme 2. Possible hydrogen atom transfer of the allylic hydrogen of the cod ligand via a putative aminyl radical species.

The possible involvement of aminyl radicals in the [MII(cod)(N3-ligand)]2+ system seems plausible. Aminyl radicals can be stabilized by coordination to a metal centre and several examples of such stable species have been reported.12,13 Reversible H-atom transfer (HAT) from a nitrogen atom of transition metal complexes with ligands possessing an amine functionality can occur and this has been extensively studied in the last decade by the group of Mayer.14 Recently, Grützmacher et al. showed that amine-diolefin RhI and IrI complexes can form stable, d8 transition metal aminyl radical complexes upon deprotonation of the amine and subsequent one-electron oxidation.10,15

These aminyl radical complexes abstract hydrogen atoms from activated Y-H bonds. Moreover, bispicolylamine (bpa) is also prone to single and double deprotonation when coordinated to rhodium or iridium diolefin species,16 and can subsequently form transient ligand16a,c,d centered radical species upon one electron oxidation.

To study the possible aminyl radical reactivity of the bpa-type ligand we need a diene ligand that is more robust than cyclooctadiene (innocent towards allylic C-H bond activation or C-C coupling). Therefore we decided to synthesize rhodium and iridium complexes with dibenzo[a,e]cyclooctatetraene (dbcot) as the supporting diolefin.17 Since the steric shielding of the metal centre proved to have a positive effect on the stability of [MII(cod)(N3-ligand)]2+ complexes, we chose bis-lutidylamine (bla) as the N3-donor ligand.

Chapter 4

72

4.2 Results and discussion

4.2.1 Synthesis and characterization of [MI(bla)(dbcot)]+ complexes

Complexes [Rh(η4-dbcot)(κ3-bla)]PF6 ([1]PF6) and [Ir(η4-dbcot)(κ3-bla)]PF6 ([2]PF6) (dbcot = dibenzo[a,e]cyclooctatetraene, bla = bis-N,N-(6-methyl-2-pyridylmethyl)amine) were synthesized from the bla ligand and the corresponding dinuclear [M(μ-Cl)(η4-dbcot)]2 compounds in methanol followed by precipitation as a PF6

– salt by adding KPF6 (Scheme 3).

N

NH

NMCl

+ MN

N

NH

+

-Cl-

1+: M = Rh2+: M = Ir

Scheme 3. Synthesis of [Rh(η4-dbcot)(κ3-bla)]+ (1+ ) and [Ir(η4-dbcot)(κ3-bla)]+ (2+).

[Rh(μ-Cl)(η4-dbcot)]2 used for the synthesis of 1+ could be generated in situ from [Rh(μ-Cl)(η2-coe)2]2 (coe = cis-cyclooctene) and dbcot. A similar approach to synthesize the iridium complex 2+ failed and resulted in unexpected formation of a new complex [Ir(κ3-bla)(H)(σ-C8H13)(η2-C8H14)]PF6 ([3]PF6) by oxidative addition of a vinylic C-H bond of coe to the iridium metal centre (Scheme 4). Therefore the successful synthesis of [2]PF6 required the use of the isolated [M(μ-Cl)(η4-dbcot)]2.

N

NH

NIrCl

+ IrN

N

NH

H

+

KPF6

-KCl

3+ Scheme 4. Synthesis of [Ir(κ3-bla)(H)(σ-C8H13)(η2-C8H14)]+ (3+).

Oxidative addition of the vinylic C–H bond of cyclooctene was previously reported for iridium complexes with HB(Pz)3

18 and Cn19 N3-supporting ligands (HB(Pz)3 = tris(pyrazolyl)borate, Cn = 1,4,7-triazacyclononane). Such reactivity is in line with comparable electronic properties of tris(pyrazolyl)borate and bis-lutidylamine type ligands, which resulted in similar reactivity of the iridium biscarbonyl complexes towards nucleophiles.20 Other electron rich (PNP) tridentate ligands also allow for such reactivity on the iridium centre.21 X-ray quality crystals of [3]PF6 were grown by layering an acetone solution of [3]PF6 with hexanes (Figure 1). The structure of 3+ is noteworthy and is an interesting example of a metal-olefin-hydride complex that does not undergo spontaneous olefin insertion into the M–H bond in coordinating solvents (e.g. acetone). [Ir(κ3-bla)(H)(σ-C8H13)(η2-C8H14)]PF6 has a distorted octahedral geometry with the bla ligand being coordinated in a fac-mode. The most sterically hindered position trans to the N3 amine nitrogen is occupied by the least sterically demanding hydride anion. The cyclooctenyl group coordinates trans to the N1 lutidyl


73

nitrogen and its C9–C10 bond has a clear double bond character (1.326(4) Å). This vinylic double bond has no significant interaction with the metal. The cyclooctenyl group has a large trans influence on the lutidyl N1 atom, which results in a rather large ( > 2.23 Å) Ir–N1 distance. The cyclooctene ligand coordinates trans to the N2 lutidyl nitrogen. As expected for an olefin coordinated to an electron deficient metal, the double bond of the cyclooctene ligand does not reveal substantial elongation.

Figure 1. X-ray structure of [Ir(H)(σ-C8H13)(η2-C8H14)(κ3-bla)]+ (3+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms and the PF6

– counterion were ommited for clarity). Selected bond distances (Å) and angles (°): Ir1–N1 2.239(3); Ir1–N2 2.136(2); Ir1–N3 2.174(3); Ir1–C1 2.200(3); Ir1–C2 2.168(3); Ir1–C9 2.041(3); Ir1–H1 1.490(3); C1–C2 1.387(5); C9–C10 1.326(4); N1–Ir1–N3 79.31(10); N1–Ir1–N2 82.12(10); N2–Ir1–N3 78.12(9); C1–Ir1–C2 37.02(12); Ir1–C9–C10 123.04(2); N1–Ir1–C9 167.63(11); N2–Ir1–C9 88.13(11); N3–Ir1–C9 91.29(11); N3–Ir1–H1 170.0(11); N2–Ir1–H1 92.2(11); N1–Ir1–H1 97.0(11); H1–Ir1–C9 91.0(11)

Single crystals of [1]PF6 and [2]PF6 suitable for X-ray structure determination were grown from acetone solutions layered with hexanes or diethyl ether respectively. The two compounds are isostructural in the solid state (Figure 2) and are best described as distorted trigonal bipyramids (tbpy). The axial positions of the trigonal bipyramid are occupied by the lutidyl (N1) nitrogen atom and a double bond of the dbcot ligand (C5–C6). The M–N1 distance is the same for both complexes within the experimental error while the axially coordinated double bond binds more strongly to iridium (resulting in shorter M–C5 and M–C6 distances and longer C5–C6 distance in 2+ compared to 1+). The equatorial sites are coordinated with lutidyl N2, amine N3 and the C1–C2 double bond of dbcot. For both complexes there is no significant difference between their M–N2 and M–N3 distances, however these nitrogen atoms are bound more strongly to iridium with on average 0.03 Å shorter M–N bonds for 2+ compared to the rhodium analogue 1+. The axially coordinated lutidyl moiety (N1) is bound more strongly to the metal than the equatorially bound N2 and N3 donors, as expected for a d8 tbpy complex.22 Due to stronger π-back bonding in the equatorial plane of tbpy complexes,22

Chapter 4

74

the C1–C2 double bond is bound stronger compared to the axially coordinated C5–C6 bond. The metal C1–C2 interaction has a significant metalla(III)cyclopropane character, resulting in a remarkable elongation of the C1–C2 distance (1.32, 1.44 and 1.47 Å for free dbcot,23 1+ and 2+ respectively). This substantial elongation of the double bonds of coordinated dbcot is indicative of strong π-acidity of this ligand.

Figure 2. X-ray structure of [Rh(η4-dbcot)(κ3-bla)]+ (1+) (left) and [Ir(η4-dbcot)(κ3-bla)]+ (2+) (right). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms, the PF6

− counter ions and solvent molecules are omitted for clarity). The crystal of 1+ reveals the presence of two independent but isostructural cations in the asymmetric unit. See Table 1 for selected bond distances and angles. Table 1. Selected bond distances (Å) and angles (°) for complexes 1+, 2+ and 22+.

[Rh(κ3-bla)(η4-dbcot)]+ [Ir(κ3-bla)(η4-dbcot)]+ [Ir(κ3-bla)(η4-dbcot)]2+

1a+ 1b+ 2+ 2a2+ 2b2+

M–N1 2.116(2) 2.113(2) 2.114(4) 2.111(3) 2.093(3)

M–N2 2.257(3) 2.259(3) 2.229(3) 2.095(3) 2.112(3)

M–N3 2.261(3) 2.249(3) 2.224(4) 2.184(3) 2.185(3)

M–C1 2.103(3) 2.101(3) 2.066(4) 2.189(3) 2.189(4)

M–C2 2.064(3) 2.062(3) 2.130(5) 2.159(3) 2.182(4)

M–C5 2.139(3) 2.147(3) 2.128(4) 2.191(3) 2.165(3)

M–C6 2.155(4) 2.155(4) 2.097(4) 2.184(4) 2.183(4)

C1–C2 1.440(5) 1.436(5) 1.472(6) 1.414(5) 1.410(5)

C5–C6 1.397(5) 1.400(5) 1.411(6) 1.390(5) 1.408(5)

C1–M–C6 40.42(13) 40.34(13) 41.41(17) 37.94(12) 37.65(13)

C2–M–C5 37.96(14) 37.99(14) 38.72(17) 37.05(13) 37.77(13)

N1–M–N2 88.97(9) 89.12(10) 91.48(12) 92.17(11) 89.91(11)

N1–M–N3 75.63(10) 75.00(9) 75.09(13) 76.57(11) 79.80(11)

N2–M–N3 76.18(10) 76.46(10) 76.55(12) 79.97(11) 76.93(12)


75

In solution 1+ is fluxional on the NMR time scale, leading to averaged 1H and 13C NMR signals for both lutidyl groups of the bla ligand and averaged signals of the two olefinic HC=CH moieties and the aromatic rings of the dbcot ligand. Analogous behavior was observed for 2+. Strong π-back bonding from the metal to the dbcot ligand results in substantial upfield shifts of both the proton and carbon resonance frequencies (6.78, 4.50, and 4.37 ppm in 1H and 133.2, 74.6 and 58.3 ppm in 13C for free dbcot, 1+ and 2+ respectively). Thus, both the X-ray and NMR analyses confirm a very strong π-bonding of dbcot to the metal.

The redox properties of 1+, 2+ were investigated using cyclic voltammetry. All complexes reveal reversible one-electron oxidation waves in the scan range between 10 and 200 mV/s. The redox potential of the bla rhodium complex is higher by 47 mV compared to the iridium analogue, which is in line with previous results showing that a variety of [Ir(cod)(N3-ligand)]+ complexes are oxidized at lower potentials than their rhodium analogs (E1/2 lowered by 70-100 mV in those cases).11 Table 2. Electrochemical data for complexes 1+ and 2+ and 4+ (vide supra). Solvent: CH2Cl2; Scan rate = 40 mV/sec; E1/2 vs. Fc/Fc+.

Complex Ea (V) ΔE (mV) E1/2 (V) If/Ib

[Rh(η4-dbcot)(κ3-bla)]PF6 ([1]PF6) 0.410 101 0.359 1.0 [Ir(η4-dbcot)(κ3-bla)]PF6 ([2]PF6) 0.364 104 0.312 1.0 [Rh(η4-dbcot)(κ3-Bn-bla)]PF6 ([4]PF6) 0.477 70 0.443 1.0 [a]ΔE = peak separation, Ea = anodic peak potential, E1/2 = half-wave potential, If/Ib = anodic peak current/cathodic peak current. 4.2.2 Synthesis and characterization of paramagnetic [MII(dbcot)(bla)]2+ complexes

In contrast to the abundant chemistry of d8-metal olefin complexes, fully characterized, stable d7 rhodium(II) and iridium(II) olefin compounds are rather scarce.24 This is caused by the general low stability of organometallic radicals, coupled with relative facile allylic C–H bond activation11,25 and Calkene–Calkene coupling26 reactions triggered by the open-shell metal centre. We expected that the use of the robust dbcot diolefin should block such undesired decomposition pathways involving the metal centre and allow for exclusive reactivity on the nitrogen atom of the coordinated bis-lutidylamine.

Scheme 5. Synthesis of open shell species [Rh(η4-dbcot)(κ3-bla)]2+ (12+ ) and [Ir(η4-dbcot)(κ3-bla)]+ (22+).

The electrochemical data suggest that stable d7 MII complexes should be obtainable from 1+ and 2+ by one-electron oxidation. Hence we investigated their chemical oxidation in solution. Treatment of 1+ and 2+ with appropriate oxidants27 led to formation of 12+ and 22+, respectively (Scheme 5). Compound 22+ proved to be

Chapter 4

76

sufficiently stable to obtain X-ray quality crystals by layering an acetone solution of 22+ onto dichloromethane at –20 °C.

Figure 3. X-ray structure of [Ir(η4-dbcot)(κ3-bla)]2+ (22+). Thermal ellipsoids are drawn with 50% probability (hydrogen atoms, the PF6

− counter ions and a disordered CH2Cl2 molecule are omitted for clarity). The crystal contains two independent cations in the asymmetric unit, only one of them is shown. See Table 1 for selected bond distances and angles.

One-electron oxidation of 2+ to 22+ results in a change of the coordination geometry

from trigonal bipyramidal to a distorted square pyramid (sqpy) with the two lutidyl donors and the two olefinic double bonds coordinated in the basal plane and the amine at the apical position, as evidenced by the single crystal X-ray diffraction. The Ir–N1 and Ir–N2 distances of 22+ are not significantly different (Δr < 3σ). The same is observed for the distance between iridium and the olefinic bonds. Lower electron density on the metal after oxidation results in increased Ir–N and decreased Ir–olefin interactions as compared to 2+. The N1, N2 atoms and the centroids of the double bonds


77

are not in one plane, due to a slight ‘twist’ of the dbcot moiety (Figure 3). The measured crystal contained two independent molecules: the first isomer has the dbcot moiety twisted clockwise, whereas the other counterclockwise. Optimization of the X-ray structure using DFT resulted in convergence to a structure where N1, N2 and the centroids of the double bonds are aligned in a plane. The slightly ‘twisted’ geometry thus seems to be a result of crystal packing forces.

The observed change of the coordination geometry from tbpy to sqpy is more frequently observed when the electron configuration of a transition metal changes from d8 to d7, and is most likely a result of the Jahn-Teller effect.28 For that reason we expect that compound 12+ has a structure similar to 22+.

We used EPR spectroscopy to further investigate the (electronic) structure of 12+ and 22+. Measured X-band EPR spectra of 12+ and 22+ could be simulated as rhombic spectra (i.e. g1 ≠ g2 ≠ g3) with well resolved (super)hyperfine interactions with the metal centre and a single nitrogen atom along the g3(z) axis (Figure 4 and Table 3). The measured g values are indicative of metal-centered radicals. As expected for heavier transition metals with larger spin-orbit couplings, the g-anisotropy (rhombicity) of the iridium complex 22+ is larger than that of the rhodium analogue 12+.

2900 3000 3100 3200 3300 3400

sim

g-value

dX''/

dB

B [Gauss]

exp

2.3 2.2 2.1 2

2250 2500 2750 3000 3250 3500 3750

sim

g-value

dX

''/d

B

B [Gauss]

exp

2.8 2.6 2.4 2.2 2 1.8

Figure 4. Experimental and simulated X-band EPR spectra of [RhII(dbcot)(bla)]2+ (frequency: 9.379235 GHz; modulation amplitude: 0.8 Gauss; power: 0.02 mW) complex 12+ (left) and [IrII(dbcot)(bla)]2+ (frequency 9.377560 GHz; modulation amplitude: 4 Gauss; power: 0.2 mW) complex 22+ (right) recorded in frozen acetone at 60K. Arrows indicate the forbidden transitions. The salt [n-Bu4N]PF6 was added to the solution to obtain a better glass. The spectrum of 42+ is similar to the spectrum of 12+.

The spectrum of the iridium(II) compound 22+ is influenced by the presence of large Ir-quadrupole interactions causing the appearance of two weak ‘forbidden’ transitions as indicated by the arrows in Figure 4.11,29 Since the spectrum does not reveal any resolved hyperfines in this region of the spectrum that could be influenced by these quadrupole interactions we did not take these interactions into account in the spectral simulations.

To get a better insight in the electronic structure of both complexes we calculated the EPR properties with DFT (Table 3). In line with the experiments, DFT predicts that 12+ and 22+ are metal centered radicals with Mulliken spin densities of 72% and 73% on rhodium and iridium respectively, and 16.5 % on the amine nitrogen atom.

Chapter 4

78

Table 3. Simulated (exp) and calculated (DFT) g-values and (super)hyperfine interactions (MHz) for compounds 12+, 22+ and 42+ (n.r. = not resolved). Directions of the principal axes of the g tensor in the molecular axis system of complexes are shown.

g11 (x) g22 (y) g33 (z) Complex Propertyexp DFT exp DFT exp DFT

12+ g-value 2.181 2.133 2.158 2.118 2.000 1.982 HFIRh n.r. -7.67 n.r. -14.2 54.0 -74.6 HFIN n.r. 36.8 n.r. 36.9 64.0 62.122+ g-value 2.368 2.320 2.294 2.257 1.942 1.863 HFIIr n.r. 97.0 n.r. 103 122.0 167 HFIN n.r. 38.4 n.r. 38.5 60.0 63.2

M

N

N

NH

g11(x)

g33(z)

g22(y)

42+ g-value 2.211 2.128 2.184 2.122 1.996 1.982 HFIRh n.r. -8.67 n.r. -16.9 58.9 -70.7 HFIN n.r. 34.1 n.r. 34.2 61.8 63.3

The radical species 12+ and 22+ are sterically shielded around the metal, which is

reflected by their relative inertness towards dioxygen in solution. Bubbling air through the solutions of 12+ or 22+ did not give rise to the formation of superoxo species nor to a noticeable decrease of the EPR signal. Apparently, shielding of the metal centre by the methyl groups of the bla ligand prevents any rapid radical type reactivity at the metal centre. However, somewhat surprisingly, in solution the complexes slowly convert to the one-electron reduced diamagnetic compounds 1+ and 2+, as evidenced by NMR spectroscopy.30 This process is not a disproportionation reaction (as commonly observed for RhII and IrII complexes), because the one-electron reduced species 1+ and 2+ are the only observed products.

Our initial hypothesis was that the increased charge of the metal (changing from 1+ to 2+ upon one electron oxidation) results in an increased acidity of the NH group of the bla ligand. Spontaneous deprotonation of the amine group would then lead to formation of an aminyl radical via an intramolecular redox reaction between the metal and the nitrogen atom. Subsequently, this aminyl radical could then abstract a hydrogen atom from the reaction medium, thus reforming the reduced complex 1+ or 2+ (Scheme 6). To check whether the re-formation of compounds 1+ and 2+ indeed proceeds via the pathway shown in Scheme 6, we studied the reactivity of 12+ and 22+ towards Brønsted bases. We also performed radical trapping experiments.

Scheme 6. Hypothetical deprotonation-HAT pathway leading to net reduction of the paramagnetic species 12+ and 22+.

Reaction of complexes 12+ and 22+ with solid K2CO3 in acetone or acetonitrile resulted

in disappearance of the characteristic EPR signals and no signal of the putative aminyl radical could be detected.31 Instead, quantitative formation of the one-electron reduced diamagnetic complexes 1+ and 2+ was observed with 1H NMR. Remarkably, the NMR


79

spectroscopy data further reveal that the reactions retain the N-H hydrogen atoms for a large part whereas formation of the N-D deuterated analogs of 1+ and 2+ would be expected if the reactions would proceed via the pathway depicted in Scheme 6. This makes the aminyl radical pathway rather doubtful, unless the reaction would be associated with a very large kinetic isotope effect.32 Radical trapping experiments33 did not allow for unambiguous detection of the proposed aminyl radicals, nor of solvent radicals.34

4.2.3 Synthesis of paramagnetic N-protected [RhII(Bn-bla)(dbcot)]2+ complex and its reactivity with K2CO3

Since the above results are inconclusive about the role of the amine/aminyl radical in the observed reduction of 12+/22+ to 1+/2+, we decided to perform a control experiment using a similar compound in which the amine is protected with a benzyl moiety. Hence we investigated the reactivity of the complex [Rh(dbcot)(Bn-bla)](PF6)2 ([4](PF6)2).

The precursor complex [Rh(η4-dbcot)(κ3-Bn-bla)]PF6 ([4]PF6) was prepared using the method described for [1]PF6 (Scheme 7). The 1H NMR signals of the dbcot moiety in compound 4+ are broadened compared to the signals of Bn-bla at room temperature and contrary to 1+ and 2+ the olefinic signals of dbcot are magnetically inequivalent giving two broad signals at δ = 4.75 and 4.24 ppm. This suggests that the rotation of dbcot is much slower in case of 4+ than in 1+ and 2+.

Scheme 7. Synthesis of [Rh(η4-dbcot)(κ3-Bn-bla)]+ (4+).

CV measurements reveal that the redox potential of the [Rh(Bn-bla)(dbcot)]+ (4+) is higher by 67 and 170 mV compared with [Rh(bla)(dbcot)]+ (1+) and the previously reported [Rh(Bn-bla)(cod)]+ 11 respectively, showing that exchange of either the bla ligand for Bn-bla ligand or cod for dbcot results in a considerable destabilization of the +II oxidation state of rhodium (see Table 2). For that reason we used AgPF6 in DCM as the oxidizing reagent to form 42+.

The EPR spectrum of 42+ is very similar to the spectrum of 12+ showing a rhombic, almost axial g-tensor, and hyperfine couplings with rhodium and nitrogen (see Table 3). As for 12+ and 22+ the DFT calculated EPR parameters are in good qualitative agreement with the experiment. The DFT calculated Mulliken spin densities at rhodium (67%) and nitrogen (18%) are comparable to those of 12+ and 22+.

Stirring an acetone-d6 solution of 42+ with solid K2CO3 again results in a color change from dark green to bright yellow, with quantitative formation of the reduced species 4+, (as evidenced by NMR spectroscopy). This is similar to the reactions observed for 12+ and 22+ and shows that the presence of the N-H group is not essential for the reduction of the MII oxidation state radical species to their MI precursors. Hence, the reaction does not necessarily involve the intermediacy of aminyl radicals and the role of the Brønsted base has to be different than originally anticipated.

Chapter 4

80

Base assisted oxidation of solvent35 seems to be a viable pathway for the reduction of species 12+/22+ (Scheme 8). Judging from the redox potentials, the dicationic radical metal complexes 12+/22+ and acetone must exist in a redox equilibrium with the closed shell metal complexes 1+/2+ and the acetone radical cation.36 Without a base, this equilibrium lays far to the left, but the acetone radical cation formed in small quantities can be easily deprotonated by the base, shifting the redox equilibrium between acetone and 12+/22+ completely to the right. Alternatively, in the presence of small quantities of water the hydroxide anion can be generated, which is known as a potential one-electron reducing agent in organic solvents.37 Thus, depending on the water content in the solvent (or base) this seems to be a plausible mechanism as well.

Scheme 8. Regeneration of the diamagnetic complex by base promoted oxidation of solvent or hydroxide anion.

At this point we cannot exclude that the pathway involving deprotonation of the N-H group can take place, taking into account the reactivity of related amine complexes of rhodium and iridium. Instant hydrogen atom abstraction by an iridium coordinated aminyl radical has been recently reported by Grützmacher et al.15c However, it is tempting to assume that base assisted solvent oxidation is the actual mechanism of reformation of 1+/2+ from 12+/22+ and that the aminyl radical pathway plays little (if any) role in this reaction. 4.3 Summary and conclusions

Open shell rhodium(II) and iridium(II) complexes with dibenzocyclooctadiene and bislutidylamine (bla) type ligands were successfully synthesized. They represent rare examples of isolable open-shell olefin complexes. The chosen diolefin ligand is inert toward radical reactions, which allowed for investigations of possible radical reactivity pathways involving the aminyl nitrogen atom of the bla ligand.

An attempt to obtain the aminyl radical complexes by deprotonation the N-H group of the bla ligand using a base led to an unexpected, selective reduction of the paramagnetic species [MII(dbcot)(bla)]2+ to their one-electron reduced closed-shell analogs [MI(dbcot)(bla)]+ (M = Rh, Ir). Control experiments with the rhodium(II) complex [RhII(dbcot)(Bn-bla)]2+ containing the Bn-bla ligand (a bla derivative having the amine protected with a benzyl group) revealed similar reactivity.

These results show that the generation of aminyl radicals from the complexes of the general formula [MII(diolefin)(N3-ligand)]2+ (M = Rh, Ir) is perhaps not occurring at all. At least it is rather unlikely that the observed reduction reactions occur via deprotonation of the amine followed by HAT. It is remarkable, however, that the


81

complexes studied are quantitatively reduced in the presence of bases to the MI species. The reduction is most likely caused by a hydroxide anion or a solvent molecule that can be deprotonated in its oxidized form. Hence the [Mn(dbcot)(N3-ligand)]n+ (M = Rh, Ir; n = 1, 2) complexes seem to be stable redox transfer agents, perhaps allowing oxidation of organic compounds under basic conditions. 4.4 Experimental

X-ray diffraction

The structures are shown in Figures 1, 2 and 3.38 The crystal data are shown in Table 4. Xray data were collected at low temperature under the control of the Nonius COLLECT software with a Nonius KappaCCD on rotating anode. The intensity data were corrected for absorption with the program SADABS. The structures were solved with the program DIRDIF and refined with SHELXL97. The Flack parameter for [1]PF6 was refined to 0.484(13) with a BASF/TWIN refinement. The structure of [2](PF6)2 has solvent accessible voids filled with disordered solvent. Their contribution to the structure factors in the refinement was taken into account with the PLATON/SQUEEZE approach. The hydrogen atom on Ir for [3]PF6 was located in a difference density map. Table 4. Crystallographic data for [1](PF6), [2](PF6), [2](PF6)2, and [3](PF6).

[Rh(κ3-bla)(dbcot)]PF6 · CH3COCH3 ([1]PF6):

[Ir(κ3-bla)(dbcot)]PF6 · (CH3CH2)2O ([2]PF6):

[Ir(κ3-bla)(dbcot)](PF6)2 · CH2Cl2 ([2](PF6)2):

[Ir(κ3-bla)(H)(σ-C8H13)(η2-C8H14)]PF6

(3[PF6]) Crystal color yellow yellow black pale yellow Empirical formula C30H29N3Rh, C3H6O, F6P C30H29IrN3, C4H10O, F6P 2(C30H29IrN3), 4(F6P),

CH2Cl2 C30H45IrN3, F6P

Formula weight 737.52 842.87 1912.37* 784.88 Temperature [K] 110 110 110 110 Radiation MoKα MoKα MoKα MoKα Wavelength [Å] 0.71073 0.71073 0.71073 0.71073 Crystal system Monoclinic Monoclinic Monoclinic Monoclinic Space group C c P 21/c P 21/c P 21/c a [Å] 28.5187(8) 9.2911(4) 21.5483(9) 9.5523(7) b [Å] 11.2036(3) 20.4418(9) 16.7416(7) 16.2517(5) c [Å] 19.4833(4) 17.3595(7) 20.9573(8) 20.1842(7) β [°] 101.351(1) 98.599(2) 110.520(2) 97.861(2) Volume [Å3] 6103.4(3) 3260.0(2) 7080.7(5) 3104.0(3) Z 8 4 4 4 Density [Mg m-3] 1.605 1.7173(1) 1.7939(1)* 1.6796(2) Absorption coefficient [mm-1]

0.682 4.212 4.027* 4.414

Theta-max 27.5 27.5 27.5 27.5 F(000) 3008 1672 3728* 1568 R1 0.0281 0.0333 0.0292 0.0234 wR2 0.0600 0.0584 0.0759 0.0496 S 1.047 1.093 1.101 1.027 * Starred items are excluding the disordered solvent contribution.

DFT calculations: Geometry optimizations were carried out with the Turbomole program package39 coupled to the PQS Baker optimizer40 at the ri-DFT41 level using the BP8642 functional and SV(P) basis set.43 Calculated EPR spectra44 were obtained with ADF program45 at the DFT, BP86,46 TZP47 level, using the Turbomole optimized geometries. General procedures: All manipulations were performed in an argon atmosphere by standard Schlenk techniques or in a glovebox. Methanol and dichloromethane were distilled under nitrogen from CaH2. Hexanes were distilled under nitrogen from Na wire. Acetone was deoxygenated using freeze-pump-thaw method. NMR experiments were carried out on a Bruker DRX300 (300 and 75 MHz for 1H and 13C respectively) on a Varian Inova 500 (500 and 125 MHz for 1H and 13C

Chapter 4

82

respectively) spectrometers. Solvent shift reference: acetone-d6 δ = 2.05 and δ = 29.84 for 1H and 13C, CDCl3 δ = 7.26 and δ = 77.23 for 1H and 13C respectively. Abbreviations used are s = singlet, d = doublet, t = triplet, m = multiplet, br = broad. Elemental analyses (CHN) were carried out by H. Kolbe Mikroanalytisches Laboratorium (Germany). X-band EPR spectroscopy measurements were performed with a Bruker EMX Plus spectrometer. Cyclic voltammograms of ~2 mM parent compounds in 1 M Bu4NPF6 electrolyte solution were recorded in a gas-tight single-compartment three-electrode cell equipped with platinum working electrode (apparent surface of 0.42 mm2), coiled platinum wire auxiliary, and silver wire pseudoreference electrodes. The cell was connected to a computer-controlled PAR Model 283 potentiostat. All redox potentials are reported against the ferrocene/ferrocenium (Fc/Fc+) redox couple. Ferrocene was used as internal standard. Magnetic susceptibility was measured on a Magway MSB Mk1 magnetic balance. [Fe(μ-C5H4COMe)Cp]PF6,

27 [Ir(dbcot)(μ-Cl)]2,48 [M(coe)2(μ-Cl)]2,

49 dibenzo[a,e]cyclooctatetraene,50 N-Benzyl-N,N-di[(6-methyl-2-pyridylmethyl)]amine,51 bis((6-methyl-2-pyridyl)methyl)amine52

have been prepared according to previously reported procedures. Dibenzo[a,e]cyclooctadiene: Although various experimental procedures are available for the synthesis of dbcot,50 we did not find a full assignment of all 13C NMR signals of this compound. 1H (500 MHz, CDCl3): δ = 7.17 (m, 4H, Ar-H); 7.08 (m, 4H, Ar-H); 6.78 (s, 4H, olefinic). 13C (125 MHz, CDCl3): δ = 137.27 (CIV); 133.44 (olefinic); 129.31; 127.03. Syntheses: [Rh(κ3-bla)(η4-dbcot)]PF6 ([1]PF6): 179 mg of [Rh(coe)2Cl]2 (0.499 mmol Rh) and 107 mg of dbcot (0.524 mmol) were dissolved in 10 ml of CH2Cl2 and the solution was stirred for 2 hours and the solution was evaporated to dryness. Next, 115 mg bla (0.506 mmol) and a mixture of 10 ml of MeOH and 10 ml of CH2Cl2 were added and stirred for 1 hour. The unreacted solid was filtered off and the remaining solvent was removed in vacuo. The thus formed solid was dissolved in 5 ml MeOH and 106 mg KPF6 were added causing precipitation of a bright yellow solid, which was filtered and washed with 2 ml MeOH. Yield: 168 mg (0.247 mmol, 49.5 %). 1H (500 MHz, acetone-d6): δ = 7.80 (t, 3J(H,H) = 7.5 Hz, 2H; Py); 7.44 (d, 3J(H,H) = 7.5 Hz, 2H; Py); 7.35 (d, 3J(H,H) = 7.5 Hz, 2H; Py); 6.62 (m, 8H; Bn); 5.29 (d[AB] br, 2J(H,H) = 15.5 Hz, 2H; N-CH2-Py); 4.50 (d, 2J(Rh,H) = 1.5 Hz, 4H; Bn-CH-); 4.28 (d[AB], 2J(H,H) = 16.5 Hz, 2H; N-CH2-Py); 3.39 (s br, 6H; Py-CH3). 13C (125 MHz, acetone-d6): δ = 161.3 (Py); 161.2 (Py); 145.6 (Bn); 139.3 (Py-C4); 126.9 (Bn); 126.7 (Bn); 126.0 (Py-C5); 121.3 (Py-C3); 74.6 (d, 2J(Rh,H) = 12.6 Hz; (CH=CH); 58.3 (N-CH2-Py); 28.6 (Py-CH3). Elemental Analysis: Calcd. C, 53.03; H, 4.30; N, 6.18; Found: 52.77; H, 4.67; N, 5.98. [Ir(κ3-bla)(η4-dbcot)]PF6 ([2]PF6): 300 mg of [Ir(dbcot)Cl]2 (0.695 mmol Ir) were suspended in 10 mL of methanol and 160 mg of bla (0.704 mmol) was added. The solution was stirred for 30 min and turned transparent yellow. 128 mg KPF6 (0.695 mmol) was added causing precipitation of a bright yellow solid, which was filtered and washed with 2 ml MeOH. Yield: 234 mg (0.304 mmol, 43.7 %). 1H (500 MHz, acetone-d6): δ = 7.88 (t, 3J(H,H) = 7.5 Hz, 2H; Py); 7.52 (d, 3J(H,H) = 7.5 Hz, 2H; Py); 7.48 (d, 3J(H,H) = 7.5 Hz, 2H; Py); 6.91 (m, 4H; Bn); 6.79 (m, 4H; Bn); 5.33 (d[AB], 2J(H,H) = 17.0 Hz, 2H; N-CH2-Py); 4.52 (d[AB], 2J(H,H) = 17.0 Hz, 2H; N-CH2-Py); 4.37 (s, 4H; CH=CH); 3.30 (s, 6H; Py-CH3). 13C (125 MHz, acetone-d6): δ = 162.9 (Py); 161.5 (Py); 148.2 (Bn); 139.7 (Py-C4); 127.2 (Bn); 126.5 (Py-C5); 126.4 (Bn); 121.5 (Py-C3); 60.2 (N-CH2-Py); 58.3(CH=CH); 29.5 (Py-CH3). Elemental Analysis: Calcd for [2]PF6·0.5H2O: C, 46.33; H, 3.89; N, 5.40 Found: C: 46.58; H: 4.29; N: 5.12.


83

[Rh(κ3-bla)(η4-dbcot)](PF6)2 ([1](PF6)2): 94.5 mg (0.139 mmol) of [1]PF6 and 35.6 mg of AgPF6 (0.141 mmol) were dissolved in 3 ml of acetone and left to react for 30 minutes. Solution was filtered using a syringe filter and the filtrate was condensed to black oil in vacuo. 8 ml of CH2Cl2 were added causing precipitation of purple solid, which was centrifuged. Yield: 84 mg (0.102 mmol, 73.4 %). Unfortunately, despite many attempts, due to intrinsic instability of [1](PF6)2 we were not able to obtain an analytically pure sample. [Ir(κ3-bla)(η4-dbcot)](PF6)2 · CH2Cl2 ([2](PF6)2): 136 mg (0.177 mmol) of [2]PF6 and 50 mg of [Fe(μ-C5H4COMe)Cp]PF6 (0.15 mmol) were dissolved in 20 ml of dichloromethane and stirred for 45 minutes in which time a brown precipitate formed and the characteristic absorbance of acetylferrocenium was gone. Solution was filtered, washed with dichloromethane and dried in vacuo. Yield: 80 mg (0.08 mmol, 53.3 %). Magnetic susceptibility: μeff = 2.06 (corrected for diamagnetism of [2]PF6, PF6, Ir

2+ and CH2Cl2; measured: 1.90).53 Elemental Analysis: Calcd for [2](PF6)2·CH2Cl2: C, 37.28; H, 3.13; N, 4.21; Found: C: 37.24; H: 3.08; N: 4.36. [Ir(κ3-bla)(H)(σ-C8H13)(η2-C8H14)]PF6 ([3]PF6): An excess of dbcot (238, 1.17 mmol) and bla (265 mg, 1.17 mmol) were added to a suspension of [{Ir(coe)2Cl}2] (523 mg, 1.16 mmol Ir) in dry methanol (40 mL) and was stirred for 18 h. The reaction mixture turned to a transparent green solution, to which NH4PF6 was added causing precipitation of a white solid. The product was filtered off and recrystallized. Yield: 270 mg, 30%. Crystals suitable for X-ray diffraction were grown by layering an acetone solution of 3[PF6] with hexanes. 1H (acetone-d6, 300 MHz): δ = 7.79 (t, 3J(H,H) = 7.8 Hz, 1H; PyA); 7.63 (t, 3J(H,H) = 7.5 Hz, 1H; PyB); 7.46 (m, 2H; PyA); 7.25 (d, 3J(H,H) = 7.5 Hz, 2H; PyB); 6.60 (s, br, 1H; N-H); 5.47 (dd[AB], 2J(H,H) = 17.7 Hz, 3J(H,H) = 8.1 Hz, 1H; N-CH2-Py); 5.15 (dd[AB], 2J(H,H) = 15.9 Hz, 3J(H,H) = 5.1 Hz, 1H; N-CH2-Py); 5.05 (t, 3J(H,H) = 7.5 Hz, 1H; (IrC-CH); 4.76 (dd[AB], 2J(H,H) = 18.6 Hz, 1H; N-CH2-Py); 4.68 (dd[AB], 2J(H,H) = 15.6 Hz, 1H; N-CH2-Py); 3.88 (m, 2H; CH=CH); 3.13 (s, 3H; PyA-CH3); 3.02 (s, 3H; PyB-CH3); 2.66 (m, 1H; coe); 2.25 (m, 1H; coe); 1.5 (m, 22H; coe); -15.84 (s, 1H, Ir-H). 13C (75 MHz, acetone-d6): δ = 164.5 (Py); 163.4 (Py); 162.3 (Py); 161.7 (Py); 140.5 (PyA); 139.6 (PyB); 127.9 (IrC=CH); 127.5 (PyA); 127.2; (Ir-C); 125.6 (PyB); 122.1 (PyA); 121.6 (PyB); 69.3 (CH=CH); 64.6 (N-CH); 63.9 (N-CH) 63.8 (CH=CH); 40.8 (CH-CH2); 33.8 (CH-CH2); 33.1 (CH-CH2); 33.0 (CH-CH2); 32.4 (coe); 32.2 (PyACH3); 31.8 (PyBCH3); 29.9 (coe); 29.7 (coe); 28.8 (coe); 28.5 (coe); 25.5 (coe); 27.8 (coe); 27.6 (coe). Elemental Analysis: Calcd (C30H45F6IrN3P): C: 45.91, H: 5.78, N: 5.35, found C: 45.83, H: 5.86, N: 5.33. [Rh(κ3-Bn-bla)(η4-dbcot)]PF6 ([4]PF6): 180 mg of [Rh(coe)2Cl]2 (0.499 mmol Rh) and 102 mg of dbcot (0.499 mmol) were dissolved in 10 ml of CH2Cl2 and the solution was stirred for 2 hours and the solution was evaporated to dryness. Next, 164 mg Bn-bla (0.517 mmol) and a mixture of 8 ml of MeOH and 8 ml of CH2Cl2 were added and stirred for 1 hour and the solvent was removed in vacuo. The thus formed solid was dissolved in 6 ml MeOH and 126 mg KPF6 were added causing precipitation of a bright yellow solid, 2 ml of H2O was added and the solution was stirred for 15 minutes after which it was filtered and washed with 2 ml MeOH. Yield: 331 mg (0.430 mmol, 86.2 %). 1H (300 MHz, acetone-d6): δ = 7.80 (t, 3J(H,H) = 7.7 Hz, 2H; Py); 7.66 – 7.41 (m, 5H; Py, Ph); 7.36 (d, 3J(H,H) = 7.6 Hz, 2H; Py); 7.03 (m, br, 4H; Bndbcot); 6.97 (m, br, 4H; Bndbcot); 5.04 (d[AB] br, 2J(H,H) = 15.8 Hz, 2H; N-CH2-Py); 4.91 (s, 2H; N-CH2-Ph); 4.75 (s, br, 4H; Bn-CH-), 4.24 (s, br, 4H; Bn-CH-); 3.86 (d[AB], 2J(H,H) = 16.0 Hz, 2H; N-CH2-Py); 3.60 (s, 6H; Py-CH3)

Chapter 4

84

13C (125 MHz, acetone-d6): δ = 162.1 (Py); 160.6 (Py); 145.0 (Bndbcot-C1); 139.8 (Py-C4); 133.5 (Bn-C1); 132.7 (Ph-C2/6); 129.7 (Ph-C3/5); 129.7 (Ph-C4); 127.3 (Bndbcot-C2/3); 127.2 (Py-C5); 122.6 (Py-C3); 63.6 (N-CH2-Ph); 61.6 (N-CH2-Py); 30.0 (Py-CH3). The 13C NMR signals of the double bond of dbcot could not be located due to the fluxional behavior of this moiety. FAB+-MS: Calcd for [4]+ (C37H35N3Rh) m/z: 624.1883; Found m/z: 624.1886 (Δ = -0.5 ppm). [Rh(κ3-Bn-bla)(η4-dbcot)](PF6)2 ([4](PF6)2): 89.4 mg of [4]PF6 (0.116 mmol) and 29.4 mg AgPF6 (0.116 mmol) were dissolved in 3 ml CH2Cl2 causing instant precipitation of metallic silver and [4](PF6)2. Due to a rather low stability of [4](PF6)2, the attempts to isolate an analytically pure sample were not successful. Hence for the reaction with base we used the approx 1:1 mol mixture of [4](PF6)2 and Ag black obtained after decanting the green supernatant and drying the black-green powder in vacuo. Reaction of radical species 12+, 22+ and 42+ with K2CO3: In a typical experiment 0.02 mmol of the metal complex 12+, 22+ or 42+ was dissolved in 1 ml of acetone-d6 and stirred with 0.4 mmol of anhydrous solid K2CO3. After the solution turned yellow, NMR spectrum was recorded showing quantitative formation of 1+, 2+ or 4+

respectively. 4.5 Acknowledgements

We thank Luis Fuente Arruga for his contributions to this chapter. We thank Dr. Maxime Siegler and Prof. Dr. Anthony Spek for the X-ray structure determinations. We thank Han Peeters for measuring the FAB+-MS mass spectrum. 4.6 Notes and References

1 Whittaker, J. W. Chem. Rev. 2003, 103, 2347-2363. 2 (a) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que, Jr. L. Chem. Rev. 2004, 104, 939-986. (b) Limberg, C.

Angew. Chem. Int. Ed. 2003, 42, 5932-5954. 3 Meunier, B. de Visser, S. P. Shaik, S. Chem. Rev. 2004, 104, 3947-3980. 4 Baik, M.-H.; Newcomb, M.; Friesner, R. A.; Lippard, S. J. Chem. Rev. 2003, 103, 2385-2419. 5 Stubbe, J.; Nocera, D. G.; Yee, C. S.; Chang, M. C. Y. Chem. Rev. 2003, 103, 2167-2201. 6 (a) Gridnev, A. A.; Ittel, S. D. Chem. Rev. 2001, 101, 3611-3659. (b) de Bruin, B.; Dzik, W. I.; Li, S.;

Wayland, B. B. Chem. Eur. J. 2009, 15, 4312-4320. (c) see Chapter 7 of this Thesis. 7 Penoni, A.; Wanke, R.; Tollari, S.; Gallo, E.; Musella, D.; Ragaini, F.; Demartin, F.; Cenini, S. Eur. J.

Inorg. Chem. 2003, 1452-1460. 8 (a) Sherry, A. E.; Wayland, B. B. J. Am. Chem. Soc. 1990, 112, 1259-1261. (b) Cui, W.; Zhang, X. P.;

Wayland, B. B. J. Am. Chem. Soc. 2003, 125, 4994-4995. (c) Cui, W.; Wayland, B. B. J. Am. Chem. Soc. 2004, 126, 8266-8274. (d) Puschmann, F. F.; Grützmacher, H.; de Bruin, B. J. Am. Chem. Soc. 2010, 132, 73-75.

9 (a) Zhang, L.; Chan, K. S. J. Organomet. Chem. 2006, 691, 3782-3787. (b) Dzik, W. I.; Reek, J. N. H.; de Bruin, B. Chem. Eur. J. 2008, 14, 7594-7599.

10 Königsmann, M.; Donati, N.; Stein, D.; Schönberg, H.; Harmer, J.; Sreekanth, A.; Grützmacher, H. Angew. Chem. Int. Ed. 2007, 46, 3567-3570.

11 Hetterscheid, D. G. H.; Klop, M.; Kicken, R. J. N. A. M.; Smits, J. M. M.; Reijerse, E. J.; de Bruin, B. Chem. Eur. J. 2007, 13, 3386-3405.

12 For an overwiew see: Hicks, R. G. Angew. Chem. Int. Ed. 2008, 47, 7393-7395. 13 Copper: (a) Mankad, N. P.; Antholine, W. E.; Szilagyi, R. K.; Peters, J. C. J. Am. Chem. Soc. 2009, 131,

3878-3880. Ruthenium: (b) Miyazato, Y.; Wada, T.; Muckerman, J. T.; Fujita, E.; Tanaka, K. Angew. Chem. Int. Ed. 2007, 46, 5728-5730. Nickel: (c) Adhikari, D.; Mossin, S.; Basuli, F.; Huffman, J. C.; Szilagyi, R. K.; Meyer, K.; Mindiola, D. J. J. Am. Chem. Soc. 2008, 130, 3676-3682. For aminyl radicals coordinated to iridium and rhodium see ref 15.

14 (a) Roth, J. P.; Yoder, J. C.; Won, T.-J.; Mayer, J. M. Science, 2001, 294, 2524-2526. (b) Yoder, J. C.; Roth, J. P.; Gussenhoven, E. M.; Larsen, A. S.; Mayer, J. M. J. Am. Chem. Soc. 2003, 125, 2629-2640. (c)


85

Mader, E. A.; Larsen, A. S.; Mayer, J. M. J. Am. Chem. Soc. 2004, 126, 8066-8067. (d) Mader, E. A.; Davidson, E. R.; Mayer, J. M. J. Am. Chem. Soc. 2007, 129, 5153-5166. (e) Wu, A.; Mayer, J. M. J. Am. Chem. Soc. 2008, 130, 14745-14754. (f) Mader, E. A.; Manner, V. W.; Markle, T. F.; Wu, A.; Franz, J. A.; Mayer, J. M. J. Am. Chem. Soc. 2009, 131, 4335-4345.

15 (a) Büttner, T.; Geier, J.; Frison, G.; Harmer, J.; Calle, C.; Schweiger, A.; Schönberg, H.; Grützmacher, H. Science 2005, 307, 235-238. (b) Maire, P.; Königsmann, M.; Sreekanth, A.; Harmer, J.; Schweiger, A.; Grützmacher, H. J. Am. Chem. Soc. 2006, 128, 6578-6580. (c) Donati, N.; Stein, D.; Büttner, T.; Schönberg, H.; Harmer, J.; Anadaram, S.; Grützmacher, H.; Eur. J. Inorg. Chem. 2008, 4691-4703.

16 (a) Hetterscheid, D. G. H. PhD Thesis, Radboud University Nijmegen, 2006. (b) Tejel, C.; Ciriano, M. A.; del Río, M. P.; van den Bruele, F. J.; Hetterscheid, D. G. H.; Tsichlis i Spithas, N.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 5844-5845. (c) Tejel, C.; Ciriano, M. A.; del Río, M. P.; Hetterscheid, D. G. H.; Tsichlis i Spithas, N.; Smits, J. M. M.; de Bruin, B. Chem. Eur. J. 2008, 14, 10932-10936. (d) Tejel, C.; del Río, M. P.; Ciriano, M. A.; Reijerse, E. J.; Hartl, F.; Záliš, S.; Hetterscheid, D. G. H.; Tsichlis i Spithas, N.; de Bruin, B. Chem. Eur. J. 2009, 15, 11878-11889.

17 Recently it was shown that replacement of cod ligand with dbcot greatly improved the (air) stability of an iridium catalyst for allylic substitutions: Spiess, S.; Welter, C.; Franck, G.; Taquet, J. -P.; Helmchen, G. Angew. Chem. Int. Ed. 2008, 47, 7652-7655.

18 (a) Fernandez, M. J.; Rodriguez, M. J.; Oro, L. A.; Lahoz, F. J. J. Chem. Soc. Dalton Trans. 1989, 2073-2076. b) Alvarado, Y.; Boutry, O.; Gutiérrez, E.; Monge, A.; Nicasio, M. C.; Poveda, M. L.; Pérez, P. J.; Ruíz, C.; Bianchini, C.; Carmona, E. Chem. Eur. J. 1997, 3, 860-873.

19 Iimura, M.; Evans, D. R.; Flood, T. C. Organometallics 2003, 22, 5370-5373. 20 Dzik, W. I.; Smits, J. M. M.; Reek, J. N. H.; de Bruin, B. Organometallics 2009, 28, 1631-1643. 21 (a) Hermann, D.; Gandelman, M.; Rozenberg, H.; Shimon, L. J. W.; Milstein, D. Organometallics 2002,

21, 812-818. (b) Friedrich, A.; Ghosh, R.; Kolb, R.; Herdtweck, E.; Schneider, S. Organometallics 2009, 28, 708-718.

22 Rossi, A. R.; Hoffmann, R.; Inorg. Chem. 1975, 14, 365-374. 23 Irngartinger, H.; Reibel, W. R. K. Acta Cryst. B. 1981, B37, 1724-1728. 24 (a) [Rh(C6Cl5)2(cod)]: García, M. P.; Jiménez, M. V.; Oro, L. A.; Lahoz, F. J.; Casas, J. M.; Alonso, P. J.

Organometallics 1993, 12, 3257-3263. (b) [Ir(C6Cl5)2(cod)]: García, M. P.; Jiménez, M. V.; Oro, L. A.; Lahoz, F. J.; Alonso, P. J. Angew. Chem. Int. Ed. 1992, 31, 1527-1529. (c) [(η5-C5Ph5)Rh(cod)]+ : Shaw, M. J.; Geiger, W. E.; Hyde, J.; White, C. Organometallics 1998, 17, 5486-5491. (d) [Ir(Me3tpa)(C2H4)]2+: de Bruin, B.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Wilting, J. B. M.; de Gelder, R.; Smits, J. M. M.; Gal, A. W. Angew. Chem. Int. Ed. 2002, 41, 2135-2138. e) for a review see: de Bruin, B.; Hetterscheid, D. G. H. Eur. J. Inorg. Chem. 2007, 211-230.

25 (a) Ogoshi, H.; Setsunu, J.; Yoshida, Z. J. Am. Chem. Soc. 1977, 99, 3869-3870. (b) Wayland, B. B.; Feng, Y.; Ba, S. Organometallics 1989, 8, 1438-1441.

26 (a) Brammer, L.; Connelly, N. G.; Edwin, J.; Geiger, W. E.; Orpen, A. G.; Sheridan, J. B. Organometallics 1988, 7, 1259-1265. b) Zhai, H.; Bunn, A.; Wayland, B. Chem. Commun. 2001, 1294-1295. c) Bunn, A. G.; Wayland, B. B. J. Am. Chem. Soc. 1992, 114, 6917-6919.

27 Both compounds could be oxidized by Ag+ in acetone. 22+ was prepared by oxidation with [Fe(η5-C5H4COMe)Cp]+ in CH2Cl2, however this oxidant was too weak to oxidize 1+. For a list of common redox agents in organometallic chemistry see: Connelly, N. G.; Geiger, W. E. Chem. Rev. 1996, 96, 877-910.

28 (a) Hetterscheid, D. G. H.; Grützmacher, H.; Koekoek, A. J. J.; de Bruin, B. Prog. Inorg. Chem. 2007, 55, 247-354. (b) Wassenaar, J.; de Bruin, B.; Siegler, M. A.; Spek, A. L.; Reek, J. N. H.; van der Vlugt, J. I. Chem. Commun. 2010, 47, 1232-1234. For examples of related [Rh(N3-ligand)(diolefin)]+/2+ complexes see also ref 11 and (c) Hetterscheid, D. G. H.; Smits, J. M. M.; de Bruin, B. Organometallics 2004, 23, 4236-4246.

29 (a) Hetterscheid, D. G. H.; J. Kaiser; Reijerse, E. J.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Smits, J. M. M.; de Gelder, R.; de Bruin B. J. Am. Chem. Soc. 2005, 127, 1895-1905.

30 In acetone-d6 a very slow rise of the signals of the diamagnetic precursor could be observed for both complexes, in case of the rhodium complex 12+ accompanied by the rise of signals of the free dbcot moiety. Kinetic measurements using UV/Vis spectrometry revealed that the decomposition of 12+ and 22+ proceeds according to a first order rate equation in the metal complex concentration with kobs = 1·10–6 and 4·10–7 s–1 respectively. Addition of H2O increases the rate of decomposition considerably. However in case of 12+ in an aselective manner.

31 The same results were obtained when exactly 1 equivalent of KOtBu was added. This base however proved to be less suitable, because if used in excess, some follow-up unselective reactivity was observed, possibly via doubly deprotonated bla complexes (ref 16).

Chapter 4

86

32 If the aminyl radical would abstract a hydrogen atom from the solvent one would expect that mostly deuterium would be incorporated. Alternatively a very large kinetic isotope effect could explain the obtained hydrogen atom, see: Huynh, M. H. V.; Meyer, T. J. Proc. Natl. Acad. Sci. USA 2004, 101, 13138-13141.

33 Evans, C. A. Aldrichimica Acta 1979, 12, 23-29. 34 Radical trapping of species 12+ in acetone solution using DMPO (DMPO = 5,5-dimethyl-1-pyrroline-N-

oxide) revealed formation of at least two species A and B, however neither of these could not be assigned to a solvent radical and most probably a rhodium radical has been captured. These species existed in 1:0.1 ratio which changed in time. The simulated parameters: I (major): giso = 2.007, AN = 48.9 MHz, AH = 56.6 MHz, ARh = 4.9 MHz; II (minor) giso = 2.0049, AN = 46.3 MHz, AH = 57.5 MHz. Species 22+ with DMPO and K2CO3 gave a complex, very low intensity signal which was impossible to simulate reliably.

35 The oxidation of CO32- to C2O6

2- seems to be less probable. Although the redox potential of the C2O62-

/CO32- couple in aqueous carbonate buffer is +0.67 V vs. Fc/Fc+ (0.21 vs. SCE), the low concentration of

carbonate anions in acetone makes this pathway energetically unfavorable. Zhang, J.; Oloman, C. W.; J. Appl. Electrochem. 2005, 35, 945-953.

36 This equilibrium lays on the side of 12+/22+ and acetone (the oxidation of acetone occurs over 1000 mV vs. the Fc/Fc+ couple11 which gives the energy difference between the two sides of the equilibrium of roughly 600-700 mV or 14-16 kcal/mol).

37 The redox potential of the –OH/•OH couple in MeCN is only +0.29 vs Fc/Fc+ (+0.96 vs NHE): (a) Sawyer, D. T.; Roberts, jr, J. K. Acc. Chem. Res. 1988, 21, 469-476. (b) calculated from the relationship NHE – E1/2 Fc/Fc+ = 630 mV: Pavlishchuk, V. V.; Addison, A. W. Inorg. Chim. Acta 2000, 298, 97-102.


39 Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346-354. 40 PQS version 2.4, 2001, Parallel Quantum Solutions, Fayettevile, Arkansas (USA); the Baker optimizer is

available separately from PQS upon request; Baker, J.; J. Comput. Chem. 1986, 7, 385-395. 41 Sierka, M.; Hogekamp, A.; Ahlrichs, R.; J. Chem. Phys. 2003, 118, 9136-9148. 42 (a) Becke, A. D.; Phys. Rev. A, 1988, 38, 3098-3100. (b) Perdew, J. P.; Phys. Rev. B, 1986, 33, 8822-

8824. 43 Schaefer, A.; Horn, H.; Ahlrichs, R.; J. Chem. Phys. 1992, 97, 2571-2577. 44 (a) G-tensor: van Lenthe, E.; van der Avoird A.; Wormer, P. E. S. J. Chem. Phys. 1997, 107, 2488. (b) A-

tensor: van Lenthe, E.; van der Avoird A.; Wormer, P. E. S. J. Chem. Phys. 1998, 108, 4783. 45 te Velde, G.; Bickelhaupt, F. M.; van Gisbergen, S. J. A.; Fonseca Guerra, C.; Baerends, E. J.; Snijders J.

G.; Ziegler, T. J. Comput. Chem. 2001, 22, 931. 46 (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B, 1988, 37, 785-789. (b) Becke, A. D. J. Chem. Phys. 1993,

98, 1372-1377. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. 47 Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chem. Acc. 1997, 97, 119-124. 48 (a) Anton, D. R.; Crabtree, R. H. Organometallics 1983, 2, 621-627. (b) Singh, A.; Sharp, P. R. Inorg.

Chim. Acta 2008, 361, 3159-3164. 49 van der Ent, A.; Onderdenlinden, A. L. Inorg. Synth. 1990, 28, 90-91. 50 Chaffins, S.; Brettreich, M.; Wudl, F. Synthesis 2002, 9, 1191-1194. Instead of Li sand (as described in

the original procedure), for performing the first step we used granular Li (0.5% Na), and obtained dbcot in 15 % overall yield.


52 Komatsu, K.; Kikuchi, K.; Kojima, H.; Urano, Y.; Nagano, T. J. Am. Chem. Soc. 2005, 127, 10197-10204.

53 Bain, G. A.; Berry, J. F. J. Chem. Edu. 2008, 85, 532-536.

Chapter 5

Selective C–C Coupling of Ir-Ethene and Ir-Carbene Radicals*

N

N N

NIr

NCMe

N

N

N

N

IrMeCN

O

OEt

N

N N

NIr

H

N2

C(O)OEt

N

N N

NIr

NCMe

H SiMe3H

H

N2

SiMe3

2+

II.4+

III

MeCN

- ethene

III

2+

III

MeCN

- ethene

+ H.

* Part of this work has been published: Reproduced in part with permission from [Dzik, W. I.; Reek, J. N. H.; de Bruin, B. Chem. Eur. J. 2008, 14, 7594-7594] Copyright © [2008] WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chapter 5

88

5.1 Introduction

Although free carbenes are intrinsically unstable, taming of their high reactivity by coordination to the metal centre has resulted in an enormous scope of catalytic transformations like alkene metathesis1 and cyclopropanation,2 C-H bond insertion,3 carbene polymerization4 and many more. Most of the above reactions proceed via diamagnetic pathways and only recently the possibly valuable radical-type involvement of open-shell transition metal carbene species in cyclopropanation reactions has been proposed,5 but without solid experimental or computational support.

Chiral cobalt (II) complexes with salen6 or porphyrin7 ligands developed by groups of Katsuki and Zhang are some of the most successful examples of cyclopropanation catalysts. Bergman and coworkers reported enantioselective carbene transfer from diazo compounds to olefins and imines mediated by RhII(benbox)Cl+, presenting a rare example of a catalytically active paramagnetic mononuclear rhodium (II) complex.8 So far, the putative open-shell carbene complexes have neither been isolated nor unambiguously characterized.

Although a variety of group 9 complexes are capable of carbene transfer catalysis,9 this area of research is clearly dominated by investigations of diamagnetic Doyle-type dinuclear acetate bridged RhII−RhII species (and alike) and a couple of mononuclear diamagnetic MIII catalysts.9,10 Remarkably, the reactivity of both mononuclear (diamagnetic) MI and mononuclear (paramagnetic) MII complexes (M = Rh, Ir)11 with diazo compounds has received almost no attention. Recent investigations on the reactivity of MI complexes with diazo compounds led to a discovery that mononuclear MI(diene) complexes (M = Rh, Ir) mediate the formation of polymers from ethyl diazoacetate.12 Motivated by these initial results, we became interested in the influence of the oxidation state on the reactivity of mononuclear Rh and Ir carbene complexes. Further inspired by our general interest in the reactivity of paramagnetic RhII and IrII species with redox non-innocent ligands, we decided to investigate the reactivity of a mononuclear IrII(ethene) complex with ethyl diazoacetate and trimethylsilyl-diazomethane, which are common carbene precursors. A recent study by Chan describes the selective formation of [(por)RhIII(CH2COOEt)] upon reaction of [(por)RhII] with ethyl diazoacetate (EDA),13 likely via an open-shell carbenoid species, which is relevant to the work described in this Chapter.

5.2 Results and discussion

We concentrated on the reactivity of the previously reported paramagnetic IrII(ethene) complex [(Me3tpa)IrII(ethene)]2+ (1) (Me3tpa = N,N,N-tris(6-methyl-2-pyridylmethyl)amine). Interestingly, complex 1 reveals both ‘metallo-radical’ and ‘alkene radical’ behavior (Scheme 1).11a,b,14 The ‘vacant’ site cis to the ethene ligand is sterically shielded by the three Me fragments of the Me3tpa ligand, and therefore not accessible for larger molecules. However, the small and linear MeCN can bind to this position, by which it induces ethene ligand centered radical reactivity. De Bruin and coworkers previously proposed that β-ethyl radicals of the type [(Me3tpa)IrIII-CH2CH2•]

2+ (2) could be intermediates in a series of radical type reactions triggered by coordination of MeCN to the otherwise stable metallo-radical 1.14 Ethene loss from intermediate 2 should be easy, giving access to the metallo-radical [(Me3tpa)IrII(NCMe)]2+ (3). Somewhat similar behavior of (por)IrII species in their reactivity with ethene has been reported.15

Selective C–C coupling of Ir-Ethene and Ir-Carbene Radicals

89

N

N NN Ir

N

N N

N IrNCMe

N

N N

N IrNCMe

2+

II.

2+

III

.

MeCN

2+

II

.- ethene

1 2 3

Scheme 1. Reactivity of [(Me3tpa)IrII(ethene)]2+ towards MeCN.

It is thus not so clear what to expect about the reactivity of complex 1 towards diazo compounds. Would they react directly with 1, or would such reaction preferably take place via 2 or 3 in the presence of MeCN? One-electron activation of diazo compounds may allow radical-type reactions of its ‘carbene’ moiety. This should expand the reactivity scope of carbenes coordinated to late transition metals beyond the well-known, standard 2e transformations.

The reaction of complex 1 towards ethyl diazoacetate (EDA) in acetone leads to a complex mixture of compounds. In contrast, the reaction of 1 with EDA in MeCN proceeds selectively as judged from the 1H NMR spectrum. The reaction results in formation of the tetracationic dinuclear C3-bridged species [(Me3tpa)(MeCN)IrIII-(CH2CH2CH(COOEt))IrIII(NCMe)(Me3tpa)]4+ (4) in 74 % isolated yield (Scheme 2).

N

N N

NIr

H

N2

C(O)OEt

N

N N

NIr

NCMe

N

N

N

N

IrMeCN

O

OEt

N

N N

NIr

N

NN

NIr

NCMe

MeCN H

N2

C(O)OEt

2+

II.

4+

III

MeCN

1

2- ethene

4

III

III

5

III

4+

MeCN- ethene

kobs2

kobs1

Scheme 2. Reaction of 1 with EDA in MeCN to give 4. In absence of EDA 1 converts to 5. The observed rate constants for the two reactions are identical (kobs1 = kobs2).

Addition of excess EDA does not have any influence on the outcome of the reaction. The reactivity of 1 with EDA in MeCN could reflect the metallo-radical behavior of 3, the ethene ligand-radical behavior of 2, or both. In absence of EDA, 1 in MeCN slowly converts to the ethylene bridged dinuclear iridium complex [(Me3tpa)(MeCN)IrIII(μ2-

Chapter 5

90

CH2CH2)IrIII(NCMe)(Me3tpa)]4+ (5).14a Complex 5, however, does not react with EDA,

so carbene insertion (from EDA) into the Ir−C bond of 5 to form 4 can be excluded as a mechanistic pathway.

Two mechanistic pathways seem conceivable: (A) dissociation of ethene from 1 with formation of 3, followed by coordination of the diazo compound to iridium (6) with subsequent loss of dinitrogen to form the ‘radical carbenoid’ 7, which in turn attacks the ethene fragment of 1 or 2 to form the new C−C bond of 4; or (B) attack of the ethene carbon-centered radical of 2 at the diazo carbon atom of EDA to form the γ-alkyl radical IrIII−CH2CH2CH•(COOEt) species 9, which would eventually couple with the Ir-centered radical 3. These routes are depicted in Scheme 3.

Scheme 3. Conceivable pathways to the C3-bridged bis-iridium species 4.

EPR measurements did not allow us to detect any intermediates during the conversion of 1 to 4; even in the presence of a large excess of EDA only gradually disappearing signals of 1 were detected. Also the reaction kinetics did not allow us to discriminate between the reaction pathways A and B in Scheme 3. The kinetic measurements revealed that the reaction is first order in [1] and zero order in [EDA], and proceeds at exactly the same rate as observed for the formation of 5 in absence of EDA (kobs1 = kobs2 = 0.0075 min–1). So ethene dissociation from 2, or MeCN coordination to 1 are the most likely rate limiting steps (Scheme 3, formation of 2 or 3).

In absence of more experimental details we resorted to DFT calculations to obtain more information about the possible reaction mechanisms leading to 4. We used methyl diazoacetate to model EDA. Given the importance of the three methyl fragments on the reactivity of [(Me3tpa)IrII(ethene)]2+, we decided to model the complete dications. Due


91

to the substantial size of the full (Me3tpa)Ir system, we did not try to calculate the transition states or energies associated with the final radical couplings in Scheme 3 (3 + 9 → 4 or 2 + 7 → 4). Besides, this would not be very useful because DFT energies of reactions involving spin-state changes, such as radical-radical coupling reactions, are generally not very reliable.16,17 The reaction pathways prior to the actual coupling, however, all deal with S = ½ systems, and the relative energies of the minima and transition states along the pathways 1 → 2 → 8 → 9 and 1 → 2 → 3 → 6 → 7 should be meaningful.

Both reaction pathways (A and B in Scheme 3) are thermodynamically favorable; with an overall stair-case decrease of free energies along the reaction coordinates (Scheme 4). Formation of 2 from 1 is slightly uphill, but this is mainly due to (gas phase) entropy effects associated with the approach of MeCN, which do not play a large role in the experimental system where MeCN is used as a solvent.

Scheme 4. Calculated free energies (energies) in kcal/mol of the intermediates and transition states in routes A and B leading to formation of 4 (see also Scheme 3).

The TS for ethene loss from 2 (TS2) represents a very low barrier (+0.3 kcal/mol) for

formation of 3. This barrier is most likely too low, as DFT tends to underestimate the iridium-ethene interactions.18 This becomes evident, and worse, upon using the hybrid HF-DFT functional b3-lyp, at which level (TZVP basis) species 2 proved unstable and converged to 3 by spontaneous ethene dissociation.19 For reasons of comparison, we thus used the BP86 (SV(P) basis) throughout. Although the actual barrier for ethene loss from 2 might be somewhat higher than calculated, it is unlikely to be very high. Therefore the involvement of species 2 can only be relevant for low barrier follow-up reactions. This is not the case in the reaction of 2 with MDA to form 8, which has a relatively large barrier (TS2 ~ +13.5 kcal/mol). In contrast, the approach of MDA to bind to the Ir centre of 3 (reaction 3 → 6) has a very flat energy profile and is essentially barrierless. Thus MDA binding to the β-carbon atom of 2 seems unlikely, since pathway A is kinetically preferred over pathway B according to the DFT calculations.

Chapter 5

92

Once formed, the MDA adducts 6 and 8 both reveal a remarkably low barrier for N2 loss (TS3 ~ 0.2 kcal/mol, TS4 is essentially barrierless once entropy corrections are applied). This is in marked contrast with substantially higher calculated barriers (in the range of 7-12 kcal/mol) for N2 loss from MDA and diazoalkane adducts of closed-shell RhI species.12a, 20

We argued that we might obtain additional experimental evidence for the reaction proceeding via pathway A by using a bulkier substituted diazo compound. Sufficient steric bulk should prevent the organometallic radicals from coupling, and if the reaction would proceed via pathway A, the thus expected final product would be a mononuclear iridium compound derived from a bulky analog of 7. In fact, trimethylsilyldiazomethane (TMSDM) proved to be a proper reagent for verifying this hypothesis.

The reaction of 1 with TMSDM in MeCN affords the mononuclear iridium compound 11 in 41% isolated yield. Compound 11 has a methylene(trimethylsilyl) group bound to the metal (Scheme 5). This is consistent with the above DFT calculations predicting the ‘radical carbenoid’ pathway (A) to be kinetically preferred over a direct C−C coupling between the diazo carbon atom and the β-carbon of ethyl radical complex 2.

N

N NN Ir

N

N N

N IrNCMe

N

N N

N IrNCMe

N

N N

NIr

NCMe

H SiMe3

N

N N

NIr

NCMe

H SiMe3H

2+

II.

2+

III

.

MeCN

2+

II

.- ethene

1 2 3

2+

III

.10

N2CHSiMe3- N2

2+

III

11

H.

Scheme 5. Formation of iridium (trimethylsilyl)methylene complex 11 consistent with a radical carbenoid pathway.

The ‘radical carbenoid’ intermediate 10 abstracts a hydrogen atom from the reaction mixture. The solvent is not the source of the hydrogen atom, as experiments in deuterated solvents did not lead to incorporation of deuterium. The most likely source is the (trimethylsilyl)diazomethane reagent.13

The key reactive species responsible for the formation of 4 and 11 thus seem to be the radical carbenoid species 7 and 10, respectively. Such (N-ligand)IrII-carbene complexes cannot be classified as regular Fischer- or Schrock-type carbenes, and their unusual electronic structure requires some additional comments (Scheme 6). As shown in Figure 1 for complex 7, the species are primarily carbon centered radicals. Most of the spin density of 7 is located at its carbenoid carbon (Mulliken spin densities: Ccarbene: 86%, Ir: 5%, Ocarbonyl: 12%, Ccarbonyl: –4%, OMe: 5%, Hcarbene: –5%).


93

N

N N

NIr

NCMe

H R

N

N N

NIr

NCMe

H R

2+

III

.

2+

II.

Scheme 6. ‘Redox non-innocence’ of open-shell TM carbene complexes; the ‘IrII(carbene)’ complexes 7 and 11 are best described as IrIII-carbon centered radicals.

The SOMO of 7 is primarily built from the carbenoid carbon p-orbital in anti-bonding combination with an Ir d-orbital with a smaller orbital coefficient (dxz if we define the z-axis along the Ir−C bond and the x-axis along the Ir−Namine bond), with some expected delocalization over the adjacent carbonyl fragment (Figure 1). The carbenoid radical 7 thus seems to be best described as a one-electron reduced Fischer-type carbene complex (Figure 2, top).22 In good agreement with occupation of an electron in the Ir−C anti-bonding orbital (π*), which reduces the Ir−C bond-order, the calculated Ir−C bond of 7 (2.02 Å) is rather long compared to related closed-shell systems (~1.85-1.91 Å).12b,20

Figure 1. SOMO (left) and spin density (right) plots of 7.

It is quite remarkable that the redox chemistry of transitionmetal carbene complexes has thus far received so little attention. The frontier orbitals of both Fischer- and Schrock-type carbenes are such that the carbene fragment must be ‘redox non-innocent’. The LUMO of a Fischer-type carbene is primarily carbon-centered, which gives the carbene fragment its intrinsic electrophilic character, but also allows the formation of a carbon centered radical upon one-electron reduction. Similarly, the carbon-centered HOMO of a Schrock-type carbene gives the carbene fragment its nucleophilic character, but also allows the formation of a carbon centered radical upon one-electron oxidation (see Figure 2).

Chapter 5

94

Figure 2. Schematic representation of the frontier orbitals involved in formation of carbon centered radicals by one-electron reduction of Fischer-type carbenes (top) or one-electron oxidation of Schrock-type carbenes (bottom).

Thus obtained carbon-centered radicals can not be described as regular Fischer- or

Schrock-type carbenes, and actually represent a new class of ‘one-electron activated carbenoids’. For group 9 elements (Co, Rh, Ir), we are aware of only one such previously claimed example.21 More general, reactivity studies of open-shell carbenoid complexes (one-electron activated carbene complexes) are in an early stage of development and only a few reports describe the formation of C−C bonds via open-shell radical carbenoid complexes.22 The redox non-innocence of more stable N-heterocyclic carbenes was addressed very recently.23 5.3 Conclusion

We demonstrated the unprecedented formation of a paramagnetic iridium bound ‘carbenoid radical’ complex that selectively couples to an IrII(ethene) species with formation of a C−C bond. A more bulky ‘carbenoid radical’, obtained by reacting trimethylsilyl diazomethane with the IrII(ethene) complex, abstracts a hydrogen atom from the reaction mixture instead of undergoing C−C coupling with an IrII(ethene) species. DFT calculations show that the unpaired electron density of both ‘carbenoid radical’ species resides mainly on the ‘carbenoid’ carbon, explaining the reactivity of the two complexes. This study expands the scope of known reactivity of diazo compounds and allows the consideration of ‘radical carbenoids’ in selective C−C bond formation reactions. In the following chapter we will show that the concept of ‘carbenoid radical’ is crucial for understanding the mechanism of olefin cyclopropanation mediated by CoII species.


95

5.4 Experimental

General procedures

All manipulations were performed in an argon atmosphere by standard Schlenk techniques or in a glove box. Acetonitrile was distilled under argon from CaH2. NMR experiments were carried out on a Bruker DRX300 (300 and 75 MHz for 1H and 13C respectively) and Varian Inova (500 MHz for 1H). Solvent shift reference: CD3CN δ = 1.94 and δ = 1.24 for 1H and 13C respectively. Abbreviations used are s = singlet, d = doublet, t = triplet, m = multiplet. Elemental analysis (CHN) was carried out by H. Kolbe Mikroanalytisches Laboratorium (Germany). Kinetic measurements were performed on Perkin-Elmer Lambda 5 UV-Vis spectrometer in a thermostated quartz Schlenk cuvette at 20 °C. X-band EPR spectroscopy measurements were performed with a Bruker EMX Plus spectrometer. Fast Atom Bombardment (FAB) mass spectrometry was carried out using a JEOL JMS SX/SX 102A four-sector mass spectrometer, coupled to a JEOL MS-MP9021D/UPD system program. Samples were loaded in a matrix solution (3-nitrobenzyl alcohol) on to a stainless steel probe and bombarded with Xenon atoms with energy of 3 keV. During the high resolution FAB-MS measurements a resolving power of 10,000 (10% valley definition) was used. Ethyl diazoacetate and 2 M solution of (trimethylsilyl)diazomethane in diethylether were obtained from Aldrich, and used without further purification. Complex 1 ([(Me3tpa)IrII(ethene)](PF6)2) was prepared as described before.14a Syntheses: [Ir(Me3tpa)(MeCN)(CH2CH2CHCOOEt)(MeCN)(Me3tpa)Ir](PF6)4 (4): 48 mg (0.057 mmol) of 1 was dissolved in 4 ml MeCN. Subsequently 6 μl (0.057 mmol) of EDA was added using a micropipette and the reaction was stirred overnight. The reaction mixture was concentrated to approx 0.5 ml and 4 ml of MeOH was added causing precipitation of 4 as a white powder. Yield: 38 mg (74%). 1H NMR (500 MHz, CD3CN): δ = 7.99 (1H, t, 7.5 Hz); 7.90 (2H, m); 7.82 (1H, t, 7.5 Hz); 7.65 (2H, m); 7.53 (1H, d, 7.5 Hz); 7.42 (2H, d, 7.5 Hz); 7.34 (4H, m); 7.25 (1H, d, 8 Hz); 7.21 (2H, d, 7.5 Hz); 7.16 (1H, d, 7.5 Hz); 7.13 (1H, d, 7.5 Hz); 5.45 (1H, d, 16 Hz); 5.38 (1H, d, 15.5 Hz); 5.24 (1H, d, 16.5 Hz); 4.88 (1H, d, 17.5 Hz); 4.85 (2H, d, 17 Hz); 4.76-4.65 (3H, m); 4.59 (2H, s); 4.48 (1H, d, 16 Hz); 3.79 (1H, m, -OCH2CH3); 3.63 (1H, d, 11.5 Hz, Ir-CH-COOEt); 3.10 (6H, s, PyCH3); 2.923 (3H, s, Ir-NCCH3); 2.917 (3H, s, Ir-NCCH3); 2.81 (1H, m, 1H, m, -OCH2CH3); 2.79 (3H, s, PyCH3); 2.65 (3H, s, PyCH3); 2.60 (3H, s, PyCH3); 2.32 (3H, s, PyCH3); 2.29 (2H, m, Ir-CH2); 1.05 (1H, m, CH2CH2CHCOOEt); 0.67 (3H, 7 Hz, OCH2CH3); -0.28 (1H, m, CH2CH2CHCOOEt). 13C NMR (75 MHz, CD3CN): δ = 182.8 (COOEt); 165.9, 165.5, 165.2, 164.9, 164.8, 164.4, 164.0, 163.8, 163.4, 162.6, 159.1, 158.2 (Py-C2/6); 141.7, 141.4, 141.2, 140.6, 140.2 (Py-C4); 129.2, 128.8, 128.7, 128.6, 128.4, 127.7, 123.2, 122.7, 122.6, 122.2, 120.3, 120.1 (Py-C3/5); 74.7, 74.2, 71.1, 71.03, 70.96, 70.6 (Py-CH2-N); 60.6 (OCH2CH3); 33.5 (C-CH2-C); 27.5, 27.30, 27.25, 27.1, 26.6 (Py-CH3); 13.9 (OCH2CH3); 11.1 (Ir-CH(COOEt)); -2.0 (Ir-CH2). Anal Calcd for C52H64F24Ir2N10O2P4: C 34.21, H 3.53, N 7.67, found: C 34.50, H 3.64, N 7.32. [Ir(Me3tpa)(MeCN)(CH2Si(CH3)3)(PF6)2] (11): 57 mg (0.068 mmol) of 1 was dissolved in 4 ml MeCN. Subsequently 34 μl (0.068 mmol) of 2M solution of (trimethylsilyl)diazomethane in Et2O was added using a micropipette and the reaction was stirred overnight. The reaction mixture was concentrated to approx 0.5 ml and 4 ml of CH2Cl2 was added, causing a little precipitation of 5. The solution was decanted and the solvent was removed under vacuum to afford brown solid. The solid was redissolved in MeCN, and EtOH was added causing precipitation of a brown oil which was discarded. Decanting and removing the solvent under vacuum yielded 26 mg (0.028 mmol – 41 %) of 11 as an off-white powder. 1H NMR (300 MHz, CD3CN): δ = 7.82 (2H, t, 7.5 Hz, PyA-H4); 7.65 (1H, t, 7.8 Hz, PyB-H4); 7.38-7.29 (5H, m, PyA-H3, PyA-H5, PyB); 7.15 (1H, d, 7.5 Hz, PyB); 5.26 (2H, d[AB], 16.2 Hz, N-CH2-PyA); 4.83 (2H, d[AB], 16.2 Hz, N-CH2-PyA); 4.63 (2H, s, N-CH2-PyB); 3.14 (3H, s, PyB-

Chapter 5

96

CH3); 2.91 (3H, s, IrNC-CH3); 2.81 (6H, s, PyA-CH3); 1.50 (2H, s, Ir-CH2-Si); -0.27 (s, 9H, Si-CH3). 13C NMR (75 MHz, CD3CN): δ = 165.7 (PyA-C6); 164.8 (PyA-C2); 162.8 (PyB-C6); 158.9 (PyB-C2); 140.8 (PyA-C4); 140.1 (PyB-C4); 128.5 (PyA); 127.9 (PyB); 122.5 (PyA); 120.2 (PyB); 74.4 (PyB-CH2-N); 71.3 (PyA-CH2-N); 27.2 (PyA-CH3); 27.0 (PyB-CH3); 5.8 (IrNCCH3); -20.3 (Ir-CH2-Si); the IrNCCH3 and Si-CH3 signals are obscured by the solvent signal. FAB+-MS: m/z = 798.2 [M-PF6

–]+, 672.3 [M-2PF6–+F-]+, 631.2 [M-2PF6

–+F–-MeCN)]+. DFT calculations

The geometry optimizations were carried out with the Turbomole program24 coupled to the PQS Baker optimizer.25 Geometries were fully optimized as minima or transition states at the BP8626 level using the SV(P) basis set27 on all atoms (small-core pseudopotential28 on iridium). All stationary points were characterized by vibrational analysis (numerical frequencies); ZPE and thermal corrections (entropy and enthalpy, 298 K, 1 bar) from these analyses are included. The thus obtained (free) energies (kcal/mol) are reported in Scheme 3. The orbital and spin density plots shown in Figure 1 were generated with Molden.29 5.5 Acknowledgements

We thank Jan Meine Ernsting and Jan Geenevasen for their assistance with the NMR experiments, Han Peeters for measuring FAB+-MS mass spectra, Taasje Mahabiersing for his assistance with the kinetic measurements and Prof. Dr. Peter H. M. Budzelaar for helpful tips and tricks concerning the DFT calculations. 5.6 Notes and References

1 (a) Schrock, R. R. Chem. Rev. 2009, 109, 3211-3226. (b) Nolan, S. P.; Clavier, H. Chem. Soc. Rev. 2010, 39, 3305-3316.

2 (a) Doyle, M. P.; Forbes, D. C. Chem. Rev. 1998, 98, 911-935. (b) Kirmse, W. Angew. Chem. Int. Ed. 2003, 42, 1088-1093. (c) Wee, A. G. H. Curr. Org. Synth. 2006, 3, 499-555 (d) Davies, H. M. L.; Hedley, S. J. Chem. Soc. Rev. 2007, 36, 1109-1119.

3 Doyle, M. P.; Duffy, R.; Ratnikov, M.; Zhou, L. Chem. Rev. 2010, 110, 704-724. 4 Jellema, E.; Jongerius, A. L.; Reek J. N. H.; de Bruin B. Chem. Soc. Rev. 2010, 39, 1706-1723. 5 For references see Chapter 6 of this Thesis. 6 (a) Niimi, T.; Uchida, T.; Irie, R.; Katsuki, T. Tetrahedron Lett. 2000, 41, 3647-3651. (b) Niimi, T.;

Uchida, T.; Irie, R.; Katsuki, T. Adv. Synth. Catal. 2001, 33, 79-88. 7 (a) Chen, Y.; Fields, K. B.; Zhang, X. P. J. Am. Chem. Soc. 2004, 126, 14718-14719. (b) Chen, Y.;

Zhang, X. P. J. Org. Chem. 2007, 72, 5931-5934. (c) Chen, Y.; Ruppel, J. V.; Zhang, X. P. J. Am. Chem. Soc. 2007, 129, 12074-12075. (d) Zhu, S.; Ruppel, J. V.; Lu, H.; Wojtas, L.; Zhang, X. P. J. Am. Chem. Soc. 2008, 130, 5042-5043. (e) Zhu, S.; Perman, J.A.; Zhang, X.P. Angew. Chem. Int. Ed. 2008, 47, 8460-8463. (f) Ruppel, J. V.; Gauthier, T. J.; Snyder, N. L.; Perman, J. A.; Zhang, X. P. Org. Lett. 2009, 11, 2273-2276.

8 Krumper, J. R.; Gerisch. M.; Suh, J. M.; Bergman, R. G.; Tilley, T. D. J. Org. Chem. 2003, 68, 9705-9710.

9 For examples of Co and Rh catalyzed reactions, see reference 2a. 10 Recently an IrIII(salen) complex proved as an efficient catalyst for cyclopropanation of a broad scope of

olefins: (a) Kanchiku, S.; Suematsu, H.; Matsumoto, K.; Uchida, T.; Katsuki, T. Angew. Chem. Int. Ed. 2007, 46, 3889-3891. (b) Suematsu, H.; Kanchiku, S.; Uchida, T.; Katsuki, T. J. Am. Chem. Soc. 2008, 130, 10327-10337.

11 For reviews about mononuclear RhII and IrII species, see: (a) Hetterscheid, D. G. H.; Grützmacher, H.; Koekoek, A. J. J.; de Bruin, B. Progr. Inorg. Chem. 2007, 55, 247-354. (b) B. de Bruin, D.G.H. Hetterscheid, Eur. J. Inorg. Chem. 2007, 211-230. (c) DeWit, D. G. Coord. Chem. Rev. 1996, 147, 209-246. (d) Pandey, K. K. Coord. Chem. Rev. 1992, 221, 1-42.

12 (a) Hetterscheid, D. G. H.; Hendriksen, C.; Dzik, W. I.; Smits, J. M. M.; van Eck, E. R. H.; Rowan, A. E.; Busico, V.; Vacatello, M.; Van Axel Castelli, V.; Segre, A.; Jellema, E.; Bloemberg T. G.; de Bruin, B. J.


97

Am. Chem. Soc. 2006, 128, 9746-9752. (b) Jellema, E.; Budzelaar, P. H. M.; Reek, J. N. H.; de Bruin B. J. Am. Chem. Soc. 2007, 129, 11631-11641.

13 Zhang, L.; Chan, K. S. Organometallics, 2007, 26, 679-684. 14 (a) Hetterscheid, D. G. H.; J. Kaiser; Reijerse, E. J.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Smits,

J. M. M.; de Gelder, R.; de Bruin B. J. Am. Chem. Soc. 2005, 127, 1895-1905. (b) Hetterscheid, D. G. H.; Bens, M.; de Bruin B. Dalton Trans. 2005, 979-984.

15 Zhai, H.; Bunn, A.; Wayland, B. B. Chem. Comm. 2001, 1294-1295. 16 Chan, K.S.; Li, X.Z.; Dzik, W.I.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 2051-2061. 17 Hybrid HF/DFT functionals (such as b3-lyp) tend to overestimate the relative stability of radicals

compared to closed-shell systems: (a) Saeys, M.; Reyniers, M.-F.; Marin, G. B.; Van Speybroeck, V.; Waroquier, M. J. Phys. Chem. A 2003, 107, 9147-9159. (b) Sustmann, R.; Sicking, W.; Huisgen, R. J. Am. Chem. Soc. 2003, 125, 14425-14434. (c) Jensen, K. P.; Ryde, U. J. Phys. Chem. A 2003, 107, 7539-7545. (d) Gosh, A. J. Biol. Inorg. Chem. 2006, 11, 712-724.

18 (a) Thewissen, S. PhD Thesis, Nijmegen, 2005. (b) Budzelaar, P. H. M.; Blok, A. N. J. Eur. J. Inorg. Chem. 2004, 11, 2385-2391.

19 Benchmark studies of 5d transition metals are lacking, but for 3d and 4d transition metals hybrid HF-DFT methods such as b3-lyp or B3LYP tend to underestimate the metal-ligand bond strengths. Deviations from experimental values are generally much smaller for pure DFT methods, such as BP86: (a) Harvey, J. N. Annu. Rep. Prog. Chem. Sect. C 2006, 102, 203-226. (b) Schultz, N. E.; Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2005, 109, 11127-11143. (c) Furche, F.; Perdew, J. P. J. Chem. Phys. 2006, 124, 044103(1)-044103(27). (d) Waller, M. P.; Braun, H.; Hojdis, N.; Bühl, M. J. Chem. Theory Comput. 2007, 3, 2234-2242. See also ref. 17c and 17d.

20 Cohen, R.; Rybtchinski, B.; Gandelman, M.; Rozenberg, H.; Martin, J. M. L.; Milstein, D. J. Am. Chem. Soc. 2003, 125, 6532-6546.

21 Ikeno, T.; Iwakura, I.; Yamada, T. J. Am. Chem. Soc. 2002, 15152-15153. 22 Sierra, M. A.; Gómez-Gallego, M.; Martínez-Álverez, R. Chem. Eur. J. 2007, 13, 736-744, and references

therein. 23 (a) Tumanskii, B.; Sheberla, D.; Molev, G.; Apeloig, Y. Angew. Chem. Int. Ed. 2007, 46, 7408-7411. (b)

Sheberla, D.; Tumanskii, B.; Tomasik, A. C.; Mitra, A.; Hill, N. J.; West, R.; Apeloig, Y. Chem. Sci. 2010, 1, 234-241 and references therein.

24 Ahlrichs, R.; Turbomole Version 5, 2002, Theoretical Chemistry Group, University of Karlsruhe. 25 PQS version 2.4, 2001, Parallel Quantum Solutions, Fayettevile, Arkansas (USA); the Baker optimizer is

available separately from PQS upon request; Baker, J. J. Comput. Chem. 1986, 7, 385-395. 26 (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098-3100. (b) Perdew, J. P. Phys. Rev. B 1986, 33, 8822-8824. 27 Schäfer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571-2577. 28 (a) Turbomole basisset library, Turbomole Version 5, see 24. (b) Andrae, D.; Haeussermann, U.; Dolg,

M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123-141. 29 Schaftenaar, G.; Noordijk, J. H. J. Comput. - Aided Mol. Des. 2000, 14, 123-134.

Chapter 6

‘Carbene Radicals’ in CoII(por)-Catalyzed Olefin Cyclopropanation*

* Part of this work has been published: Reproduced in part with permission from [Dzik, W. I.; Xu, X.; Zhang, X. P.; Reek, J. N. H.; de Bruin, B. J. Am. Chem. Soc. 2010, 132, 10891-10902] Copyright [2010] American Chemical Society

Chapter 6

100

6.1 Introduction

Transition metal-catalyzed olefin cyclopropanation with diazo reagents is one of the most attractive methods to prepare functionalized cyclopropanes, which have found a myriad of fundamental and practical applications. Highly diastereo- and enantioselective catalytic systems, which are mainly based on diamagnetic CuI and binuclear Rh2- paddlewheel complexes, have been successfully developed to catalyze certain types of cyclopropanation reactions.1 While excellent results with styrenes and other electron-rich olefins have been obtained with these catalysts, their performance with electron-deficient olefins (e.g. acrylates) is, however, problematic due to the electrophilic nature of the metal-carbene intermediate.

Following the discovery that cobalt(II) complexes are capable of (stereo- and enantioselective) olefin cyclopropanation reactions,2 various cobalt-based catalytic systems have been developed. So far, the most successful cobalt-based catalysts are complexes with salen3 and porphyrin4 ligands developed by the groups of Katsuki and Zhang, respectively. Chiral cobalt-porphyrin cyclopropanation catalysts are unprecedented in their reactivity, stereocontrol, and their ability to affect cyclopropanation with (near) stoichiometric amounts of alkenes avoiding carbene dimer formation.5 Another intriguing feature of the cobalt(II) porphyrin systems is their effectiveness in cyclopropanation of electron-deficient olefins like methyl acrylate or acrylonitrile, while they perform rather poorly in the cyclopropanation of aliphatic alkenes (Scheme 1).4d,f This reactivity is remarkable, and suggests some nucleophilic character of the carbene transfer intermediate in these reactions. This is unexpected, because in analogy with Cu- and Rh-based systems one would expect the formation of electrophilic (Fischer-type) transition-metal carbene intermediates from diazo esters with relatively late transition metals like cobalt. This points to a very different character of the carbene transfer intermediate for CoII(por)-based systems compared to the generic late transition metal electrophilic (Fischer-type) carbene intermediate.

Scheme 1. Cobalt(II) porphyrin-catalyzed cyclopropanation of electron-deficient olefins.

Understanding these reactions in terms of a detailed reaction mechanism, thus explaining the above-mentioned unique reactivities and exceptional selectivities, will be important for future developments in (stereo)selective cyclopropanation reactions, especially regarding the use of diazo compounds (such as diazomalonates and diazoacetoacetates) and alkene substrates, for which high selectivities have not yet been achieved. Detailed mechanistic insights in these reactions may eventually allow us to expand the scope to other CoII(por)-mediated ring-closing reactions (e.g. formation of five-membered rings from dienes and carbenes).

The available mechanistic information so far from previously reported kinetic studies and spectroscopic investigations using CoII(salen) and CoII(por) systems is fragmented and susceptible to different mechanistic interpretations. Reported DFT studies are restricted to simplified models of CoII(salen) systems and provide only limited information about small substrate models (ethene and diazoacetaldehyde).6 As a result, questions regarding the (electronic) structures of the proposed cobalt carbene

‘Carbene Radicals’ in CoII(por)-Catalyzed Olefin Cyclopropanation

101

intermediates and transition states, as well as an estimate of the energy barriers on the basis of computational studies with realistic models, are unanswered. Especially, a rationale behind the remarkable activity of CoII(por) systems towards electron-deficient olefins is an important question to answer. Clearly, the mechanism of CoII-catalyzed cyclopropanation is poorly understood, and requires detailed attention.5

In this Chapter we address the above questions in a combined experimental and computational approach. EPR spectroscopic studies and complementary DFT-EPR-property calculations allowed us to detect and characterize the elusive carbene-transfer intermediate operating in CoII(por) systems, providing valuable new information about its unusual (electronic) structure. In addition, we performed a full mechanistic DFT study of the CoII(por)-catalyzed cyclopropanation of methyl acrylate, styrene and propene with methyl diazoacetate.

However, before we describe the details of our studies, let us first summarize the most relevant experimental mechanistic information available from previous reports in the next section (section 6.2). 6.2 Background and available experimental mechanistic information

Some kinetic and spectroscopic investigations of CoII(TPP) (TPP = tetraphenylporphyrin) catalyzed styrene cyclopropanation by ethyl diazoacetate (EDA) have been reported by Cenini and coworkers.7 This system performed relatively well in the cyclopropanation of styrenes, while its reactivity towards relatively electron-rich aliphatic alkenes was rather poor. The kinetic studies revealed a first order rate dependence on [catalyst], [EDA] and [styrene]8 concentrations.9 The authors claimed the IR and NMR spectroscopic observation of a relatively stable ‘carbene’ species CoII(TPP)(CHCOOEt), formed upon addition of EDA to CoII(TPP), in which the :CHCOOEt moiety ends up bridging between the metal and a pyrrole nitrogen atom.7 In general, carbene moieties can bind to transition metal porphyrin complexes in two ways: (1) classic coordination of the carbene carbon solely to the transition metal, where this ‘terminal carbene’ is generally thought to result in a metal-carbon double bond; and (2) as a ‘bridging carbene’ between the metal and a pyrrolato nitrogen of the porphyrin moiety, formed by insertion of the ‘terminal carbene’ into the M−N bond (see Figure 1).10-15

Figure 1. Possible coordination modes of carbenes to transition metal porphyrin complexes.

The IR detection of the ‘carbene’ species, CoII(TPP)(CHCOOEt) was independently reported by Yamada and coworkers,16 although with a different carbonyl stretch frequency (νCO = 1597 cm–1 in DCM) than the one reported by Cenini and coworkers7 (νCO = 1722 cm–1 in benzene). Yamada and coworkers assigned the IR signals to a ‘terminal carbene’ species, albeit with a cobalt-carbon single bond, and proposed that this species has substantial unpaired spin density delocalized over the carbene carbon

Chapter 6

102

and its neighboring carbonyl moiety, in analogy with the ‘terminal carbene’ species proposed to be formed from CoII(salen) and methyl diazoacetate (based on IR spectroscopy and complementary DFT calculations).17 A similar mechanism was proposed by Zhang and coworkers in their original report on the discovery of CoII(por)-catalyzed olefin cyclopropanation.4a For IrII complexes, we demonstrated experimentally and computationally that carbene ligands generated from diazo esters like EDA at low-spin open-shell d7 group 8 transition metals indeed behave as so-called ‘redox non-innocent’ ligands.18,19 In analogy, the elusive ‘terminal carbene’ CoII(por)(CHCOOEt) species are perhaps not true ‘carbenes’ in the classic sense, but rather carbon centered ‘carbene radicals’ (see Scheme 2). This clearly adds to the complexity of our understanding of the (electronic) structure of ‘carbene’ intermediates in CoII(por)-mediated cyclopropanation reactions (vide infra).

Scheme 2. Redox non-innocent behavior of carbene ligands in open-shell d7 group 8 transition metal complexes.

The proposed CoII(por)(CHCOOEt) ‘carbene’ species could not be isolated, because a fast reaction with additional EDA immediately produced diethyl maleate and (at higher EDA concentrations) the catalytically inactive diamagnetic species CoIII(TPP)(CH2COOEt).7 The latter species was characterized by X-ray diffraction and must be formed from CoII(TPP)(CHCOOEt) by hydrogen atom abstraction from EDA or the solvent, indeed pointing to a significant radical character of the ‘carbene’ species. In the presence of styrene, no intermediates could be detected at all, and EDA dimerization was suppressed. Addition of the radical scavenger TEMPO (TEMPO = 2,2,6,6-tetramethylpiperidine N-oxide) substantially slowed down the cyclopropanation reaction, but no irreversible reaction occurred between CoII(TPP) and TEMPO.7,20-22 Hence, an intermediate with substantial unpaired spin density at one of the organic moieties must play an important role in the mechanism. 6.3 Results and discussion

6.3.1 EPR spectroscopy and ESI-MS spectrometry

We decided to study the reaction of CoII(por) species with ethyl diazoacetate (EDA) by EPR spectroscopy and ESI mass spectrometry in an attempt to detect and characterize the putative CoII(por)(CHCOOR) ‘carbene’ species. Detection of this species with EPR spectroscopy in combination with complementary DFT calculations should allow us to characterize their electronic structure in detail. Quite remarkably, despite the proposed unusual carbene radical character of these CoII(por)(CHCOOR) species,4a,16,17 their detection with EPR spectroscopy has not been reported.

Hence, in an attempt to investigate the electronic structure of the elusive carbene adducts, we investigated the reaction of the porphyrin complexes Co(TPP) and Co(3,5-DitBu-ChenPhyrin)4b (Figure 2) with EDA by EPR spectroscopy. Both these complexes are active catalysts for olefin cyclopropanation, but Co(3,5-DitBu-ChenPhyrin) was shown to be much more active and selective.4


103

Figure 2. Cobalt(II) porphyrin complexes used in this study.

A solution of Co(TPP) in toluene at 40 K gave exactly the same EPR spectrum as reported previously by Van Doorslaer & Schweiger.23 Quite remarkably, addition of 4 eq. of EDA to the toluene solution at RT led to instant and complete disappearance of the EPR signals (measured in the range of 5–70 K). Although we cannot exclude that some diamagnetic material was being formed in this reaction (e.g. formation of CoIII(TPP)(CH2COOEt) by hydrogen atom abstraction7), this does not seem to be the main reason for the EPR silence, considering the clear observation of [Co(TPP)(CHCOOEt)]+ species representing the dominant signals in the ESI-MS spectrum of the same batch (vide infra). This was confirmed by measuring 1H NMR spectra of Co(TPP) directly after addition of 4 equivalents of EDA in benzene-d6, which show only the presence of paramagnetically broadened Co(TPP)-derived signals, free EDA, diethyl maleate, and minute amounts of the diamagnetic species CoIII(TPP)(CH2COOEt), in agreement with previous observations.7 Hence it seems that the Co(TPP)(CHCOOEt) radical species (irrespective of its exact structure) is EPR silent. The EPR silence can perhaps be explained by the presence of one or more exited states with energies close to the ground state causing rapid electron spin relaxation, in analogy with arguments proposed to explain the EPR silence of IrII(por) species.24 However, the exact reason for the EPR silence of Co(TPP) samples in the presence of EDA is presently not clear and contrasts markedly with the results obtained with CoII(3,5-DitBu-ChenPhyrin) described below. Co(3,5-DitBu-ChenPhyrin) shows a quite complex EPR spectrum in toluene at 40 K (Figure 3), indicating the presence of 2-3 paramagnetic cobalt species. EPR parameters of cobalt porphyrins are highly influenced by their ligand surroundings,25 and it is known that Co(TPP) forms 1:1 and 1:2 adducts with toluene.23 The approx. gx,y > 3 and gz < 2 and large ACo values observed for Co(3,5-DitBu-ChenPhyrin) in toluene are indicative of the expected square planar cobalt porphyrin complex, and agree qualitatively with the reported data for Co(TPP) in toluene.

N

N N

N

Co

N

N N

N

Co

HNO

HNO

NHO

NHO

H

H H

H

(S)(S)

(S)(S)

Co(TPP) Co(3,5-DitBu-ChenPhyrin)

Chapter 6

104

1000 2000 3000 4000 5000

g-value

without EDAd

X''/

dB

B [Gauss]

EDA added

8 6 4 3 2

Figure 3. X-band EPR spectrum of [Co(3,5-DitBu-ChenPhyrin)] before (top) and after (bottom) addition of EDA. Spectra were recorded in frozen toluene at 40 K.

Addition of 4 eq. of EDA to the toluene solution at RT resulted in a considerable change of the spectrum (measured at 40 K), and revealed full conversion from a 4-coordinate complex to a mixture of axially coordinated Co(3,5-DitBu-ChenPhyrin)(L) species. The spectrum could be simulated (Figures 3-5) as a mixture of three species I, II and III in a rough ratio of 5 to 1 to 0.8 (as derived from the spectral simulations).26 The simulated EPR parameters of these species are listed in Table 1.

The g and ACo values of species II are in the range of strong field carbon adducts (CO, isocyanides) of CoII porphyrins,27 whereas the signals of species I are indicative of coordination of a weak field ligand (comparable to H2O).23 Therefore, we assign the signals of species I to the EDA adduct CoII(3,5-DitBu-ChenPhyrin)(EDA), with EDA coordinated to cobalt either via its carbon, carbonyl or dinitrogen moiety. The signals of species II can be assigned to the ‘bridging carbene’ species CoII(3,5-DitBu-ChenPhyrin)(CHCOOEt) (Figure 1, right).

Remarkably, the EPR parameters of the third species III are indicative for an ‘organic radical’ (Figure 5, bottom), but the simulations also reveal resolved hyperfine couplings with cobalt and a proton. We therefore assign these signals to the ‘terminal carbene’ species CoII(3,5-DitBu-ChenPhyrin)(CHCOOEt), which we take as the first direct experimental evidence for its carbon-centered radical character (Figure 1, left). It thus seems that the ‘bridging carbene’ II and ‘terminal carbene’ III isomeric forms of CoII(3,5-DitBu-ChenPhyrin)(CHCOOEt) exist in dynamic equilibrium with each other in solution.26

The EPR parameters of the ‘bridging carbene’ species CoII(por)(CHCOOMe) (II) and the ‘terminal carbene’ species CoII(por)(CHCOOMe) (III) were also calculated with DFT methods, using the non-substituted porphyrin ring (por), and methyl esters (from MDA) instead of ethyl esters. The DFT data of these simplified models are in good qualitative agreement with the experimental data of species II and III, respectively (Table 1). DFT calculated EPR parameters of methyl diazoacetate (MDA) adducts of


105

CoII(por) do not agree at all with the experimental parameters of species I, but this is not surprising because it is known that classical DFT approaches do not perform well in predicting the EPR parameters of undistorted Co porphyrin systems.28 However, the EPR parameters of species I most likely belong to an axial adduct with a weak field ligand, because its EPR parameters qualitatively agree with those reported for Co(TPP)(H2O).23

2400 2800 3200 3600 4000

g-valued

X''/

dB

B [Gauss]

3 2.5 2

exp

sim

Figure 4. Experimental and simulated X-band EPR spectra of the species formed upon addition of EDA to a toluene solution of Co(3,5-DitBu-ChenPhyrin). Experimental conditions: T = 40 K, frequency = 9.377430 GHz, modulation amplitude = 5 gauss, power = 2 mW.

2400 2800 3200 3600 4000

g-value

dX''/

dB

B [Gauss]

3 2.5 2

I

II

III

Figure 5. Individual components I, II and III contributing to the EPR spectrum shown in Figure 4.

Chapter 6

106

Table 1. Experimental(a) and DFT calculated(b) EPR parameters.(c)

gx gy gz ACox ACo

y ACoz AH

x AHy AH

z

I (a) 2.558 2.520 2.004 255 208 270 - - - Co(TPP)(H2O) (e) 2.505 2.505 2.014 265 265 315 II (a) 2.228 2.120 2.005 110 115 300 - - - Co(por)(CHCOOMe) ‘bridging carbene’ (b)

2.332 2.183 2.070 −39 234 491 - - -

III (a) 2.060 2.048 2.030 40 56 nr(d) 160 40 nr(d) Co(por)(CHCOOMe) ‘terminal carbene’ (b)

2.010 2.002 1.975 −67 −19 6 −77 −50 −13

(a) Parameters from spectral simulations. (b) Orca, b3-lyp/TZVP. (c) Hyperfine couplings in MHz. (d) nr = not resolved (e) taken from reference 23.

The same reaction mixtures used for the above EPR investigations were also analyzed with electrospray ionization mass spectrometry (ESI-MS). The ESI-MS spectrum of Co(TPP) exposed to 4 eq of EDA in toluene reveals a major signal corresponding to [Co(TPP)(CHCOOEt)]+ (m/z = 757.20), along with very low intensity signals corresponding to [Co(TPP)]+ (m/z = 671.10) and [Co(TPP)(CHCOOEt)2]

+ (m/z = 843.30).

Similar results were obtained for CoII(3,5-DitBu-ChenPhyrin). Major peaks of [Co(3,5-DitBu-ChenPhyrin)(CHCOOEt)]+ (m/z = 1425.7) were detected in both toluene and CD2Cl2. Peaks corresponding to the alkyl species [Co(3,5-DitBu-ChenPhyrin)(CH2COOEt)]+ (m/z = 1426.7) in toluene and [Co(3,5-DitBu-ChenPhyrin)(CHDCOOEt)]+ (m/z = 1427.7) in CD2Cl2 were also observed. They are indicative for hydrogen atom abstraction from the solvent, in agreement with the radical deactivation pathway of the Co(TPP)(CHCOOEt) carbene species reported by Cenini and coworkers.7 The spectrum in CD2Cl2 revealed also a strong signal corresponding to [Co(3,5-DitBu-ChenPhyrin)(CHCOOEt)2]

+ (m/z = 1511.7). This signal was also present in toluene, but in lower intensity.

The observed 2:1 carbene:Co adducts likely have a comparable structure to that reported for [CoIII(OEP)(CHCOOEt)2]NO3 (OEP = octaethylporphyrin) (i.e. with two ‘carbene’ moieties bridging between the cobalt and pyrrole nitrogen atoms, and located on opposite sides of the porphyrin ring; see Figure 11, species F’).12d However, cyclopropanation of a pyrrole double bond, insertion of the carbene into a C–H bond or a Büchner ring expansion cannot be ruled out.29 6.3.2 Electronic structures of the ‘terminal carbene‘ and ‘bridging carbene’ species

The singly occupied molecular orbitals (SOMO) and spin density plots of the ‘bridging carbene’ and ‘terminal carbene’ isomers of CoII(por)(CHCOOMe) were calculated using DFT methods and are presented in Figure 6.

According to these calculations, the ‘bridging carbene’ complex is clearly a metal centered radical, with its unpaired electron residing mainly in the cobalt 3dz2 orbital, in good agreement with the measured EPR parameters (Table 1).


107

Figure 6. Spin density (left) and SOMO (right) plots of the DFT optimized ‘bridging carbene’ (top, Mulliken spin density on Co: 99.9%) and ‘terminal carbene’ (bottom, Mulliken spin densities: Ccarbene: 81%; Ocarbonyl: 14%; Co: 9%; OMe: 4%; Hcarbene: –4%; Ccarbonyl: –3.0%) isomers of CoII(por)(CHCOOMe).

The spin distribution in the ‘terminal carbene’ complex is strikingly different. The unpaired electron resides mainly on the ‘carbene’ carbon, i.e. the α-carbon of the methyl 2-ylidene-acetate moiety, and is slightly delocalized over the neighboring cobalt and oxygen atoms. This electronic structure agrees well with the EPR spectrum of III, which is indicative of an ‘organic radical’. Hence the ‘terminal carbenes’ are best described as carbon centered radicals (or ‘carbene radicals’) rather than true transition metal carbene moieties in the classic sense.

Species II and III represent an interesting example of equilibrium between two redox isomers. The ‘bridging carbene’ II is a d7 CoII complex, while the d6 CoIII terminal carbene complex III has the unpaired electron located on the ‘carbene ligand’. This is a situation somewhat similar to the recently reported dynamic interconversion between two [Rh(trop2PPh)(PPh3)] electromers30 or valence tautomers31 of CoIII(catecholato)– CoII(semiquinato).32

Occupation of spin density on the ‘carbene’ carbon is not completely unexpected. The LUMO of an electrophilic Fischer-type carbene is carbon-centered, and allows for one-electron reduction to a corresponding carbon-centered radical with external reducing agents.33 Casey and coworkers showed that reduction of chromium pentacarbonyl

Chapter 6

108

alkoxyaryl complex by one electron led to a carbon centered radical anion, persistent at −50 °C, as evidenced by EPR spectroscopy.34 Since then, several other group 6 carbene radical anions have been reported to have nucleophilic carbon character and to undergo e.g. dimerization, coupling with electron-deficient alkenes and hydrogen atom abstraction reactions.35 So clearly Fischer-type carbene ligands are redox non-innocent and intramolecular electron transfer from the electropositive metal centre to the carbene moiety cannot be excluded as an alternative pathway for formation of ‘carbene radicals’. This is exactly what happens for the ‘terminal carbene’ species CoII(por)(CHCOOMe) (Figure 7A). A schematic representation of the relevant orbital interactions is shown in Figure 7.

The interaction with the singlet :CHCOOMe moiety pushes the energy of the cobalt dz2 orbital above the energy of the antibonding MO constructed from the carbon py orbital and cobalt dyz orbital, thus resulting in intramolecular electron transfer from cobalt to the carbene moiety (Figures 7A and 7B). Partial occupation of the Co−C π* antibonding orbital thus substantially reduces the metal-carbon bond order. Hence the ‘carbene’ species loses typical Fischer-type character, gains unusual radical reactivity, and becomes more nucleophilic. However, its open-shell radical character, putting discrete unpaired spin density on the ‘carbene’ carbon, is an important difference from any closed-shell Fischer and Schrock-type carbene descriptions.

Exactly the same electronic structure arises if the bonding is considered as a triplet carbene interacting with the CoII(por) species (Figure 7C). Formation of a Co−C σ-bonding pair from the unpaired electrons in the cobalt dσ orbital and the triplet carbene sp2 orbital leaves an unpaired electron in the py orbital, effectively generating the same carbene radical ligand. Formal oxidation state counting leads to a cobalt(III)-carbene anion radical in each case.

Figure 7. Redox non-innocent behavior of carbenes coordinated to open-shell CoII(por) species explained by a simplified MO bonding scheme.


109

The radical character of the ‘carbene’ carbon atom of Co(por)(CHCOOR) species is likely responsible for the formation of the alkyl species CoIII(por)(CH2COOEt) through hydrogen atom abstraction.7 A similar behavior was observed for RhII(TMP) (TMP = tetramesitylporphinato) which reacts with diazo compounds to form RhIII(TMP)(alkyl) species.36 In the previous Chapter we reported comparable redox non-innocent behavior of carbenes bound to paramagnetic •IrII species.18,19 The resulting IrIII(•CHR) ‘carbene radicals’ react with (Ir-coordinated) ethene to form new C–C bonds (or undergo similar hydrogen atom abstraction for more bulky ‘carbenes’).18 6.3.3 Mechanistic DFT studies

Having established a clear picture of the (electronic) structure of the Co(por)(•CHCOOR) ‘carbene radical’ species in the previous sections, we next addressed their potential as intermediates in the catalytic cyclopropanation reactions with computational DFT methods. Computational methods

The BP86 functional is generally accurate in predicting the geometries of transition metal complexes,37 and has an excellent performance, especially for 3d metal compounds.38 Extensive DFT computational studies of cobalt porphyrins and related cobalamin derivatives have illustrated that the observed Co−C bond dissociation enthalpies (BDE) and internuclear distances are best reproduced by the non-hybrid BP86. Hence for reactions in which the making and breaking of cobalt-carbon bonds plays an important role, especially in cases where open-shell intermediates are involved, the use of the BP86 functional is preferred over B3LYP.39 The BP86 functional also gave accurate predictions of the thermodynamic energies and kinetic barriers associated with catalytic chain transfer40 and degenerative radical exchange processes of radical polymerization reactions controlled by Co(por) complexes.41 The Co−C bond should play an important role in predicting the energies and barriers for the Co(por)-catalyzed cyclopropanation reactions described in this paper. We therefore consistently used the non-hybrid BP86 functional in our studies.

Generally, CoII(por) systems and their adducts CoII(por)(L) have a low-spin d7 doublet (S = ½) ground state,42 and related DFT calculations reported by Yamada and coworkers have shown that the contribution of higher spin states to the cyclopropanation mechanism in reactions mediated by Co(salen) models can be neglected6b (even at the hybrid B3LYP level, which is known to favor higher spin states stabilities43). Hence, all calculations were performed with complexes in their doublet (S = ½) spin states. The reported free energies in kcal mol–1 are obtained from the calculated internal energies at the TZVP basis set on all atoms, adjusted for the zero-point energy, entropy, and approximately for the condensed phase reference volume (see Experimental Section). We used the non-functionalized cobalt porphyrin Co(por) as a smaller model of the experimental Co(TPP) complex. In addition we included some calculations using the Co(porAmide) model containing a 2-acetamidophenyl substituted porphyrinato ligand as a smaller model of the experimental Co(3,5-DitBu-ChenPhyrin) system, in order to account for possible effects of hydrogen bonding interactions between the carbonyl moiety of MDA and the ligand amide functionalities (Figure 8).4f

Chapter 6

110

Figure 8. Cobalt porphyrin models used in the DFT studies. 6.3.3.1 Formation of the ‘carbene’ complexes from CoII(por) and MDA

We first focused on the formation of the terminal and bridging ‘carbene’ species Co(por)(•CHCOOR) from MDA and the CoII(por). Experimentally, cobalt complexes with amide functionalized porphyrins exhibited higher activities in alkene cyclopropanation reactions as compared to Co(TPP). In view of these higher activities, we also investigated the same reactions with the amide-functionalized CoII(porAmide) model, in order to study the possible effect of hydrogen bonding interactions between the amide functionality of the porphyrin and the carbonyl group of the diazoacetate on the rate of ‘carbene’ formation (Figure 9).4f

Figure 9. Relevant hydrogen bonding interactions in formation and stabilization of the ‘carbene’ species.

The calculated free energies for the reaction of MDA with Co(por) and Co(porAmide) are plotted in Figure 10.

Figure 10. Free energy changes for the reaction of EDA with CoII(por) (black) and with CoII(porAmide) involving hydrogen bonding of the carbonyl group with the amide moiety (dashed blue). Selected bond distances (Å) are presented as well.


111

The DFT calculations predict the availability of three possible MDA adducts CoII(por)(MDA) from the CoII(por) species A and MDA. The carbonyl O-bound adduct B” has the lowest energy, +4.8 kcal mol−1 uphill relative to A. The nitrogen-bound adduct B’ is slightly higher in energy (+5.1 kcal mol−1), and the carbon-bound adduct B has the highest energy (+8.9 kcal mol−1). The carbon-bound adduct B,44 is however, the only productive adduct, allowing the elimination of dinitrogen (N2) to form the radical carbene species C via a relatively high barrier transition state TS1 (+13.1 kcal mol−1 from B; +22.0 kcal mol−1 from A). This barrier is in good qualitative agreement with the relatively slow cyclopropanation kinetics of Co(TPP).4a,7,45

Once formed, the ‘terminal carbene’ species C readily collapses to form the somewhat more stable ‘bridging carbene’ species C’. The energy difference between these species is, however, not large (1.5 kcal mol−1), suggesting that these species should be in dynamic equilibrium with each other. The DFT calculations are in good agreement with the results from the EPR measurements. The calculated barriers for converting C into C’ (12.9 kcal mol−1) and vice versa (14.4 kcal mol−1) are low enough to establish a fast equilibrium at room temperature.

Hydrogen bonding between the amide N-H of Co(porAmide) and the carbonyl oxygen of MDA (Figure 9) lowers the energy of the carbon-bound adduct B, the transition state TS1 and the carbene radical C by 1.7, 3.1 and 3.1 kcal mol−1, respectively. The overall barrier for formation of species C from Co(porAmide) and MDA is lowered to 18.9 kcal mol−1, in good qualitative agreement with the experimentally observed faster cyclopropanation reactions with amide functionalized cobalt porphyrins.4b-f,45

Figure 11. Reaction of MDA with ‘bridging carbene’ species C’.

Recently, a mechanistic study on Os(TTP)-mediated cyclopropanation indicated that the trans bis-’terminal carbene’ osmium species is the active catalyst (TTP = 5,10,15,20-tetra-p-tolylporphyrinato).46 Hence, we also investigated the possibility of formation of analogous cobalt bis-carbene species (see Figure 11). In the absence of olefin, slow formation of a bis-carbene species from the ‘bridging carbene’ CoII(por)(CHCOOMe) species C’ and MDA is both kinetically possible (ΔG‡ = 25.3

Chapter 6

112

kcal mol−1) and thermodynamically favorable (ΔG° = −4.3 kcal mol−1), according to the DFT calculations. These calculations are in agreement with the detection of the species in the ESI-MS spectra in low intensities (vide supra). However, their formation under catalytic conditions in the presence of olefin is clearly kinetically disfavored from follow-up reactivity of C’ with the olefinic substrate, which proceeds with much lower barriers (vide infra). Consequently, bis-carbene species CoII(por)(CHCOOR)2 should not play an (important) role under the catalytic conditions.

The trans bis-carbene species F (−1.8 kcal mol−1) having one ‘terminal carbene’ and one ‘bridging carbene’ moiety can convert readily (ΔG‡ = 12.0 kcal mol−1) to the somewhat more stable trans bis-‘bridging carbene’ species F’ (−4.3 kcal mol−1), according to the DFT calculations. Other trans bis-carbene isomers were found higher in energy, especially the trans bis-’terminal carbene’ isomer F’’ (+27.3 kcal mol−1 relative to F’). Hence, the involvement of trans bis-’terminal carbene’ species analogous to those proposed to be involved in Os(TPP)-mediated cyclopropanation can be safely excluded in the case of Co(por) systems. 6.3.3.2 Olefin cyclopropanation via ‘carbene radical’ addition to olefins

We tried to find transition states for direct reactions of olefins with the MDA adduct B, CoII(por)(MDA), in order to investigate the possibility of cyclopropanation proceeding via C−C coupling between the diazo adduct and olefin with simultaneous dinitrogen loss from the coordinated MDA moiety, as proposed by Cenini and coworkers.7 Despite several attempts in approaching the problem with different constraint geometry variations, however, we were unable to find such transition states. Similarly, we were unable to find transition states for the reaction of the ‘bridging carbene’ species C’ with olefins. Therefore, we focused on the ‘terminal carbene’ species C as the remaining and most logical carbene transfer intermediate in the cyclopropanation mechanism. All free energies of the stationary points are referenced to the ‘bridging carbene’ species C’, which should be the dormant state species in the catalytic cyclopropanation cycle according to the DFT calculations.

We investigated the reaction of the ‘terminal carbene’ radical species C with three different types of alkenes: styrene, methyl acrylate and propene. We chose propene as a model for aliphatic alkenes which generally do not form cyclopropanes effectively in CoII(por)-catalyzed cyclopropanation with diazo compounds. Styrene is a benchmark substrate for cyclopropanation catalyzed by various transition metals, whereas methyl acrylate is an example of an electron-deficient olefin that is successfully cyclopropanated by CoII(por) catalysts, but not by closed-shell Cu and Doyle-type Rh2 catalysts.

The computed carbene transfer mechanism from the ‘terminal carbene’ species C clearly proceeds via a stepwise radical process, which contrasts with the concerted carbene transfer processes generally observed for closed-shell late transition metal catalysts in olefin cyclopropanation. The reaction involves the (irreversible) addition of the carbon-centered ‘radical carbene’ species C to the olefin, thus generating the Co(por)(CHCOOMe−CH2−CHR•) species D having its unpaired electron spin density primarily localized at the γ-carbon of the ‘alkyl’ moiety (see Figures 12 and 13). Formation of the species D is associated with relatively large barriers (TS2). These barriers are somewhat lower, but very comparable to the barrier (TS1) for formation of C, which is in excellent agreement with first order kinetics in both [EDA] and [styrene] observed experimentally in the Co(TPP)-catalyzed styrene cyclopropanation.7,47


113

Figure 12. Computed pathways for trans-cyclopropanation of propene (ΔG°PR, green), styrene (ΔG°ST, blue) and methyl acrylate (ΔG°MA, red) mediated by Co(por)(CHCOOMe). Free energies in kcal mol−1 relative to the resting state ‘bridging carbene’ species C’. Selected bond distances (Å) are presented as well.

5kcalmol-1

Figure 13. Computed pathways for cis-cyclopropanation of propene (ΔG°PR, green), styrene (ΔG°ST, blue) and methyl acrylate (ΔG°MA, red). Free energies in kcal mol−1 relative to the resting state ‘bridging carbene’ species C’. Selected bond distances (Å) are presented as well.

Chapter 6

114

Once formed, the ‘γ-alkyl radical’ type species D readily collapse to form the corresponding cyclopropanes, regenerating the starting CoII(por) species A to continue the catalytic cycle. This reaction can be described as a concerted radical type C−C bond formation with simultaneous homolysis of the Co−C bond. The overall computed carbene transfer steps are consistent with the expected bond length changes for the obtained two-step radical processes (see Figures 12 and 13).

The ring closure reactions from D are very low barrier processes (TS3 < 2.6 kcal mol−1 in all cases). The barriers are actually so low that apart from cyclopropanation no other follow-up reactivity can be expected. For this reason, despite their carbon radical character, the species C and D cannot, initiate free radical polymerization reactions.48

These TS3 barriers are even lower than the barriers expected for Cβ−Cγ bond rotation in these functionalized alkyls.49 As a result, the diastereoselectivity of cyclopropanation is predetermined in the C−C bond forming transition state TS2. In that sense, the computed two-step radical-type mechanism is not different from a concerted carbene transfer mechanism. We calculated pathways for formation of both the trans-cyclopropanes (Figure 12) and the cis-cyclopropanes (Figure 13) to gain some insight in the nature of diastereoselectivity of the CoII(por)-based catalytic systems. The geometries of the stationary points along the pathway for cis- and trans-cyclopropanation are very similar, and the differences are mainly energetic.

The two-step radical mechanism nicely explains the high activity of CoII(por) catalysts in the cyclopropanation of electron-deficient olefins like methyl acrylate, and their poor performance in cyclopropanation of electron-rich aliphatic alkenes. One-electron reduction of the redox non-innocent carbene ligand (by the CoII centre) should give the carbene carbon atom more nucleophilic character (Figure 7). Consequently, the relative energies of the transition states TS2 should depend on the electrophilicity of the olefin (that is quantified by the global electrophilicity parameter ω).50 The ω values are 0.60 eV for propene, 1.13 eV for styrene and 1.51 eV for methylacrylate,51 and correlate well with the calculated barriers. The relative stabilities of the species D are strongly dependent on the ability of the olefin R group (R = CH3, Ph, COOMe) to stabilize the γ-carbon radicals. Reported radical stabilization energies52 (RSE) are in the order: •CH(Me)CH3 (−5.6 kcal mol−1) < •CH(COOMe)CH3 (−9.6 kcal mol−1) < •CH(Ph)CH3 (−14.3 kcal mol−1) and are in good qualitative agreement with the relative stabilities of the species D (stabilization with Me < COOMe < Ph). These RSE values should partly influence TS2 as well, hence explaining the relatively low TS2 barrier for styrene (that cannot be explained by the ω value alone).

For steric reasons, the TS1cis barriers for cis-cyclopropanation (Figure 13) are higher than the TS1trans barriers for trans-cyclopropanation (Figure 12) in all cases. While the differences are very small for cyclopropanation reactions of propene and styrene, dipolar repulsion between the two carboxymethyl groups results in a quite substantial difference (2.4 kcal mol−1) between the barriers for cis- and trans-cyclopropanation of methyl acrylate. This corresponds well with the high experimental trans:cis selectivity for cyclopropanation of methyl acrylate with EDA catalyzed by D2-symmetric chiral porphyrin CoII complexes.4d

The calculated trans:cis ratios for cyclopropanation of methyl acrylate, styrene and propene using the Co(por) catalyst are shown in Table 2. The calculated values are in excellent agreement with the available experimental data. The high trans:cis ratio obtained in cyclopropanation of methyl acrylate mediated by CoII(3,5-DitBu-


115

ChenPhyrin) therefore seems to be general for CoII(por) systems, and does not seem to be caused by its porphyrin steric bulk. Table 2. DFT calculated barriers for olefin addition to radical carbene C, and calculated and experimental trans:cis ratios at room temperature.

Olefin ΔGtrans ΔGcis trans:cis Expt.

methyl acrylate 17.5 19.9 98:02 99:01[a] Styrene 18.2 18.5 68:32 75:25[b] Propene 21.3 21.5 55:45 n.d. [a] CoII(3,5-DitBu-ChenPhyrin), ref 4d; [b] CoII(TPP), ref 4a 6.3.3.3 Suppression of side product formation by ‘carbene dimerization'

One of the advantages of using CoII(por) catalysts for cyclopropanation is their markedly suppressed ‘carbene’ dimerization activity under the practical catalytic conditions. Dimerization of diazo compounds to fumarates and maleates is the main side reaction in cyclopropanation mediated by most other catalysts (e.g. closed-shell Cu and Doyle-type Rh2 catalysts). To suppress this common side reaction, slow addition of the diazo compound to the reaction mixture and/or use of excess olefins are usually required. In the case of the CoII(por)-based systems, these precautionary procedures are not necessary, as only trace amounts of carbene dimerization products are generally formed under the catalytic conditions.4,7 In the absence of olefin, CoII(por) species do mediate slow carbene dimerization from the diazo substrates. This leads almost exclusively to the formation of maleates; apparently formation of the thermodynamically favored fumarates is kinetically suppressed. We investigated these phenomena computationally to get some insight in the reasons behind the low dimerization activity and high maleate selectivity of CoII(por)-based catalytic systems.

5kcalmol-1

Figure 14. Formation of dimethyl maleate by dimerization of methyl diazoacetate. Selected bond distances (Å) are presented as well.

Chapter 6

116

The calculated pathway for dimerization of MDA reveals two similar and relatively high energy transition states TS4 (23.0 kcal mol−1) and TS4’ (23.1 kcal mol−1) for C−C bond formation between the ‘terminal carbene’ species C and MDA (Figure 14). These reactions are followed by loss of dinitrogen from the diazo compound with simultaneous formation of dimethyl maleate in both cases. Formation of the kinetic product dimethyl maleate instead of the thermodynamically more stable dimethyl fumarate is a result of the steric influence between the porphyrin ring and the attacking diazo acetate (Figure 14). We were unable to find a transition state for formation of dimethyl fumarate. Since the calculated barrier for dimer formation is higher than the barriers for radical addition of C to methyl acrylate or styrene (< 20 kcal mol−1), carbene dimerization is expected to be suppressed, which is in good agreement with experimental observations under the catalytic conditions (i.e. in the presence of olefin). 6.4. Summary and Conclusions

On the basis of the results from the combined experimental (EPR and ESI-MS measurements) and computational (DFT calculations) studies, a catalytic cycle for CoII(por)-mediated cyclopropanation of olefins is proposed to involve several unusual key intermediates as summarized in Figure 15.

Figure 15. Catalytic cycle for Co(por) catalyzed cyclopropanation of olefins.

The reaction proceeds via an unprecedented two-step radical addition-substitution pathway, in which the redox non-innocent behavior of the terminal carbene ligand in intermediate C plays a key role (Figure 15). The CoII(por) catalyst A reacts with the diazo ester compound to form a transient adduct B, which loses dinitrogen in a rate-limiting step (TS1) to form the ‘terminal carbene’ intermediate C. The ‘terminal


117

carbene’ C exists in equilibrium with the somewhat more stable ‘bridging carbene’ C’. Both C and C’ species were detected experimentally in the reaction mixture of CoII(3,5-DitBu-ChenPhyrin) with EDA by EPR spectroscopy. The electronic structure of C is noteworthy. The species is a carbon-centered radical and is best described as a one-electron reduced Fischer-type carbene. These results clearly underline the general importance of redox non-innocent ligands53 and represent a rare example of the involvement of a ligand radical in organometallic catalysis.18,19

The ‘bridging carbene’ species C’ is not capable of carbene transfer to the olefin, and appears to be a dormant state in the catalytic cycle. The cobalt-centered radical C’ has to transform back to the carbon-centered radical C to allow carbene transfer to the olefin. This proceeds via radical addition of the ‘carbene radical’ C to the C=C double bond of the olefin to form a γ-alkyl radical intermediate D. The intermediates D then readily collapses in an almost barrierless ring closure reaction (TS3) to form the cyclopropane.

Addition of the ‘terminal carbene’ C to the olefin (TS2) proceeds with comparable barrier as its formation (TS1), thus explaining first order kinetics in both substrate and catalyst. Formation of C can be accelerated by stabilization of C and TS1 via hydrogen bonding. Calculated barriers for CoII(por)-mediated carbene dimerization are higher than the highest barriers for the olefin cyclopropanation, thus explaining the suppression of carbene dimerization under catalytic conditions.

The proposed radical-type mechanism (Figure 15) agrees well with all available mechanistic and kinetic information, and readily explains the excellent performance of CoII(por)-based systems in the cyclopropanation of electron-deficient olefins. The new insights obtained from these studies shed light on how to address further selectivity issues in catalytic cyclopropanation and will aid future development of new catalytic systems for remaining difficult substrates. Furthermore, these mechanistic insights may lead to the development of new type of radical ring closure processes by CoII(por) catalysts (e.g. formation of five-membered rings from properly designed 1,4-dienes and carbene sources). 6.5. Experimental

EPR and ESI-MS

Sample preparation: Toluene was distilled under nitrogen from sodium wire; CD2Cl2 was degassed using freeze-pump-thaw method and dried over 4 Å molecular sieves. Co(3,5-DitBu-ChenPhyrin)4b and Co(TPP)54 were prepared according to published procedures. They were dissolved in toluene (cCo(ChenPhyrin) = 2.7·10–3 M; cCo(TPP) = 10–4 M) or CD2Cl2 (cCo(ChenPhyrin) = 5.4·10–3 M) and each solution was transferred into a Teflon-valved EPR tube in a N2-filled glovebox and their CW X-band EPR spectra were recorded. Co(TPP) gave exactly the same spectrum as reported by Van Doorslaer and Schweiger.23 Next, 4 eq of EDA were added under N2 atmosphere and the samples were shaken for 30-40 seconds and subsequently frozen in liquid nitrogen. Experimental X-band EPR spectra of these mixtures were recorded on a Bruker EMX spectrometer equipped with a He temperature control cryostat system (Oxford Instruments). The spectra were simulated by iteration of the anisotropic g values, (super)hyperfine coupling constants and line widths. We thank Prof. F. Neese for a copy of his EPR simulation program. After 24 hours, the samples were diluted with MeOH to the approx. concentration of 10–5 M and their ESI-MS spectra were recorded. ESI-MS measurements were performed on a Shimadzu LCMS-2010A liquid chromatograph mass spectrometer by direct injection of the diluted sample to the ESI probe.

Chapter 6

118

DFT

Geometry optimizations were carried out with the Turbomole program package55 coupled to the PQS Baker optimizer56 at the ri-DFT level using the BP8657 functional and the resolution-of-identity (ri) method.58 We used the SV(P) basis set59 for the geometry optimizations of all stationary points. All minima (no imaginary frequencies) and transition states (one imaginary frequency) were characterized by numerically calculating the Hessian matrix. ZPE and gas-phase thermal corrections (entropy and enthalpy, 298 K, 1 bar) from these analyses were calculated. Improved energies were obtained with single point calculations at DFT/BP86 level using the Turbomole def-TZVP basis set.60 Estimated condensed phase (1 L mol–1) free energies, entropies and enthalpies were obtained from these data by neglecting the enthalpy RT term and subsequent correction for the condensed phase reference volume SCP

= SGP + Rln(1/24.5). Calculated EPR

spectra were obtained with Orca61 at the DFT, b3-lyp,62 TZVP60 level, using the Turbomole optimized geometries. 6.6. Acknowledgements

We thank Xue Xu and Prof. Dr. X. Peter Zhang for provision of Co(3,5-DitBu-ChenPhyrin) and their contributions to the chapter. 6.7 Notes and References

1 For reviews on this topic see for instance: (a) Doyle, M. P.; Forbes, D. C. Chem. Rev. 1998, 98, 911-935. (b) Kirmse, W. Angew. Chem. Int. Ed. 2003, 42, 1088-1093. (c) Wee, A. G. H. Curr. Org. Synth. 2006, 3, 499-555 (d) Davies, H. M. L.; Hedley, S. J. Chem. Soc. Rev. 2007, 36, 1109-1119.

2 (a) Nakamura, A.; Konishi, A; Tatsuno, Y; Otsuka, S. J. Am. Chem. Soc. 1978, 100, 3443-3448. (b) Nakamura, A.; Konishi, A; Tsujitani, R; Kudo, M.; Otsuka, S. J. Am. Chem. Soc. 1978, 100, 3449-3461.

3 (a) Niimi, T.; Uchida, T.; Irie, R.; Katsuki, T. Tetrahedron Lett. 2000, 41, 3647-3651. (b) Niimi, T.; Uchida, T.; Irie, R.; Katsuki, T. Adv. Synth. Catal. 2001, 33, 79-88.

4 (a) Huang, L.; Chen, Y.; Gao, G.-Y. Zhang, X. P. J. Org. Chem. 2003, 68, 8179-8184. (b) Chen, Y.; Fields, K. B.; Zhang, X. P. J. Am. Chem. Soc. 2004, 126, 14718-14719. (c) Chen, Y.; Zhang, X. P. J. Org. Chem. 2007, 72, 5931-5934. (d) Chen, Y.; Ruppel, J. V.; Zhang, X. P. J. Am. Chem. Soc. 2007, 129, 12074-12075. (e) Zhu, S.; Ruppel, J. V.; Lu, H.; Wojtas, L.; Zhang, X. P. J. Am. Chem. Soc. 2008, 130, 5042-5043 (f) Zhu, S.; Perman, J.A.; Zhang, X.P. Angew. Chem. Int. Ed. 2008, 47, 8460-8463 (g) Ruppel, J. V.; Gauthier, T. J.; Snyder, N. L.; Perman, J. A.; Zhang, X. P. Org. Lett. 2009, 11, 2273-2276.

5 For a highlight on the topic see: Doyle, M. P. Angew. Chem. Int. Ed. 2009, 48, 850-852. 6 According to these DFT calculations with small models of the actual catalysts and substrates, strictly

planar models seem to cyclopropanate ethene via a concerted carbene transfer mechanism from discrete CoII-carbene intermediates to the olefin, but more flexible analogs prefer a pathway via four-membered metallacyclic intermediate. In their earlier studies,6a the same authors described the possibility of a radical pathway involving with 5-coordinate CoII(salen)(L) for cyclopropanation with CoII(salen) species in the presence of an axially coordinating neutral ligand L: (a) Ikeno, T.; Iwakura, I.; Yabushita, S.; Yamada, T. Org. Lett. 2002, 4, 517-520. (b) Iwakura, I.; Ikeno, T.; Yamada, T. Org. Lett. 2004, 6, 949-952.

7 Penoni, A.; Wanke, R.; Tollari, S.; Gallo, E.; Musella, D.; Ragaini, F.; Demartin, F.; Cenini, S. Eur. J. Inorg. Chem. 2003, 1452-1460.

8 The order in [styrene] is somewhat more complex. A bell-shaped rate dependence on [styrene] was observed pointing to a first order dependence at low styrene concentrations. The decreased rates at higher styrene concentrations were ascribed to substrate inhibition kinetics by the authors, but catalysts poisoning by an impurity (e.g. the stabilizer in styrene) cannot be excluded, as the authors may have used non-purified styrene samples.

9 The same group recently reported first order kinetics in [catalyst] and [EDA], but zero order in [styrene] for related CoII(salen) systems, suggesting that for these systems the energy barrier for the cyclopropanation step is considerably lower than that for formation of the active carbene-transfer intermediate: Caselli, A.; Buonomenna, M. G.; de Baldironi, F.; Laera, L.; Fantauzzi, S.; Ragaini, F.; Gallo, E.; Golemme, G.; Cenini, S.; Drioli, E. J. Mol. Cat. A: Chem. 2010, 317, 72-80.

10 For a discussion on the possible binding modes and the factors that determine them see: Tatsumi, K.; Hoffmann, R. Inorg. Chem. 1981, 20, 3771-3784.


119

11 Reactions of diazo esters with CoIII(por) complexes were extensively studied by the group of Johnson.12 Most notably the reactions exclusively led to formation of ‘bridging carbene’ species, where the carbene ended-up bridging between cobalt and one of the pyrrolato nitrogen atoms of the porphyrin ring.12 Also a bridging bis-carbene adduct was characterized crystallographically.12d Terminal carbene species could not be isolated. Recent investigations on reactions of diazoacetates with CoIII(salen) complexes suggest that also in this system a bridging carbene is preferred over a terminal carbene.13

12 (a) Batten, P.; Hamilton, A.; Johnson, A. W.; Shelton, G.; Ward, D. J. Chem. Soc., Chem. Comm. 1974, 550-551. (b) Johnson, A. W.; Ward, D.; Batten, P; Hamilton, A. L.; Shelton, G.; Elson, C. M. J. Chem. Soc., Perkin Trans. I 1975, 2076-2085. (c) Johnson, A. W.; Ward, D. J. Chem. Soc., Perkin Trans. I 1977, 720-723; (d) Batten, P; Hamilton, A. L.; Johnson, A. W.; Mehendran, M.; Ward, D.; King, T. J. J. Chem. Soc., Perkin Trans. I 1977, 1623-1628.

13 Koehn, S. K.; Gronert, S.; Aldajaei, J. T. Org. Lett. 2010, 12, 676-679. 14 Related bridging carbene porphyrin complexes were also crystallographically characterized. Nickel: (a)

Chevrier, B.; Weiss, R. J. Am. Chem. Soc. 1976, 98, 2985-2990. Ruthenium: (b) Seyler, J. W.; Safford, L. K.; Fanwick, P. E.; Leidner, C. R. Inorg. Chem. 1992, 31, 1545-1547. Iron with a bridging vinylidene species: (c) Chevrier, B.; Weiss, R.; Lange, M; Chottard. J-C.; Mansuy, B. J. Am. Chem. Soc. 1981, 103, 2899-2901. (d) Latos-Grazynski, L.; Cheng, R-J; La Mar, G. N.; Balch, A. L. J. Am. Chem. Soc. 1981, 103, 4270-4272. (e) Olmstead, M. M.; Cheng, R-J; Balch, A. L. Inorg. Chem. 1982, 21, 4143-4148.

15 For Fe(por)(Cl)-nitrene complexes the formation of terminal vs. bridging nitrenes seems to be dependent on the electronic properties of the nitrene ligand: a) Mahy, J-P.; Battioni, P.; Mansuy, D.; Fischer, J.; Weiss, R.; Mispelter, J.; Morgenstern-Badarau, I.; Gans, P. J. Am. Chem. Soc. 1984, 106, 1699-1706. b) Mahy, J-P.; Battioni, P.; Bedi, G.; Mansuy, D.; Fischer, J.; Weiss, R.; Morgenstern-Badarau, I. Inorg. Chem. 1988, 27, 353-359.

16 Iwakura, I.; Tanaka, H.; Ikeno, T.; Yamada, T. Chem. Lett. 2004, 33, 140-141. 17 Ikeno, T.; Iwakura, I.; Yamada, T. J. Am. Chem. Soc. 2002, 124, 15152-15153. 18 (a) See Chapter 5 of this Thesis. (b) Dzik, W. I.; Reek. J. N. H.; de Bruin, B. Chem. Eur. J. 2008, 14,

7594-7599. 19 For detailed overviews of open-shell reactivity in open-shell group 8 organometallic chemistry, including

other examples of redox non-innocent ligands, see: (a) de Bruin, B.; Hetterscheid, D. G. H.; Koekkoek, A. J. J.; Grutzmacher, H.; Prog. Inorg. Chem. 2007, 55, 247-354. (b) de Bruin, B.; Hetterscheid, D. G. H.; Eur. J. Inorg. Chem. 2007, 2, 211-230.

20 Fantauzzi, S.; Gallo, E.; Rose, E.; Raoul, N.; Caselli, A.; Issa, S.; Ragaini, F.; Cenini, S. Organometallics 2008, 27, 6143-6151.

21 Related RhII(TMP) systems bind TEMPO reversibly (Keq ~ 104), but undergo irreversible hydrogen atom and methyl group abstraction reactions from TEMPO at higher temperatures: Chan, K. S.; Li, X. Z.; Dzik, W. I.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 2051-2061.

22 Irreversible binding of TEMPO to related IrII species to form diamagnetic TEMPO adducts has been reported: Hetterscheid, D. G. H.; Kaiser, J.; Reijerse, E.; Peters, T. P. J.; Thewissen, S.; Blok, A. N. J.; Smits, J. M. M.; de Gelder, R.; de Bruin, B. J. Am. Chem. Soc. 2005, 127, 1895-1905.

23 Van Doorslaer, S.; Schweiger, A. Phys. Chem. Chem. Phys. 2001, 3, 159-166 24 Zhai, H.; Bunn, A.; Wayland, B. B. Chem. Commun. 2001, 1294-1295. 25 Ozarovski, A.; Lee, H. M.; Balch, A. L. J. Am .Chem. Soc. 2003, 125, 12606-12614. 26 As the EPR experiments were performed by freeze-quenching the EPR tube in liquid N2, the detected

distribution of the species II and III does not necessarily reflect their equilibrium distribution. 27 Wayland, B. B.; Sherry, A. E.; Bunn, A. G. J. Am. Chem. Soc. 1993, 115, 7675-7684. 28 (a) Atanasov, M.; Daul, C. A.; Rohmer, M. M.; Venkatachalam, T. Chem. Phys. Lett. 2006, 427, 449-454;

For a more general overview on the performance of DFT in predicting EPR parameters see: (b) Neese, F. J. Biol. Inorg. Chem. 2006, 11, 702-711. (c) Neese, F. Coord. Chem. Rev. 2009, 253, 526-563.

29 Gomes, A. T. P. C.; Leão, R. A. C.; Alonso, C. M. A. ; Neves, M. G. P. M. S.; Faustino, M. A. F.; Tomé, A. C.; Silva. A. M. S.; Pinheiro, S.; de Souza, M. C. B. V.; Ferreira, V. F.; Cavaleiro, J. A. S. Helv. Chim. Acta 2008, 91, 2270-2283

30 Puschmann, F. F.; Harmer, J.; Stein, D.; Rüegger.; de Bruin, B.; Grutzmacher, H. Angew. Chem. Int. Ed. 2010, 49, 385-389.

31 For an excellent overview, see: Sato, O.; Tao, J.; Zhang, Y.-Z. Angew. Chem. Int. Ed. 2007, 46, 2152-2187.

32 (a) Poneti, G.; Mannini, M.; Sorace, L.; Saintctavit, P.; Arrio, M.-A.; Otero, E.; Criginski Cezar, J.; Dei, Angew. Chem. Int. Ed. 2010, 49, 1954-1957. (b) Dapporto, P.; Dei, A.; Poneti, G.; Sorace, L. Chem. Eur. J. 2008, 14, 10915-10918. c) Beni, A.; Dei, A.; Laschi, S; Rizzitano, M.; Sorace, L. Chem. Eur. J. 2008, 14, 1804-1813.

Chapter 6

120

33 Block, T. F.; Fenske, R. F.; Casey, C. P. J. Am. Chem. Soc. 1976, 98, 441-443. 34 Krusic, P. J.; Klabunde, U.; Casey, C. P.; Block, T.F. J. Am. Chem. Soc. 1976, 98, 2015-2018. 35 For an excellent overview on one electron activated (group 6) metal Fischer carbenes see: Sierra, M. A.

Gómez-Gallego, M.; Martínez-Álvarez, R. Chem. Eur. J. 2007, 13, 736-744. 36 Zhang, L.; Chan, K. S. Organometallics 2007, 26, 679-684. 37 Schultz, N. E.; Zhao, Y.; Truhlar, D.G. J. Phys. Chem. A 2005, 109, 11127-11143. 38 Furche, F.; Perdew, J. P. J. Chem. Phys. 2006, 124, 044103 39 See for example: (a) Jensen, K. P.; Ryde, U. J. Phys. Chem. A 2003, 107, 7539-7545. (b) Kuta, J.;

Potchkovskii, S.; Zgierski, M. Z.; Kozlowski, P. M. J. Comput. Chem. 2006, 27, 1429-1437. (c) Qi, X. J.; Li, Z.; Fu, Y.; Guo, Q. X.; Liu, L. Organometallics 2008, 27, 2688-2698.

40 (a) see Chapter 7 of this Thesis. (b) de Bruin, B.; Dzik, W. I.; Li, S.; Wayland, B. B. Chem. Eur. J. 2009, 15, 4312-4320.

41 Li, S.; de Bruin B.; Peng, C.-H.; Fryd, M.; Wayland, B. B. J. Am. Chem. Soc. 2008, 130, 13373-13381. 42 Walker, F. A. in The porphyrin handbook vol 5, 2000, 160-163 eds Kadish, K. M.; Smith, K. M.; Guilard,

R. Academic Press. 43 (a) Harvey, J. N.; Poli, R.; Smith, K. M. Coord. Chem. Rev. 2003, 238-239, 374-361. (b) Harvey, J. N.

Annu. Rep. Prog. Chem. Sect. C 2006, 102, 203-226 and references therein 44 A related C-bound EDA adduct of RhIII(por)(I) has been detected at low temperature: Maxwell, J. L.;

Brown, K. C.; Bartley, D.W.; Kodadek, T. Science 1992, 256, 1544-1547. 45 Reliable experimental energy barriers are not available. 46 Hamaker, C. G.; Djukic, J-P.; Smith, D. A.; Woo, L. K. Organometallics 2001, 20, 5189-5199. 47 The TS2 barrier is the most ‘organic’ reaction in the cyclopropanation cycle, with the least influence of

cobalt. Hence, it is possible that the TS2 barriers are somewhat underestimated by the use of the BP86 functional. The hybrid b3-lyp functional is probably better in predicting the barrier to TS2, but is not suitable for all other reaction steps in the catalytic cycle.

48 To initiate olefin radical polymerization, the barrier for cyclopropane ring closure (TS3) would have to be higher than 8 kcal mol−1, i.e. higher than the activation energy for addition of olefinic monomers to a growing radical polymer chain: Fisher, H.; Radom, L. Angew. Chem. Int. Ed. 2001, 40, 1340-1371.

49 The barrier for rotation of a C–C bond in propane molecule is 3.4 kcal/mol and we use it as the lower limit of the expected energy of rotation of the Cβ–Cγ bond in species D: Carey, F. A.; Sundberg, R. J. Andvanced Organic Chemistry Part A: Structure and Mechanisms 5th Ed. Springer 2007 p 142.

50 Parr, R. G.; v. Szentpály L.; Liu, S. B. J. Am. Chem. Soc. 1999, 121, 1922-1924. 51 (a) Das, T. K.; Banerjee, M. J. Phys. Org. Chem. 2010, 23, 148-155. (b) Domingo, L. R.; Pérez, P.;

Contreras, R. Tetrahedron 2004, 60, 6585-6591. 52 Zipse, H Top. Curr. Chem. 2006, 263, 163-189 and references therein. 53 Chirik, P. J. Wieghardt, K. Science 2010, 327, 794-795. 54 Sugimoto, H.; Kuroda, K. Macromolecules 2008, 41, 312-317. 55 Ahlrichs, R.; Turbomole Version 5, 2002, Theoretical Chemistry Group, University of Karlsruhe. 56 PQS version 2.4, 2001, Parallel Quantum Solutions, Fayettevile, Arkansas (USA); the Baker optimizer is

available separately from PQS upon request; Baker, J. J. Comput. Chem. 1986, 7, 385-395. 57 (a) Becke, A. D. Phys. Rev. A, 1988, 38, 3098-3100. (b) Perdew, J. P. Phys. Rev. B 1986, 33, 8822- 8824. 58 Sierka, M.; Hogekamp, A.; Ahlrichs, R. J. Chem. Phys. 2003, 118, 9136-9148. 59 Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571-2577. 60 Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chem. Acc. 1997, 97, 119-124. 61 Neese, F. ORCA – an ab initio, Density Functional and Semiempirical program package, version 2.7.0,

University of Bonn 2009. 62 (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B, 1988, 37, 785-789. (b) Becke, A. D. J. Chem. Phys. 1993,

98, 1372-1377. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.

Chapter 7

Hydrogen Atom Transfer in Reactions of Organic Radicals with (por)CoII• and in Subsequent

Addition of (por)Co–H to Olefins*

* Part of this work has been published: Reproduced in part with permission from [de Bruin, B.; Dzik, W. I.; Li, S.; Wayland, B. B. Chem. Eur. J. 2009, 15, 4312-4320] Copyright © [2009] WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Chapter 7

122

7.1 Introduction

From the previous chapter it becomes evident that cobalt porphyrin systems can be successfully used to promote catalytic radical-type reactions of the non-innocent carbene ligand with olefins. One of the elementary steps in this reaction is the homolytic cleavage of the cobalt-carbon bond resulting in the formation of a cyclopropane. The ease of homolytic cleavage of the weak Co–C bond, resulting in reversible formation of alkyl radicals, is also the modus operandi of the cobalamine based enzymes and this feature has been also successfully employed in polymer chemistry, opening new avenues in radical polymerizations.

Reactions of metal-centered radicals (M•) and organo-metal complexes (M–R) with oligomer radicals (•C(CH3)(X)CH2P) are implicated in several forms of controlled radical polymerization of olefins (Scheme 1).1-4

Scheme 1. Reactions of metal-centered radicals and organo-metal complexes with organic radicals. (A) Reactions of organic radicals with metal-centered radicals relevant to control of radical polymerization. (B) Exchange of organic radicals with the organic units in organo-metal complexes.

Reactions of organic radicals (•C(CH3)(X)R) with metal-centered radicals (M•) either produce an organometallic complex (M−C(CH3)(X)R) or a metal hydride (M−H) and an olefin (CH2=C(X)R) by the metallo-radical M• abstracting a β-hydrogen atom from the organic radical •C(CH3)(X)R (Scheme 1A).5

An additional process that may occur in this type of systems is the exchange of radicals in solution (•C(CH3)(X’)R’) with the organic group (•C(CH3)(X)R) in the organo-metal complex (M−C(CH3)(X)R) (Scheme 1B).6

Occurrence of these organo-radical reactions with metal complexes during the radical polymerization of a monomer such as vinyl acetate (CH2=CH(OAc); −OAc = −OC(O)Me) provides several mechanisms to attain control over the radical polymerization process. A wide range of metal-centered radicals and organo-metal complexes manifest at least a portion of the reactions shown in Scheme 1. An expanding range of transition metal species including complexes of CrI,7 MoIII,8 FeI,9 V0,10 TiIII,11 and CoII 1-4,12 have found applications in controlling radical polymerization of olefins. The best chain transfer catalysts reported so far are low spin cobalt(II) complexes4 and organo-cobalt(III) species, which function as latent storage sites for organo-radicals required to obtain living radical polymerization.1 Cobalt porphyrins reveal each of the processes shown in Scheme 1, and they are functional models to illustrate experimentally how these processes provide control over radical polymerization through both living radical polymerization and chain transfer catalysis.

Reaction of the cyanoisopropyl radical •C(CH3)2CN from AIBN (2,2’-azobisisobutyronitrile) with cobalt tetramesityl-porphyrin ((TMP)CoII•) in the presence of vinyl acetate (CH2=CH(OAc)) results in formation of transient (TMP)Co−C(CH3)2CN,13,14 and (TMP)Co−H complexes which proceed to form

Hydrogen Atom Transfer in Reactions of Organic Radicals with (por)CoΙΙ•

123

(TMP)Co−CH(OAc)CH3 as the thermodynamically favored organometallic product (equations 1-5). AIBN 2 •C(Me)2CN + N2 (1)

(TMP)CoII• + •C(Me)2CN (TMP)Co−C(Me)2CN (2)

(TMP)CoII• + •C(Me)2CN (TMP)Co−H + CH2=C(Me)CN (3)

(TMP)Co−H + CH2=CH(OAc) (TMP)Co−CH(OAc)CH3 (4)

(TMP)Co−P + •P* kp m kex

kex

(TMP)Co−P* + •P kp m

(5)

These reactions precede the onset of a living radical polymerization of vinyl acetate mediated by organo-Co(TMP) complexes.6 Detailed kinetic-mechanistic studies of this reaction system provided an unusual opportunity to observe and determine rate constants for the exchange of radicals in solution with the organic groups in organometallic species like that depicted in Scheme 1B (equation 6).6

(TMP)Co−R + •R* (TMP)Co−R* + R• (6)

Recently, combined experimental studies and DFT calculations examining the pathway and kinetics for the exchange of radicals from solution with the organo-groups in organo-cobalt tetramesitylporphyrin ((TMP)Co−R) complexes (Scheme 1B) were reported. Rapid radical exchange is an important pathway to obtain a narrow polymer molecular weight distribution in a living radical polymerization by a degenerative transfer mechanism1 that is observed for this system.3

In this Chapter we examine the formation of cobalt hydride, organo-cobalt complexes and mechanistic features of the hydrogen transfer processes shown in Scheme 1A and equations 2-4 using DFT to obtain more insights to controlled radical polymerization of olefins, and in particular catalytic chain transfer mediated by cobalt complexes. The relevance of these processes exceeds the chemistry of metal mediated radical polymerizations, as hydrogen atom transfer pathway can be an alternative for olefin insertion into an M–H bond, especially when no cis-vacant sites are available in the coordination sphere of the metal. 7.2 Results and discussion

7.2.1 DFT computational methods

Extensive DFT computational studies of cobalamin derivatives have illustrated that the observed Co−C bond dissociation enthalpies (BDE) and internuclear distances are better reproduced by the nonhybrid BP86 functional than by the widely used B3LYP functional.15,16 The hybrid B3LYP functional and related hybrid HF/DFT functionals tend to overestimate the correlation terms that stabilize open-shell species, while nonhybrid functionals like BP86 are thought to underestimate the open-shell stabilization terms. The empirical observation for the case of alkyl cobalamins15,16 and related cobalt porphyrins6 is that the nonhybrid BP86 functional gives the best representation of the Co−C (BDE) and that the B3LYP functional substantially underestimates Co−C BDE values because of overestimating radical stabilities.15,17 For these reasons we used the non-hybrid BP86 functional with the TZVP basis set on all atoms. The calculated internal energies are adjusted for zero-point energy, entropy, and

Chapter 7

124

approximately for solvent effects (see Experimental Section). The non-functionalized porphyrinato ligand (por2–) was used as a smaller model of the experimentally used tetramesitylporphyrinato (TMP2–) and tetraphenylporphyrinato (TPP2–) ligands.

7.2.2 Closed-shell descriptions of radical pairs

From a strictly theoretical point of view the DFT method may not give the most precise description of radical coupling and atom abstraction reactions. Ab initio ROHF and MCSCF/CASSCF treatments produce results that are more easily and straightforwardly interpretable as biradical properties. In contrast, DFT describes the properties of species in terms of “equivalent non-interacting electrons”, thereby excluding easy interpretation of two-electron properties such as biradical character. However, at present it is the only practical approach to study realistic models of complicated transition metal species. The broken-symmetry formalism, commonly used to describe radical reactions and properties of biradical species, generally produces very reasonable energies despite the fact that it does not correspond to pure spin states.18 In fact, even for cases where DFT produces a "closed-shell configuration", the actual system may still have considerable “biradical character”.19,20 Therefore, we use the standard broken-symmetry approach here to calculate energy profiles, but refrain from attaching much significance to the point at which DFT switches between "closed-shell" and "broken-symmetry" solutions. 7.2.3 (por)Co−R Bond Dissociation Enthalpies (BDE) (R = H, C(CH3)2CN, CH(OAc)CH3)

Calculated thermodynamic parameters for the homolytic dissociation of the (por)Co−R bond (equation 7) in (por)Co−H, (por)Co−C(CH3)2CN, and (por)Co−CH(OAc)CH3 are listed in Table 1. (por)Co−R (por)CoII• + R• (7) Table 1. Computed (BP86, TZVP, cosmo ε = 2.28; benzene) thermodynamic parameters for Co−R homolysis of (por)Co−R complexes (R = H, C(CH3)2CN and CH(OAc)CH3).

∆G° (kcal mol−1) ∆H° (kcal mol−1) ∆S° (cal mol−1 K−1)

(por)Co−H +47.2 +50.6 +11.4(por)Co−C(CH3)2CN +4.5 +14.6 +33.6(por)Co−CH(OAc)CH3 +19.1 +27.4 +28.0

The ability to obtain reliable computed thermodynamic bond dissociation parameters for (por)Co−R species requires accurate representations for both the closed-shell (por)Co−R complexes and the open-shell (por)CoII• and R• species. The calculated properties are evaluated in the following sections and compared with the available experimental data. 7.2.3.1 (por)CoII•

Unrestricted DFT calculations at the BP86 level using the TZVP basis set for all atoms correctly describes (por)CoII• as a low spin (S = ½) complex with an 2A1g ground state arising from a (dxy)

2(dz2)1(dxz,yz)4 electron configuration. The 2A1g ground state is

experimentally observed by EPR studies21 and the computed electronic structure is closely related to that obtained by using the ADF program.22


125

An interesting feature of the U-DFT computed electronic structure is that the half filled dz2 orbital is lower in energy than the filled dxz,yz orbitals. Larger unfavourable inter-electronic repulsions for electrons in dz2 orbital compared to the dxz,yz orbitals (which are expanded by mixing with the por π* orbitals) appears to be the dominant contribution in producing the unusual result.23

The effectiveness in using DFT to estimate bond dissociation energy parameters is dependent on the accuracy of the calculated energy value for (por)CoII•, which is assessed by comparison with results from experimental studies. 7.2.3.2 (por)Co−H

The DFT estimate of 50.6 kcal mol−1 for the Co−H bond dissociation enthalpy (BDE) of (por)Co−H (equation 8) (Table 1) represents one of the lower BDE values known for transition metal-hydrogen bonds.24 (por)Co−H (por)CoII• + •H (8)

The computed Co−H BDE is consistent with the inability to obtain NMR observable concentrations of (por)Co−H complexes from the reaction of (por)CoII• with H2 (PH2 < 4 atm, T = 298K) which places an upper limit of ~53 kcal mol−1 (i.e. half the BDE of H2 + 1 kcal mol−1) on the (por)Co−H BDE. Rapid hydrogen atom abstraction by (por)CoII• complexes from •C(CH3)2CN which has a C−H BDE ~50 kcal mol−1 (equation 9),14 indicates that the Co−H BDE can not be much smaller than 47 kcal mol−1 (below which value the thermodynamic contribution to the overall kinetic barrier would become > 3 kcal mol−1). •C(CH3)2CN CH2=C(CH3)CN + •H (9)

The computed (por)Co−H BDE of ~51 kcal mol−1 thus falls within the experimental limits (47 < Co-HBDE < 53), and is an indication of the reliability of the BP86 functional for DFT calculations applied to this class of complexes. 7.2.3.3 (por)Co−C(CH3)2CN and (por)Co−CH(OAc)CH3 Bond Homolysis

The thermodynamics (∆H° = 17.8 kcal mol−1, ∆S° = 23 cal K−1 mol−1) 14 and activation parameters (∆H‡ = 19.5 kcal mol−1, ∆S‡ = 12 cal K−1 mol−1)25 for (TPP)Co−C(CH3)2CN bond homolysis have been experimentally evaluated by equilibrium and dynamic measurements. The Co−C bond dissociation enthalpy for (por)Co−CH(OAc)CH3 complexes is too large for determination by the equilibrium method used for (TPP)Co−C(CH3)2CN. However, differences in the Co−R BDE values are dominated by the relative stability of the radicals (R•), which are reflected by changes in the C−H BDE values of the organic radicals. The calculated difference in the C−H BDE values for H−CH(OAc)CH3 and H−C(CH3)2CN of 8 kcal mol−1 thus places the (por)Co−CH(OAc)CH3 BDE at ~26 kcal mol−1. The computed (por)Co−C(CH3)2CN BDE of 14.6 kcal mol−1 is somewhat smaller than the measured BDE of 17.8 kcal mol−1 and the computed (por)Co−CH(OAc)CH3 BDE may be slightly overestimated at 27.4 kcal mol−1. Nonetheless, the computed Co−R BDE values are remarkably close to the

Chapter 7

126

experimental values, which provides confidence in the reliability of the computational method for this class of cobalt porphyrin complexes.

Constraint geometry optimizations of the (por)Co−R species with elongated Co−C bonds only leads to regular energy increase until complete bond dissociation to (por)CoII

and R•. Formation of (por)Co−R from (por)CoII and R• is thus clearly barrierless on the

electronic energy surface, and thus the most important factor contributing to the barrier for the radical capture must be the entropy required to bring together the organic radical and the cobalt complex. In order to find the transition states for radical capture, the free energy landscape would need to be computationally searched, which is not possible with the applied static DFT methods.26 7.2.3.4 (por)Co−R Bond Distances

The calculated Co−R bond distances for (por)Co−H, (por)Co−C(CH3)2CN, and (por)Co−CH(OAc)CH3 are in accord with expectations and fit well with the experimental data currently available (Figure 1). The calculated Co−H internuclear distance of 1.444 Å is in the range for first row transition metal M−H bond distances and close to the sum of the atomic radii. The calculated Co−C bond of 1.982 Å in (por)Co−CH(OAc)CH3 is in the range for relatively strong Co−C bonds and matches well with the mean Co−C distance of 1.98 Å reported by Stolzenberg for organo-Co(por) complexes.27 The longer Co−C distance of 2.060 Å calculated for (por)Co−C(CH3)2CN is consistent with the much smaller observed Co−C BDE.

Figure 1. Co-R bond lengths of (por)Co−H, (por)Co−CH(OAc)Me and (por)Co−C(Me)2CN. 7.2.4 Hydrogen Atom Transfer

Results of DFT (BP86) computations probing the pathway by which the separated metal-centred and organic radicals ((por)CoII•, R•) interact and ultimately react by β-hydrogen atom transfer to form (por)Co−H and the respective olefins, and the microscopic reverse through addition of (por)Co−H to the olefins to form eventually (por)Co−R species are given in Tables 2 and 3 and Figures 3 and 4. 7.2.4.1 H-atom Transfer between •C(CH3)2CN and (por)CoII•

Approach of the metal-centered radical (por)CoII• to a methyl group of •C(CH3)2CN reveals a local internal energy minimum for a close contact radical pair (Figure 3, species C). The transition state (TS1) for H-atom transfer from the cyano-isopropyl radical (CN)(CH3)2C• to (por)CoII• yielding (por)CoIII−H was readily found by first exploring energy profiles through constrained geometry optimizations with fixed Co−H and H−CH2C(CH3)CN bond distances. Starting from the close-contact radical pair C (broken-symmetry solution, <S2> = 0.817), shortening the Co−H bond resulted in a switch to a closed-shell solution (<S2> = 0) at Co−H distance less than 1.50 Å (Figure 2).


127

1.4 1.5 1.6 1.7 1.8 1.9 2.0

0

1

2

3

4

(0.72)

(0.59)

(0.82)

(0.45)

(0)

(0)

(0)

(0.26)

Rel

ativ

e E

nerg

y (k

cal/m

ol)

Co-H distance (Å)

(0)

C'

C

Figure 2. Energy profile (bp86, SV(P)) obtained with constrained geometry optimizations of species C and C’ with fixed Co−H bond distances (<S2> expectation values (SV(P)) between brackets).

Constrained elongation of the C−H bond followed by an unconstrained transition-state search revealed the transition state TS1 shown in Figure 3 and Figure 5.

The computed Co−H (1.48 Å) and C−C (1.37 Å) bond distances of the transition state (TS1) are only slightly elongated compared to the separated (por)Co−H (1.44 Å) and CH2=C(CH3)CN (1.36 Å) species. The transition state appears like a weakly interacting adduct of (por)Co−H with the olefin.

The (ZPE corrected) internal energy and enthalpy are lowered by the interaction of (por)CoII and the organic radical to form the radical pair intermediate C and TS1, but the free energy increases (∆G°298K = +2.0 kcal mol−1 and ∆G‡

298K = +3.8 kcal mol−1 for C and TS1, respectively) due to the entropy loss associated with bringing the reactants together.

In qualitative agreement with a fast experimental H-atom abstraction (kβ-H(333K) = 5 · 105 M−1 s−1),6 the calculated free energy change from the separated radicals to the transition state is low (∆G‡

333K = +4.4 kcal mol−1; approx. rate constant 9 · 109 M-1 s-1), perhaps even too low compared to the experimental value (∆G‡

333K = +10.9 kcal mol−1) estimated from the rate constant. This underestimation might be partly explained by sterics as the metal centre in the applied (por)Co model in the DFT calculations is not at all hindered contrary to the experimental (TMP)Co system. Moreover standard transition state theory (TST) becomes less reliable with small barriers.28

The reverse reaction of (por)Co−H adding with methacrylonitrile to produce separated radicals also has a relatively small activation free energy (ΔG‡

298K = +8.9 kcal/mol), which complies with the experimentally observed fast insertion reaction.

Chapter 7

128

2kcal/mol

2kcal/mol

Figure 3. Energy changes (kcal mol−1) for H-transfer from •C(CH3)2CN to (por)CoII• along with diagnostic Co−H, Co−C, C−C, and C−H internuclear distances (results taken from Table 2a). (a) Enthalpy change (∆H°) and internal energy change (∆EZPE), (b) Free energy change (∆G°298K).

Figure 4. Energy changes (kcal mol−1) for vinyl acetate insertion to (por)Co−H along with diagnostic Co−H, Co−C, C−C, and C−H internuclear distances (results taken from Table 2b). (a) Enthalpy change (∆H°) and internal energy change (∆EZPE), (b) Free energy change (∆G°298K).


129

7.2.4.2 H-atom Transfer between (por)CoIII−H and CH2=CH(OAc)

Computations of the pathway for addition of (por)Co−H to vinyl acetate (CH2=CH(OAc)) were guided by the results from the methacrylonitrile system, and applying a transition state geometry optimization revealed a single transition state (TS2, Figure 4) for the hydrogen transfer process in the vinyl acetate system. Minimization of the internal energy from TS2 resulted in the close contact radical pair G (analogous to C), for which BP86 still produced a “closed-shell” solution (<S2> = 0). Since the closed-shell and broken symmetry solutions are very close in energy even for C, this detail is probably not very relevant. Separating the radicals from the radical pair G is very fast and has only a small barrier which was not found computationally. The transition state (TS2) for the hydrogen atom transfer ((por)Co•…H…H2C(OAc)(H)C•) (Figure 4, Table 2) has structural parameters very similar to those of the separated (por)Co−H and CH2=CH(OAc) species and has features like an adduct of the metal hydride and the olefin. Table 2. Computed thermodynamic values for species relevant to β-hydrogen atom transfer to (por)CoII• and the microscopic reverse addition of (por)Co−H to olefins in reaction between (por)CoII• and •C(CH3)2CN [a].

∆E ∆EZPE ∆G°298K ∆H° ∆S°

B Co−R −15.1 −14.1 −4.5 −14.6 −32.6 A separated radicals 0.0 0.0 0.0 0.0 0.0 C biradical −3.4 −4.4 2.0 −4.0 −20.1 TS1 transition state 1.5 −1.8 3.8 −1.7 −18.4 D Co-H + olefin −0.4 −3.2 −5.1 −2.9 7.4

Table 3. Computed thermodynamic values for species relevant to β-hydrogen atom transfer to (por)CoII• and the microscopic reverse addition of (por)Co−H to olefins in reaction between (por)CoII• and CH2=CH(OAc)CH3

[a].

∆E ∆EZPE ∆G°298K ∆H° ∆S°

E Co−R −28.3 −27.2 −19.1 −27.4 −28.0 F separated radicals 0.0 0.0 0.0 0.0 0.0 G biradical −6.1 −8.4 -2.7 −8.0 −18.1 TS2 transition state −5.9 −8.9 -2.2 −9.4 −24.4 H Co−H + olefin −8.1 −11 −13.6 −10.5 10.5

[a] DFT (BP86, TZVP, cosmo ε = 2.28; benzene). Electronic energies (ΔE), ZPE (zero-point energy) corrected electronic energies (∆EZPE), standard free energies (∆G°), and standard enthalpies (∆H°) in kcal mol−1. Standard entropies (∆S°) are in cal mol−1 K−1. Condensed phase (1 L mol−1) free energies, entropies, and enthalpies were obtained from gas phase (24.5 L mol−1 at 1 bar, 298K) calculations by correction for the condensed phase reference volume. 7.2.5 Structural Features of the Intermediates and Transition States in H-transfer Reactions

The bond length changes along the reaction pathways are shown in Figures 3 and 4. The bond lengths change regularly on going from •C(CH3)2CN via TS1 to the alkene CH2=C(CH3)CN, with the expected elongation of the C−H bond, shortening of the C−C bond to form a C=C double bond and shortening of the Co−H bond to form (por)CoIII−H (Figure 3). The H3C−C and NC−C bonds of C are intermediate between those of CH2=C(CH3)CN and •C(CH3)2CN, as expected for a close-contact radical pair. The structure of TS1 appears roughly as a close contact of (por)Co−H with

Chapter 7

130

CH2=C(CH3)CN, representing a late transition state for H-atom transfer from •C(CH3)2CN to (por)CoII• or an early transition state for Co−H addition to the olefin. The Co−H bond shortens from 1.926 Å in the contact radical pair (C) to 1.479 Å in the transition state (TS1) and finally to 1.444 Å in (por)CoIII−H. The C−H internuclear distance elongates on going from the radical pair C (1.154 Å) to TS1 (1.788 Å). The C−C bond shortens on going from •C(CH3)2CN (1.505 Å) via C (1.480 Å) and TS1 (1.373 Å) to CH2=C(CH3)CN (1.356 Å). All these changes, as well as those observed along constrained search pathways, are quite regular and indicate a transition state close to the separated hydride and olefin, without any indications for a sudden switch to a charge-separated intermediate anywhere along the reaction path. The bond lengths also change regularly in the vinyl acetate system in going from CH2=CH(OAc) via TS2 and radical pair G to the •CH(OAc)CH3 radical, and reveal the expected shortening of the C−H bond, elongation of the C−C bond in transformation of •CH(OAc)CH3 into CH2=CH(OAc), and elongation of the Co−H bond along the pathway (Figure 4). The structure of TS2 is very similar to TS1, and also represents a late transition state for H-atom transfer from the radical •CH(OAc)CH3 to (por)CoII• or an early transition state for Co−H addition to the olefin.

Figure 5. Optimized geometries of the transition states of H-atom transfer from •C(CH3)2CN to (por)CoII• (TS1, left) and vinyl acetate addition to (por)Co−H (TS2, right).

The close relationship between the two transition states is clearly illustrated by the optimized geometries shown in Figure 5. The primary difference between the methacrylonitrile and vinyl acetate reaction systems is that the •C(CH3)2CN radical is stabilized by ~ 11 kcal mol−1 relative to •CH(OAc)CH3. The (por)CoII• is a highly efficient trap for the •CH(OAc)CH3 radical because of the strong (por)Co-CH(OAc)CH3 bond. The reverse dissociation of the organo-cobalt complex into radicals which occurs quickly for (por)Co−C(CH3)2CN (k(333K) ~ 2 · 102 s−1) is very slow for (por)Co−CH(OAc)CH3 (k(333K) ~ 1 · 10-5 s−1).6

7.2.6 Change of Spin State along the Reaction Coordinate

The hydrogen atom transfer from the methyl group of the radicals (S = ½) (•C(CH3)2CN, •CH(OAc)CH3) to the cobalt(II) metal-centered radical (S = ½) to form the closed-shell diamagnetic (por)Co−H and olefin species obviously involves a spin state change at some stage or trajectory along the reaction coordinate. The transition states (TS1 and TS2, Figure 5) for both reactions are calculated to be closed-shell singlets with internuclear distances similar to the diamagnetic cobalt hydride and olefin products. Although this is suggestive for closed-shell transition states, these BP86 results do not necessarily imply that the actual transition states do not have any biradical character left. As described above, identifying the position on the reaction coordinate


131

where the system switches between an open-shell and closed-shell configuration can not be confidently determined from results of DFT calculations. The BP86 computational procedure finds an open-shell singlet solution for the radical pair C between (por)CoII• and •C(CH3)2CN at a Co…H distance of 1.93 Å, which has high biradical character (<S2> = 0.796). As the cobalt approaches closer to the methyl hydrogen the system converts to a closed-shell configuration (<S2> = 0), which upon optimization converges to the somewhat higher energy (ΔE = +2.9 kcal/mol, TZVP basis set) closed-shell singlet solution C’ with a Co−H distance of about 1.6 Å. Minimization of the close-contact [(por)CoII• •CH(OAc)CH3] radical pair G at the unrestricted BP86 level produced a closed-shell singlet species, with a Co−H distance of 1.60 Å (Figure 4). It seems reasonable that the more stable radical (•C(CH3)2CN) might require a shorter distance and more complete bonding with cobalt prior to converting to the closed-shell singlet, but this (as well as interpreting TS1 and TS2 as 100% closed-shell transition states) is probably an over-interpretation of the results from DFT calculations. 7.2.7 Addition of Metal Hydrides to Olefins by Multi-step Hydrogen Atom Transfer Pathways

The most common pathway for transition metal hydrides to add to olefins is by concerted addition which uses a cis coordination site adjacent to the M−H group to bind the olefin prior to intramolecular hydrogen migration. This process is commonly referred to as “migratory insertion” (Figure 6A/B).29 Hydrogen atom transfer from the metal hydride is an alternative mechanism that has been observed for metal hydrides with small M−H bond dissociation enthalpies and when the resulting organic radical is particularly stable (Figure 6C).30

Figure 6. Mechanisms for olefin insertion into M−H bonds. (A) Concerted migratory 1,2-insertion of a cis-coordinated olefin. (B) Concerted migratory 2,1-insertion of a cis-coordinated olefin. (C) Radical type multistep addition via hydrogen atom transfer to a non-coordinated olefin.

Chapter 7

132

Hydrogenation and hydroformylation of styrene catalyzed by HCo(CO)4 was confirmed to involve a hydrogen atom transfer mechanism by observing CIDNP from radical pairs in these processes.31 The (por)Co−H species is a special case, because it does not have an accessible site adjacent to the Co−H, and a multi-step hydrogen atom transfer process is necessary for addition to the olefin at a finite rate. Computed energy changes during the addition of (por)Co−H with CH2=C(CH3)CN and CH2=CH(OAc) are given in Table 2 and 3, and illustrated in Figures 3 and 4. The multi-step process that involves hydrogen atom transfer from (por)Co−H to the olefin with formation of radicals occurs with relatively low activation free energy for both methacrylonitrile (∆G‡ ~ 8.9 kcal mol−1) which has a highly stabilized radical and vinyl acetate (∆G‡ ~ 11.4 kcal mol−1) with a relatively high energy radical. This result serves as a reminder that addition reactions of metal hydrides with unsaturated substrates could occur by a non-concerted multi-step pathway that involves an effective hydrogen atom transfer. Even when a cis coordination site adjacent to the M−H bond is readily available, a hydrogen atom transfer path should be considered as a potential alternate mechanism for M−H addition reactions. 7.3 Conclusions

DFT calculations using the non-hybrid BP86 functional produce a relatively accurate representation of the electronic structures of (por)CoII• (S = ½) and the bond dissociation thermodynamic parameters of (por)Co−H, (por)Co-C(CH3)2CN, and (por)Co−CH(OAc)CH3. DFT calculations and experimental results identify low activation free energy multi-step pathways that involve hydrogen atom transfer in both the formation of (por)Co−H from reactions between (por)CoII• and •C(CH3)2CN and for the net olefin insertion into the Co−H bond of (por)Co−H. The computed transition states for these hydrogen transfer reactions have structures suggestive of an olefin adduct of (por)Co−H. Results from this study suggest that the hydrogen atom transfer pathway could be important in numerous addition reactions of M−H complexes with unsaturated substrates. The low barrier for the multi-step single-site insertion process suggests that even for systems that do have a cis-vacant site, the radical-type insertion might compete with classic migratory insertion. First row transition metal hydrides that have smaller M−H BDE values compared to 2nd and 3rd row transition metal complexes are likely candidates for hydrogen atom transfer, particularly with olefins that form relatively stable radicals. 7.4 Experimental Section

Geometry optimizations were carried out with the Turbomole program package32 coupled to the PQS Baker optimizer33 at the ri-DFT level using the BP8634 functional and the resolution-of-identity (ri) method.35 We used the SV(P) basis set36 for the geometry optimizations of all stationary points. All minima (no imaginary frequencies) and transition states (one imaginary frequency) were characterized by numerically calculating the Hessian matrix. ZPE and gas-phase thermal corrections (entropy and enthalpy, 298 K, 1 bar) from these analyses were calculated. Improved energies were obtained with single point calculations at DFT/BP86 level using the Turbomole def-TZVP basis set.37 Estimated condensed phase (1 L mol–1) free energies, entropies and enthalpies were obtained from these data by neglecting the enthalpy RT term and subsequent correction for the condensed phase reference volume SCP

= SGP + Rln(1/24.5). Solvent corrections


133

from single point cosmo calculations (ε = 2.28; benzene) were applied for all species as well. The thus obtained (free) energies (kcal mol−1) are reported in Tables 1-3, Figure 3 and Figure 4. The ‘real’ energy εs of the (multi-determinant) open-shell singlet species C (singlet biradical) was estimated from the energy ε0 of the optimized single-determinant broken symmetry unrestricted BP86/TZVP (<S0

2> = 0.796) solution and the energy ε1 from a separate unrestricted ms = 1 calculation at the same geometry with the same functional and basis set (<S1

2> = 2.019), using the

approximate spin correction formula proposed by Yamagushi:38 2

02

1

12

002

1

SS

SSs −

−≈ εεε

Single point calculations and optimizations at the b3-lyp,39 level using the Turbomole def-TZVP basis for the reactions involving spin states changed gave poor results compared to experimentally derived energy values. 7.5 Acknowledgements

We thank Shan Li and Prof. Dr. Bradford Wayland for their contributions to the chapter. We thank Prof. Dr. Peter H. M. Budzelaar (University of Manitoba) for helpful discussions. 7.6 Notes and References

1 Wayland, B. B.; Peng, C.-H.; Fu, X.; Lu, Z.; Fryd, M. Macromolecules 2006, 39, 8219-8222. 2 (a) Wayland, B. B.; Poszmik, G.; Mukerjee, S. L.; Fryd, M. J. Am. Chem. Soc. 1994, 116, 7943-7944. (b)

Peng, C.-H.; Fryd, M.; Wayland, B. B. Macromolecules 2007, 40, 6814-6819. (c) Asandei, A. D.; Moran, I. W. J. Am. Chem. Soc. 2004, 126, 15932-15933. (d) Gridnev, A. A.; Ittel, S. D.; Fryd, M.; Wayland, B. B. Organometallics 1993, 12, 4871-4880. (e) Tang, L.; Norton, J. R. Macromolecules 2006, 39, 8229-8235.

3 Peng, C.-H.; Scricco, J.; Li, S.; Fryd, M.; Wayland, B. B. Macromolecules 2008, 41, 2368-2373. 4 Gridnev, A. A.; Ittel, S. D. Chem. Rev. 2001, 101, 3611-3659. 5 Tsou, T. T.; Loots, M.; Halpern, J. J. Am. Chem. Soc. 1982, 104, 623-624. 6 Li, S.; Peng, C.-H.; Fryd, M.; Wayland, B. B.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 40, 13373-

13381, and references therein. 7 (a) Abramo, G. P.; Norton, J. R. Macromolecules 2000, 33, 2790-2792. (b) Tang, L.; Norton, J. R.

Macromolecules 2004, 37, 241-243. 8 Le Grognec, E.; Claverie, J.; Poli, R. J. Am. Chem. Soc. 2001, 123, 9513-9524. 9 Gibson, V. C.; O'Reilly, R. K.; Wass, D. F.; White, A. J. P.; Williams, D. J. Macromolecules 2003, 36,

2591-2593. 10 Choi, J.; Norton, J. R. Inorg. Chim. Acta 2008, 361, 3089-3093. 11 Asandei, A. D.; Saha, G. Macromolecules 2006, 39, 8999-9009. 12 (a) Debuigne, A.; Caille, J.-R.; Jerome, R. Angew. Chem. Int. Ed. 2005, 44, 1101-1104. (b) Detrembleur,

C.; Debuigne, A.; Bryaskova, R.; Charleux, B.; Jerome, R. Macromol. Rapid Commun. 2006, 27, 37-41. (c) Debuigne, A.; Champouret, Y.; Jerome, R.; Poli, R.; Detrembleur, C. Chem. Eur. J. 2008, 14, 4046-4059.

13 Wayland, B. B.; Gridnev, A. A.; Ittel, S. D.; Fryd, M. Inorg. Chem. 1994, 33, 3830-3833. 14 Woska, D. C.; Xie, Z. D.; Gridnev, A. A.; Ittel, S. D.; Fryd, M.; Wayland, B. B. J. Am. Chem. Soc. 1996,

118, 9102-9109. 15 Jensen, K. P.; Ryde, U. J. Phys. Chem. A 2003, 107, 7539-7545. 16 Kuta, J.; Patchkovskii, S.; Zgierski, M. Z.; Kozlowski, P. M. J. Comput. Chem. 2006, 27, 1429-1437. 17 (a) Saeys, M.; Reyniers, M.-F.; Marin, G. B.; van Speybroeck V.; Waroquier, M. J. Phys. Chem. A 2003,

107, 9147-9159. (b) Sustman, R.; Sicking, W.; Huisgen, R. J. Am. Chem. Soc. 2003, 125, 14425-14434. 18 (a) Bill, E.; Bothe, E.; Chaudhuri, P.; Chlopek, K.; Herebian, D.; Kokatam, S.; Ray, K.; Weyhermueller,

T.; Neese, F.; Wieghardt, K. Chem. Eur. J. 2005, 11, 204-224. (b) Ray, K.; Weyhermueller, T.; Neese, F.; Wieghardt, K. Inorg. Chem. 2005, 44, 5345-5360. (c) Petrenko, T.; Ray, K.; Wieghardt, K. E.; Neese, F. J. Am. Chem. Soc. 2006, 128, 4422-4436. (d) Ray, K.; Petrenko, T.; Wieghardt, K.; Neese, F. Dalton Trans. 2007, 1552-1566.

Chapter 7

134

19 Budzelaar, P. H. M.; de Bruin, B.; Gal, A. W.; Wieghardt, K.; van Lenthe, J. H. Inorg. Chem. 2001, 40, 4649-4655.

20 (a) Staroverov, V. N.; Davidson, E. R. J. Am. Chem. Soc. 2000, 122, 186-187. (b) Staroverov, V. N.; Davidson, E. R. Int. J. Quantum Chem. 2000, 77, 316-323. (c) Staroverov, V. N.; Davidson, E. R. J. Am. Chem. Soc. 2000, 122, 7377-7385.

21 (a) Wayland, B. B.; Mohajer, D. J. Am. Chem. Soc. 1971, 93, 5295-5296. (b) Wayland, B. B.; Abd-Elmageed, M. E. J. Am. Chem. Soc. 1974, 96, 4809-4814. (c) Lin, W. C. Inorg. Chem. 1976, 15, 1114-1118.

22 Liao, M.-S.; Scheiner, S. J. Chem. Phys. 2002, 117, 205-219. 23 The electronic configuration of open-shell systems with DFT methods is a bit complicated by the fact that

DFT treats the alpha spin (spin up) and the beta spin orbitals (spin down) separately. The energy arrangement of the orbitals which are mainly built from the Co d-orbitals in [(por)CoII] is as follows:

x2-y297: -2.080 eV

94: -5.0134 eV93: -5.0147 eV

yz

z290: -5.6038 eV

88: -6.1780 eV

xz

xy

97: -2.1429 eV

93: -4.5921 eV92: -4.5985 eV

94: -3.6611 eV

91: -4.9679 eV

x2-y2

yz

z2

xz

xy

HOMO

SOMO

LUMO

alpha beta

Often the SOMO is also the HOMO, but not in this case. The alpha orbitals 94 and 93 (mainly dxz and dyz) are nicely oriented along the xz and yz axes defined by the porphyrin, and as expected are higher in energy than the dz2 orbital due to unfavourable anti-bonding π-interactions with the porhyrin. Due to the fact that the dxz and dyz orbitals interact with the porhyrin π-orbitals, they are expanded with respect to the dz2 orbital, and thus spin-pairing is easier in the dxz and dyz orbitals. In consequence beta orbitals 93 and 92 (mainly dxz and dyz) end-up lower in energy than the beta orbital 94 (mainly dz2). The beta orbitals 93 and 92 are spin paired with the alpha orbitals 94 and 93. In this way orbital 90 (mainly dz2) is actually the half-filled orbital. Indeed, a spin density plot reveals a clean dz2 orbital. In good agreement with experimental data, we can thus assign the DFT d-electron configuration of (por)CoII as follows: dxy

2, dz21,dxz

2,dyz

2, dx2-y20.

24 Uddin, J.; Morales, C. M.; Maynard, J. H.; Landis, C. R. Organometallics 2006, 25, 5566-5581. 25 Woska, D. C.; Wayland, B. B. Inorg. Chim. Acta 1998, 270, 197-201. 26 With the applied static DFT methods, there exist no methods to find transition states which exist only in

the free energy landscape, but not in the electronic energy landscape. A way to treat such problems correctly would be Car-Parrinello molecular dynamics, which is beyond the scope of this work.

27 Stolzenberg, A. M.; Cao, Y. J. Am. Chem. Soc. 2001, 123, 9078-9090. 28 (a) Laidler, K. J; King, M. C. J. Phys. Chem. 1983, 87, 2657-2664. (b) One has to take into account that

for very fast reactions solvent and diffusion effects contribute substantially to the condensed phase barriers, and thus the calculated gas phase barriers are then likely to be underestimated.

29 Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Organometallic Chemistry of Transition Metals: Principles and Use, Pt. 2. 1989, p 397.

30 (a) Eisenberg, D. C.; Norton, J. R. Isr. J. Chem. 1991, 31, 55-66. (b) Sweany, R. L.; Halpern, J. J. Am. Chem. Soc. 1977, 99, 8335-8337. (c) Sweany, R.; Butler, S. C.; Halpern, J. J. Organomet. Chem. 1981, 213, 487-492. (d) Halpern, J. Pure Appl. Chem. 1986, 58, 575-584. e) Roth, J. A.; Orchin, M. J. Organomet. Chem. 1979, 182, 299-311. f) Nalesnik, T. E.; Orchin, M. J. Organomet. Chem. 1980, 199, 265-269. (g) Ungvary, F.; Marko, L. Organometallics 1982, 1, 1120-1125. (h) Wassink, B.; Thomas, M. J.; Wright, S. C.; Gillis, D. J.; Baird, M. C. J. Am. Chem. Soc. 1987, 109, 1995-2002. (i) Shackleton, T. A.; Baird, M. C. Organometallics 1989, 8, 2225-2232.

31 Bockman, T. M.; Garst, J. F.; King, R. B.; Marko, L.; Ungvary, F. J. Organomet. Chem. 1985, 279, 165-169.

32 Ahlrichs, R.; Turbomole Version 5, 2002, Theoretical Chemistry Group, University of Karlsruhe.


135

33 PQS version 2.4, 2001, Parallel Quantum Solutions, Fayettevile, Arkansas (USA); the Baker optimizer is available separately from PQS upon request; Baker, J. J. Comput. Chem. 1986, 7, 385-395.

34 (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098-3100. (b) Perdew, J. P. Phys. Rev. B 1986, 33, 8822- 8824. 35 Sierka, M.; Hogekamp, A.; Ahlrichs, R. J. Chem. Phys. 2003, 118, 9136-9148. 36 Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571-2577. 37 Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chem. Acc. 1997, 97, 119-124. 38 (a) S. Yamanaka, T. Kawakami, H. Nagao, Y. Yamaguchi, Chem. Phys. Lett. 1994, 231, 25-33. (b) E.

Goldstein, B. Beno, K. N. Houk, J. Am. Chem. Soc. 1996, 118, 6036-6043. (c) Q. Knijnenburg, D. Hetterscheid, T. M. Kooistra, P. H. M. Budzelaar, Eur. J. Inorg. Chem. 2004, 1204-1211.

39 (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 1372-1377. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (d) Calculations were performed using the Turbomole functional "b3-lyp", which is not identical to the Gaussian "B3LYP" functional.

Chapter 8

Perspective:

Ligand Radicals in New, Catalytic Reactions

Chapter 8

138

8.1 Introduction

Unlike the uncontrolled reactivity of free organic radicals in solution, the reactivity of transition metal coordinated ‘ligand radicals’ can proceed in a highly selective manner. Several bio-enzymatic processes take effective advantage of the controllable reactivity of ligand radicals, thus allowing highly chemo-, regio-, diastereo- and enantio-selective radical-type reactions to achieve transformations that are not easy to achieve via closed-shell reaction pathways. These enzymatic radical-type reactions frequently exploit the features of so-called ‘redox non-innocent’ ligands; ligands that are easier to oxidize or reduce than the transition metal to which they coordinate. The combined, cooperative action of the radical-type reactivity of the redox non-innocent ligand and the substrate coordinating and activating properties of the transition metal, allows the enzyme to catalyze ‘difficult reactions’ with unprecedented rates and selectivities. This concept has inspired many synthetic chemists to employ similar principles in the field of organometallic chemistry, the playground of catalytic synthetic-organic transformations. Several stoichiometric open-shell organometallic reactions involve prior formation of ‘ligand radicals’, followed by ligand-centered transformations. Many of such ligand-centered radical-type reactions proceed in a controlled and selective manner, most frequently leading to metal-carbon and carbon-carbon radical-couplings. For the more sterically hindered systems selective ligand-centered hydrogen atom abstractions are a common alternative reaction pathway. These selective ligand-centered radical-type reactions are synthetically useful in organometallic synthesis, allowing interesting ligand transformations and the formation of tethered binuclear organometallic compounds, for example. Some of the observed stoichiometric transformations serve to explain unwanted side-reactions and catalyst deactivation pathways in existing (open-shell and closed-shell) catalytic reactions. Other unusual and selective radical-type ligand-ligand C−C and C−O coupling reactions are inspiring from an organic-synthetic point-of-view, and may eventually lead to applicable new catalytic transformations. In this Thesis, we have shown that synthetic catalytic reactions actually can operate via selective organometallic radical-coupling processes, thus allowing highly chemo-, regio-, diastereo- and even enantioselective turnover (see Chapter 4). New insights in the catalytic mechanism of these Co(por) mediated olefin cyclopropanation reactions revealed that the ‘redox non-innocence’ of the terminal carbene ligand plays a crucial role, and the cooperative action of the ‘carbene radical’ and the transition metal is an essential feature in the substrate activation and conversion process. These new insights clearly show that bio-inspired, enzyme-like approach can be employed in synthetic organometallic catalysis to achieve selectivity in radical-type transformations. If this is possible for radical-type olefin cyclopropanation reactions, why should it not be possible for other radical-type transformations?

Let us set our imagination free and think about the possible new reactions that could be developed on the basis of the redox non-innocence of the carbene ligand.

8.2 Broadening the scope of reactivity of carbene radicals

It is now well established that the carbene radicals allow for easy cyclopropanation of otherwise difficult (electron-poor) substrates.1 One of the possible ways to expand the scope of their reactivity could be the utilization of carbene precursors

Perspective: Ligand Radicals in New, Catalytic Reactions

139

with a conjugated double bond (Figure 1A). A properly designed precursor (i.e. leading to radicals being highly stabilized at the carbon containing the R2 group) should allow for additions of the γ carbon radical to olefinic substrates. Addition to an olefin would lead to an ε radical and subsequent ring closure could yield a cyclopentene derivative.

Another route toward highly substituted cyclopentenes could be possible with functionalized conjugated dienes (Figure 1B). This reaction would proceed via intermediates similar to the ones in pathway A and again, proper design of the reacting diene should play a key role in allowing such reactivity.

Figure 1. Theoretical radical mechanisms involving redox non-innocence of carbenes leading to highly functionalized cyclopentenes.

The calculated spin density plot of a possible intermediate from pathway A is presented in Figure 2. DFT calculations suggest that in analogy to the one-electron reduced vinylcarbenes of Cr and W,2 the radical-type reactivity should be possible on the γ carbon.

Figure 2. Calculated (Turbomole, RI-DFT, BP86, SV(P)) spin density plot of the Co(por)C(COOMe)(CHCHPh) complex showing substantial delocalization of the unpaired electron over the α and γ carbon atoms. The numbers represent the calculated spin densities on the respective atoms.

3.2Co

O

O

46.8

-18.4

49.9

-11.1

14.0

-6.7

16.7

-6.8

14.7

2.8

-3.0

Chapter 8

140

Another field in which the carbene radical complexes could be used is perhaps organometallic controlled radical co-polymerization of olefins with diazo compounds (Figure 3).3,4

Figure 3. Possible incorporation of carbene radicals into the growing polyolefin chain.

In such a hypothetical reaction, the radical carbene complex could occasionally react with the growing polymeryl radical chain built from C2 units. This would allow the incorporation of C1 units into the polyolefinic polymer. Subsequent homolytic cleavage of the M–C bond would regenerate the active polymeryl radical chain (elongated by one carbon atom of the diazo compound) and allow its further radical polymerization, just like in the standard OMRP. The metal radical could react further with either a diazo compound to form a new radical carbene, or with the growing polymer chain to form a dormant species. Combining C2 with C1 radical polymerization could lead to functionalized polymers with interesting structures and properties.

To attain high incorporation of the C1 units into the polymer chain, very reactive diazo compounds should be utilized since the barriers for formation of radical carbenes from diazoacetates are rather high1a,b and the amount of the free metal radicals during the metal mediated radical polymerization is very low.5 Thus, diazoalkanes should be good candidates as the radical carbene sources in combination with olefinic monomers which allow for easy cleavage of M–C bonds (i.e. their radicals should have high radical stabilization energy) like methyl methacrylate.

8.3 Redox non-innocence in other catalytic reactions?

Nitrenes are nitrogen analogues of carbenes (Figure 3) and could also be redox non-innocent. Metal nitrenes are involved in aziridination of olefins or C-H insertion reactions.6 Possible radical pathways were postulated,7,8 but up to now no spectroscopic evidence for the existence of nitrene radical ligands has been presented.

Figure 4. Hypothetical redox non-innocent behavior of nitrenes.

Perspective: Ligand Radicals in New, Catalytic Reactions

141

Our preliminary DFT calculations suggest that the putative tosylnitrene

intermediate9 in cobalt(II) porphyrin catalyzed aziridinations indeed has a substantial spin density located on the nitrogen ligand, showing a strong analogy with carbene radicals (Figure 4).10 Therefore it seems that detection of nitrene radicals experimentally with EPR spectroscopy could be feasible and would provide a definite proof for the redox-non innocence of the nitrene group. As a result, reactions analogous to the ring-closure reactions shown in Figure 1 could be possible with nitrene precursors, which would provide a convenient new synthetic route to prepare functionalized dihydro-pyrroles.

Figure 5. Calculated (Turbomole, RI-DFT, BP86, SV(P)) spin density plot of the Co(por)NSO2Ph complex showing substantial localization of the the unpaired electron on the nitrogen p orbital. Calculated spin densities: Co: 27.9%; N: 62.8%; S: –0.2%; O: 7.9 and 6.5%.

8.3 Conclusions

Throughout the Thesis we have shown that the concept of ligand redox-non innocence can be utilized not only in stoichiometric, but also catalytic reactions. One electron activation of the coordinated ligand leads to a dramatic change in the reactivity pathways allowing for reactions with otherwise not reactive substrates. The reactivity of radical carbenes described in Chapters 1, 5 and 6 is an archetypical example of this phenomenon. In this Perspective we propose that one can take advantage of the unusual electronic structure of the carbene radical in catalytic reactions other than cyclopropanation. Formation of five membered rings incorporating a carbene or nitrene moiety in reaction with (di)olefin or radical co-polymerization of carbenes with olefins are examples of such possible transformations.

Hence, we firmly believe that open-shell organometallic chemistry in general has a bright (catalytic) future. Much can be expected from future computational studies and conscientious spectroscopic investigations of open-shell organometallic compounds, which should provide valuable and detailed new information about their intriguing and unusual electronic structures. This will be of crucial importance for further synthetic developments, especially in the realm of catalysis.

Chapter 8

142

Notes and References

1 (a) See Chapter 6 of this Thesis. (b) Dzik, W. I.; Xu, X.; Zhang, X. P.; Reek, J. N. H.; de Bruin, B. J. Am. Chem. Soc. 2010, 132, 10891-10902. (c) Zhu, S.; Perman, J. A.; Zhang, X. P. Angew. Chem. Int. Ed. 2008, 47, 8460-8463 (d) Chen, Y.; Ruppel, J. V.; Zhang, X. P. J. Am. Chem. Soc. 2007, 129, 12074-12075.

2 Sierra, M. A.; Ramírez-López, P.; Gómez-Gallego, M.; Lejon, T.; Mancheño, M. J. Angew. Chem. Int. Ed. 2002, 41, 3442-3445.

3 For an overview of TM controlled radical polymerization of olefins, including organometallic controlled radical polymerization see for instance: di Lena, F.; Matyjaszewski, K. Prog. Polym. Sci. 2010, 35, 959-1021.

4 C1 polymerization of carbenes has been recently reviewed: Jellema, E.; Jongerius, A. L.; Reek J. N. H.; de Bruin B. Chem. Soc. Rev. 2010, 39, 1706-1723.

5 For example, the dissociation equilibrium constant at 333K for organo-cobalt complexes formed in Co(por) mediated radical polymerization of methyl acrylate was estimated as 1.15·10–10: Peng, C.-H.; Li, S.; Wayland, B. B. Inorg. Chem. 2009, 48, 5039-5046.

6 For a review about nitrene transfer reactions catalyzed by metalloporphyrins see: Fantauzzi, S.; Caselli, A.; Gallo, E. Dalton Trans. 2009, 5434-5443.

7 (a) Mahy, J.-P.; Bedi, G.; Battioni, P.; Mansuy, D. J. Chem. Soc. Perkin Trans. II 1988, 1517-1524. (b) Simonato, J.-P.; Pécaut, J.; Scheidt, W. R.; Marchon, J.-C. Chem. Commun. 1999, 989-990. (c) Au, S.-M.; Huang, J.-S.; Yu, W.-Y.; Fung, W.-H.; Che, C.-M. J. Am. Chem. Soc. 1999, 121, 9120-9132.

8 DFT studies on the mechanism of Cu catalyzed aziridinations point to the intermediacy of an open-shell (singlet or triplet) metal-nitrene species: (a) Brandt, P.; Södergren, M. J.; Andersson, P. G.; Norrby, P. -O. J. Am. Chem. Soc. 2000, 122, 8013-8020. (b) Comba, P.; Lang, C.; Lopez de Laorden, C.; Muruganantham, A.; Rajaraman, G.; Wadepohl, H.; Zajaczkowski, M. Chem. Eur. J. 2008, 14, 5313-5328.

9 Ruppel, J. V.; Jones, J. E.; Huff, C. A.; Kamble, R. M.; Chen, Y.; Zhang, X. P. Org. Lett. 2008, 10, 1995-1998.

10 This analogy can be also found in the oxo ligand of cytochrome P450 which is redox non-innocent and undergoes C–H bond insertion proceeding via a two step radical process.

Summary

143

Summary Radicals are often considered to be too reactive to be selective. For free radicals this might be true, but in the coordination sphere of a transition metal highly chemo-, regio-, diastereo- and enantioselective radical-type reactions become feasible, with many bio-enzymatic radical-type biosynthetic transformations serving as illustrative examples. The main objective of this Thesis is to show that also in organometallic chemistry radical reactions are synthetically useful, and can even lead to selective catalytic reactions. The explorative nature of the investigations that led to these conclusions further led to a number of interesting closed-shell reactivity patterns (Chapter 2 and 3).

Selective radical-type atom transfer reactions of open-shell organometallic compounds often require the presence of unpaired electron (radical) density at a reactive ligand site, and hence the so-called “ligand redox non-innocence” is an important feature in these reactions. This concept is explained in Chapter 1 on the basis of a literature survey of the intriguing open-shell organometallic chemistry of paramagnetic group 9 (Co, Rh, Ir) complexes bearing carbonyl, carbene and alkene ligands (and alike). One-electron activation of alkene, carbonyl and carbene ligands reveal a remarkable variety of selective radical-type reactions. These ligands have low lying empty orbitals that easily accept an unpaired electron from the metal in open-shell complexes. This triggers radical-type reactivity on their carbon atoms (Figure 1), which leads (among others) to formation of new C–H or C–C bonds by hydrogen atom transfer or reductive coupling to form products that cannot be obtained via closed-shell pathways.

Chapter 2 describes the synthesis and redox properties of rhodium and iridium

carbonyl complexes [MI(CO)n(Me3tpa)]+ (M = Rh, Ir, n = 1, 2) supported by the Me3tpa ligand. The iridium bis-carbonyl complex undergoes a selective, irreversible one-electron oxidation which leads to an iridium mono-carbonyl hydride complex (Figure 2). In the investigated systems the CO moiety behaves as a redox innocent ligand, and oxidation of the complexes leads to selective metal centered radical-type reactivity in the case of iridium (Figure 2). This contrasts with previous investigations of Wayland, wherein carbonyl adducts of RhII(por) were clearly shown to behave as carbon-centered radicals. Furthermore, two electron carbonyl ligand activation pathways were revealed for the closed-shell [MI(CO)n(Me3tpa)] systems. Reaction of the iridium bis-carbonyl complex with water (or methanol) results in formation of hydroxycarbonyl-hydride (or methoxycarbonyl-hydride) complexes, which allows the isolation of iridium analogs of all subsequent intermediates relevant in the water gas shift reaction (Figure 3). Reaction of the closed-shell rhodium carbonyl

M Mn n+1

M Mn n+1C C

M Mn n+1

O

O

R1

R2

R1

R2 Figure 1. Redox non-innocentbehavior of alkene, CO andcarbene ligands.

N

NN

NIr

CO

CO

N

N NN Ir

CO

H

+

I

+

III

2+

Ag+

Acetone

H2O (traces)

Figure 2. Metal centered reactivity of [Ir(CO)2(Me3tpa)] triggeredby one electron oxidation.

Summary

144

complex with dioxygen in the presence of water leads to selective formation of a carbonate complex.

In Chapter 3 we investigated the impact of the donor strength of tri- and tetradentate

bispicolylamine type ligands on the structure of rhodium carbonyls. For more electron donating bpa ligand binuclear triscarbonyl bridged species are favored over terminal carbonyls which are formed with weaker donating alkyl-bpa ligands. Interestingly, rhodium carbonyl complexes with the tetradentate trispicolyl-amine ligand exist in a dynamic equilibrium between the terminal and biscarbonyl bridged dinuclear species (Figure 4).

In Chapter 4 we investigated the possible role of aminyl radicals in the reactivity of

open-shell rhodium and iridium complexes. Unexpected one-electron reduction of [MII(dbcot)(bla)]2+ complexes (dbcot = dibenzocyclooctadiene, bla = bislutidylamine), is observed in the presence of a base (Figure 5). Control experiments with [RhII(dbcot)(Bz-bla)]2+ (Bn-bla = N-benzyl-bislutidylamine) show that these reactions can also proceed without the involvement of aminyl radicals, but rapid hydrogen atom abstraction from the solvent by an aminyl radical ligand cannot be fully excluded. In any case, these complexes did not allow the formation of detectable aminyl radical ligands upon deprotonation, thus complicating a study of their reactivity. This behavior is in marked contrast with the reported easy formation of Rh-stabilized aminyl radical ligands for structurally similar complexes, as reported by Grützmacher and coworkers. Possible reduction pathways involving base assisted oxidation of solvent or water are discussed.

Chapter 5 describes the reactivity of the paramagnetic complex

[IrII(ethene)(Me3tpa)]2+ with diazo compounds (used as carbene precursors). Here, the carbene ligand, generated at the open-shell IrII metal, clearly behaves as a redox active

N

N

N

NIr

H

O OH

CO

N

N

N

NIr

CO

CO

N

N

N

NIr

HH

CO

+

III

+

I

H2O/Acetone Acetone

+

III- CO2

Figure 3. Isolated iridium analogs of intermediates relevant in the water gas shift reaction.

N

N N

NRh

CO

N

NN

NRh

N

N N

NRh

O

O

+

I

2+

I

I2

Figure 4. Dynamic equilibrium between monouclear[Rh(3-tpa)(CO)]+ and dinuclear [Rh(4-tpa)(-CO)]2

2+.

Figure 5. Electrochemical behavior of rhodium and iridium dbcot complexes of bislutidine type ligands.

Summary

145

ligand, and to the best of our knowledge the ‘redox non-innocent’ behavior of (Fischer-type) carbenes of group 9 transition metals was demonstrated for the first time with these investigations. Radical-type C–C and C–H bond forming reactions occurring at the ‘carbene radical’ ligand and DFT investigations on their mechanism are presented in this Chapter. Based on these results we give a description of the electronic structure of one-electron activated carbenes. The ‘carbene radicals’ are best described as carbon centered radicals bound to a closed-shell transition metal centre (Figure 6).

In Chapter 6 we show that ‘carbene radicals’ are not just some chemical curiosity, but actually play an important, previously unknown role in existing catalytic cyclopropanation reactions mediated by cobalt-porphyrin complexes. Paramagnetic CoII(por) complexes are effective catalysts for olefin cyclopropanation, and the most successful cobalt-based cyclopropanation catalyst is the chiral cobalt-porphyrin CoII(3,5-DitBu-ChenPhyrin) complex developed by Zhang and coworkers (Figure 7). It is unprecedented in its reactivity, stereocontrol, and the ability to affect cyclopropanation with (near) stoichiometric amounts of alkenes avoiding carbene dimer formation. Another intriguing feature of this CoII porphyrin system is its effectiveness in cyclopropanation of electron-deficient olefins like methyl acrylate or acrylonitrile. This reactivity is remarkable, and very different from the typical electrophilic Fischer-type carbene intermediate associated with CuI and Doyle-type Rh2-based systems, thus

pointing to a very different character of the carbene transfer intermediate for CoII(por)-based systems.

In this chapter we investigated the reaction mechanism of this catalyst through EPR spectroscopic and DFT computational investigations.

The cyclopropanation mechanism clearly proceeds

via a radical-type mechanism, which involves the formation of reactive carbon-centered radicals within the coordination sphere of cobalt. We show that the key cobalt-carbene intermediate has a strong carbon radical character, and is in fact a ‘carbene radical’

N

N N

NIr

NCMe

H R

N

N N

NIr

NCMe

H R

2+

III

2+

II

Figure 6. Redox non-innocent behavior of the carbene bound to[Ir(Me3tpa)(NCMe)]2+. Resonance structures (top) and the DFTcalculated spin density (right).

N

N N

NCo

HN

O

O

O

H

N

N N

N

Co

HNO

HNO

NHO

NHO

H

H H

H

(S)(S)

(S)(S)

Co(3,5-DitBu-ChenPhyrin)

Figure 7. CoII(3,5-DitBu-Chen-Phyrin) (left) and its H-bond donation to the ‘carbene radical’ ligand in the key catalytic intermediate (top).

Summary

146

similar to the iridium-carbene radicals described in Chapter 5 (Figure 8). Its nucleophilic and radical character facilitates the cyclopropanation of electron poor olefins, and suppresses carbene dimerization activity. To our knowledge this is the first and only documented example in which ligand redox-non innocence plays a key role in an organometallic catalytic transfomation. Addition of the ‘carbene radical’ to the olefin is followed by a low-barrier transition state for cyclopropane ring closure. The latter process is best described as a radical-type C–C bond coupling with simultaneous homolytic splitting of the cobalt-carbon bond. Hence, besides the ‘redox non-innocence’ of the carbene, also the intrinsic relative weakness of the Co–C bond plays a crucial role in this reaction.

N

N N

NCo

N

N N

NCo H

H COOR

N2

H COOR

R'H

N2

H COOR

R' H

N

N N

NCo

COORNN

N

N N

NCo

H COOR

N

N N

NCo

H COOR

HR'

A further interesting feature of this catalyst is that the aryl-dicyclopropane-carboxamido functionalities of the 3,5-DitBu-ChenPhyrin ligand (Figure 7) are potent H-bond donors. This lowers the activation barrier for formation of the carbene radicals from the diazo ester substrate, thus leading to faster reactions and higher selectivities. With this radical-type mechanism being firmly established, CoII(por)-catalyzed olefin cyclopropanation provides an illustrative example of a synthetic catalytic reaction that operates via a radical-coupling process, but still allows highly selective turnover.

In Chapter 7 we computationally investigated

the role of cobalt metallo-radicals in catalytic chain transfer (CCT) in cobalt mediated radical polymerization reactions. The computational study focuses on the reversible hydrogen atom transfer reactions between cobalt porphyrins and organic radicals (Figure 9). The calculations give insight into the CCT mechanism. Results from this study further suggest that the hydrogen atom transfer pathway could be important in numerous reactions in which an olefin is inserted into a weak MH bond.

Figure 9. Reversible hydrogen atom transferbetween CoII and organic radicals.

Figure 8. Spin density plot of the Co(por) ‘carbene radical’ (top) and the catalytic cycle of the cyclopropanation reaction involving radical carbenes.

Summary

147

The main finding of this Thesis is that the concept of ligand redox non-innocence can be utilized not only in stoichiometric, but also in catalytic reactions. We showed that one-electron activation of the coordinated ligand results in a dramatic change in the reactivity pathways, which may also allow for reactions with otherwise unreactive substrates. We also claim that a better understanding of the behavior of unsaturated ligands bound to paramagnetic (Group 9) metals can lead to useful, new (catalytic) transformations. Thus, in the Perspective we speculate on the possible involvement of carbene radical complexes in new (catalytic) reactions, which might potentially lead to valuable new products via unprecedented catalytic pathways. We propose that reactions of radical carbenes with conjugated dienes or reactions of radical vinylcarbenes with olefins could lead to functionalized cyclopentenes. We also postulate that nitrene ligands should be redox active and thus react via radical type mechanisms.

Samenvatting

149

Samenvatting Van radicalen wordt vaak aangenomen dat ze te reactief zijn voor selectieve reacties. Voor vrije radicalen is dit misschien terecht, maar in de coördinatiesfeer van een overgangsmetaal worden zeer chemo-, regio-, diastereo- en enantioselectieve radicaalreacties mogelijk, waarvoor vele bio-enzymatische radicaal-type biosynthetische transformaties als voorbeeld dienen. De hoofddoelstelling van dit Proefschrift is aan te tonen dat ook in de organometaalchemie, radicaalreacties synthetisch nuttig zijn, en dat ze zelfs kunnen leiden tot selectieve katalytische reacties. Het exploratieve karakter van het onderzoek dat heeft geleid tot deze conclusies, heeft tevens een aantal interessante resultaten opgeleverd op het gebied van nieuwe ‘closed-shell’ reactiviteit (Hoofdstuk 2 en 3).

Selectieve radicaal-type atoomtransfer reacties van ‘open-shell’ organometaalverbindingen hebben meestal ongepaarde elektronen dichtheid (radicalen) nodig op het reactieve ligand. Het feit dat deze liganden ‘redox niet-onschuldig’ zijn is dus een belangrijke eigenschap in deze reacties. Dit concept wordt in detail uitgelegd in Hoofdstuk 1 op basis van een literatuurstudie van de intrigerende ‘open-shell’ organometaalchemie van paramagnetische groep 9 (Co, Rh, Ir) complexen met carbonyl, carbeen en alkeen (en vergelijkbare) liganden. Eén-elektron activering van alkeen, carbonyl and carbeen liganden laat een opmerkelijke variëteit van selectieve radicaalreacties zien. Deze liganden hebben laaggelegen lege orbitalen die gemakkelijk een ongepaard elektron accepteren van het metaal in ‘open-shell’ complexen. Dit triggert radicaalreactiviteit op de koolstofatomen van deze liganden (Figuur 1), en kan (onder andere) leiden tot de vorming van nieuwe C–H of C–C bindingen door waterstofatoom transfer of reductieve ligand-ligand koppelingen om producten te vormen die niet verkregen kunnen worden via ‘closed-shell’ routes.

Hoofdstuk 2 beschrijft de synthese en redox eigenschappen van rhodium en iridium

carbonyl complexen met het Me3tpa ligand ([MI(CO)n(Me3tpa)]+ (M = Rh, Ir, n = 1, 2)). Het iridium bis-carbonyl complex ondergaat een selectieve, onomkeerbare één-elektron oxidatie die leidt tot een iridium mono-carbonyl hydride complex (Figuur 2). In de onderzochte systemen gedraagt de CO groep zich als een redox onschuldig ligand, en oxidatie van de complexen leidt tot selectieve metaal gecentreerde radicaalreactiviteit in geval van iridium (Figuur 2). Dit is in contrast met eerdere onderzoeken van Wayland, waarin duidelijk werd aangetoond dat carbonyl adducten van RhII(por) zich gedragen als koolstof-gecentreerde radicalen. Verder werden nieuwe reactiepaden onthuld voor de 2-elektronen activering van carbonyl liganden van ‘closed-shell’ [MI(CO)n(Me3tpa)] systemen. Reactie van het iridium bis-

M Mn n+1

M Mn n+1C C

M Mn n+1

O

O

R1

R2

R1

R2 Figuur 1. Redox niet-onschuldig gedrag van alkeen, CO and carbeen liganden.

N

NN

NIr

CO

CO

N

N NN Ir

CO

H

+

I

+

III

2+

Ag+

Acetone

H2O (traces)

Figuur 2. Metaal gecentreerde reactiviteit van [Ir(CO)2(Me3tpa)] veroorzaakt door één-elektron oxidatie.

Samenvatting

150

carbonyl complex met water (of methanol) resulteert in de vorming van hydroxycarbonyl-hydride (of methoxycarbonyl-hydride) complexen. Deze complexen maken het mogelijk om iridium analogen van alle opeenvolgende relevante tussenproducten in de water-gas-shift-reactie te isoleren (Figuur 3). De reactie van het ‘closed-shell’ rhodium carbonyl complex met zuurstof in de aanwezigheid van water leidt tot de selectieve vorming van een carbonaat complex.

In Hoofdstuk 3 onderzochten we het effect van de donorsterkte van tri- en

tetradentaat bispicolylamine type liganden op de structuur van rhodium carbonylen. Voor het sterker elektron donerende bpa ligand is de vorming van binucleaire gebrugde triscarbonyl species gunstiger ten opzichte van eindstandige carbonylen die worden gevormd met zwakker donerende alkyl-bpa liganden. Hierbij is het interessant om op te merken dat eindstandige rhodium carbonyl complexen met het tetradentaat trispicolyl-amine ligand in een dynamisch evenwicht voorkomen met de dinucleaire bis-carbonyl gebrugde species (Figuur 4).

In Hoofdstuk 4 is de mogelijke rol van aminyl radicalen in de reactiviteit van ‘open-

shell’ rhodium en iridium complexen onderzocht. In de aanwezigheid van een base werd een onverwachte één-electron reductie van [MII(dbcot)(bla)]2+ complexen (dbcot = dibenzocyclooctadieen, bla = bislutidylamine) geobserveerd (Figure 5). Controle experimenten met [RhII(dbcot)(Bz-bla)]2+ (Bn-bla = N-benzyl-bislutidylamine) laten zien dat deze reacties ook kunnen verlopen zonder de betrokkenheid van aminyl radicalen, maar snelle waterstofatoom abstractie van het oplosmiddel door een aminyl radicaal ligand kan niet volledig worden uitgesloten. In ieder geval lieten deze complexen de detectie van dergelijke aminyl radicaal liganden door deprotonering niet toe, wat een studie van hun reactiviteit compliceerde. Dit gedrag is in sterk contrast met de gerapporteerde

N

N

N

NIr

H

O OH

CO

N

N

N

NIr

CO

CO

N

N

N

NIr

HH

CO

+

III

+

I

H2O/Acetone Acetone

+

III- CO2

Figuur 3. Geïsoleerde iridium analogen van relevante tussenproducten van de water-gas-shift-reactie.

N

N N

NRh

CO

N

NN

NRh

N

N N

NRh

O

O

+

I

2+

I

I2

Figuur 4. Dynamisch evenwicht tussen het monoucleaire [Rh(κ3-tpa)(CO)]+ en het dinucleaire [Rh(κ4-tpa)(μ-CO)]2

2+.

Figuur 5. Electrochemisch gedrag van rhodium en iridium dbcot complexen met bislutidine type liganden.

Samenvatting

151

eenvoudige vorming van Rh-gestabiliseerde aminyl radicaal liganden aan structureel gelijksoortige complexen, zoals gerapporteerd door Grützmacher en collega's. Verder worden in dit hoofdstuk mogelijke reductie reactiepaden besproken waarbij oxidatie van het oplosmiddel of water met behulp van een base betrokken is.

Hoofdstuk 5 beschrijft de reactiviteit van het paramagnetische complex

[IrII(ethene)(Me3tpa)]2+ met diazo verbindingen (gebruikt als carbeen uitgangstoffen). Het carbeen ligand, dat wordt gegenereerd op het ‘open-shell’ IrII metaal, gedraagt zich hier duidelijk als een redox actief ligand. Voor zo ver bij ons bekend werd dit ‘redox niet-onschuldige’ gedrag van (Fischer-type) carbenen aan group 9 overgangsmetalen voor het eerst aangetoond in dit onderzoek. Radicaalreacties die plaatsvinden op het ‘carbeen radicaal’ ligand en leiden tot vorming van C–C en C–H bindingen, alsmede DFT berekeningen aan het mechanisme van deze reacties, worden in dit Hoofdstuk gepresenteerd. Gebaseerd op deze resultaten geven we een beschrijving van de elektronische structuur van één-elektron geactiveerde carbenen. De ‘carbeen radicalen’ kunnen het best beschreven worden als koolstof-gecentreerde radicalen gebonden aan een ‘closed-shell’ overgangsmetaal (Figuur 6).

In Hoofdstuk 6 laten we zien dat 'carbeen radicalen' niet slechts een chemische

eigenaardigheid zijn, maar een belangrijke, voorheen onbekende rol spelen in de bestaande cyclopropaneringsreacties gekatalyseerd door kobalt-porfyrine complexen. Paramagnetische CoII(por) complexen zijn effectieve katalysatoren voor cyclopropanering van olefines. De meest succesvolle kobalt-gebaseerde

cyclopropanerings-katalysator is het chirale kobalt-porfyrine CoII(3,5-

DitBu-ChenPhyrin) complex ontwikkeld door Zhang en collega’s (Figuur 7). Deze katalysator is ongekend in zijn reactiviteit, stereocontrole, en zijn vermogen om cyclopropanering te laten verlopen met (bijna)

N

N N

NIr

NCMe

H R

N

N N

NIr

NCMe

H R

2+

III

2+

II

Figuur 6. Redox niet-onschuldig gedrag van een carbeen gebonden aan [Ir(Me3tpa)(NCMe)]2+. Resonantie structuren (boven) en de DFT berekende spin dichtheid (rechts).

N

N N

NCo

HN

O

O

O

H

N

N N

N

Co

HNO

HNO

NHO

NHO

H

H H

H

(S)(S)

(S)(S)


Figuur 7. CoII(3,5-DitBu-Chen-Phyrin) (links) en het doneren van een H-binding aan het ‘carbeen radicaal’ ligand in het belangrijkste intermediair van de katalystische reactie (boven).

Samenvatting

152

stoichiometrische hoeveelheden alkeen om carbeen dimerisatie te vermijden. Een ander kenmerk van deze intrigerende CoII porfyrine systemen is hun effectiviteit in cyclopropanering van elektron-deficiënte olefines zoals methyl acrylaat of acrylonitril. Deze reactiviteit is opmerkelijk en heel anders dan de typische elektrofiele Fischer-type carbeen intermediairen van de CuI en Doyle-type Rh2-gebaseerde systemen, en wijst dus op een heel ander karakter van het carbeen transfer intermediair voor CoII(por)-gebaseerde systemen.

In dit hoofdstuk onderzochten we het reactiemechanisme van deze katalysator met behulp van EPR spectroscopie en DFT berekeningen.

Het cyclopropaneringsmechanisme verloopt duidelijk via een radicaalmechanisme, waarbij reactieve koolstof-gecentreerde radicalen binnen de coördinatiesfeer van kobalt worden gevormd. We tonen aan dat het cruciale kobalt-carbeen intermediair een sterk koolstof radicaal karakter heeft, en eigenlijk een ‘carbeen radicaal’ is, soortgelijk aan de iridium-carbeen radicalen beschreven in Hoofdstuk 5 (Figuur 8). Het nucleofiele en radicaal-type karakter maakt de cyclopropanering van electron-arme olefines mogelijk, en onderdrukt carbeen dimerisatie activiteit. Voor zover bij ons bekend is dit het eerst gedocumenteerde voorbeeld waarin redox-niet-onschuldige liganden een hoofdrol spelen in een katalytische transformatie in de organometaalchemie. Additie van het ‘carbeen radicaal’ aan het olefine wordt gevolgd door een overgangstoestand met een lage barrière voor cyclopropaan ringsluiting. Dit laatste proces kan het beste beschreven worden als een radicaal-type C–C koppeling met gelijktijdige homolytische splitsing van de kobalt-koolstof binding. Dus, naast het ‘redox niet-onschuldige’ karakter van het carbeen, speelt ook de intrinsiek relatief zwakke Co–C binding een cruciale rol in deze reactie.

N

N N

NCo

N

N N

NCo H

H COOR

N2

H COOR

R'H

N2

H COOR

R' H

N

N N

NCo

COORNN

N

N N

NCo

H COOR

N

N N

NCo

H COOR

HR'

Verder is een interessante eigenschap van deze katalysator dat de aryl-dicyclopropaan-carboxamido functionaliteiten van het 3,5-DitBu-ChenPhyrin ligand (Figuur 7) potente H-binding donoren zijn. Dit verlaagt de activeringsbarrière voor de vorming van de carbeen radicalen vanuit het diazo ester substraat en leidt dus tot snellere reacties en hogere selectiviteit. Met het vaststellen van dit radicaal mechanisme, biedt de CoII(por)-gekatalyseerde olefine cyclopropanering een illustratief voorbeeld van een synthetische katalytische reactie die opereert via een radicaal-koppelingsproces, maar toch heel selectieve turnovers toestaat.

Figuur 8. Spindichtheid plot van het Co(por) ‘carbeen radicaal’ (boven) en de katalytische cyclus van de cyclopropaneringsreactie waar radicaal carbenen bij betrokken zijn.

Samenvatting

153

In Hoofdstuk 7 onderzochten we middels DFT

berekeningen de rol van kobalt metallo-radicalen in katalytische ketenoverdracht (CCT) in kobalt gemedieerde radicaalpolymerisatie reacties. De berekeningen concentreren zich op de omkeerbare waterstofatoom transfer reacties tussen kobalt porfyrines en organische radicalen (Figuur 9) en geven inzicht in het CCT mechanisme. Resultaten van deze studie suggereren verder dat de waterstofatoom overdrachtsroute van belang kan zijn in tal van reacties waarin een olefine wordt geïnserteerd in een zwakke M−H binding.

De belangrijkste bevinding van dit Proefschrift is dat het concept van redox niet-

onschuldige liganden niet alleen in stoichiometrische, maar ook in katalytische reacties kan worden benut. We toonden aan dat één-elektron activering van het gecoördineerde ligand resulteert in een dramatische verandering van de reactiviteit die reacties met anders niet reactieve substraten mogelijk zou kunnen maken. Dit Proefschrift laat verder zien dat een beter inzicht in het gedrag van onverzadigde liganden gebonden aan paramagnetische (Groep 9) metalen kan leiden tot nuttige, nieuwe (katalytische) transformaties. In het Perspectief speculeren we over de mogelijke betrokkenheid van radicaal carbeen complexen in nieuwe (katalytische) reacties, die potentieel kunnen leiden tot waardevolle nieuwe producten via, tot op heden, onbekende katalytische reactiepaden. Wij stellen voor dat de reacties van radicaal carbenen met geconjugeerde diënen of reacties van radicaal vinylcarbenen met olefinen zouden kunnen leiden tot gefunctionaliseerde cyclopentenen. Verder postuleren we dat nitreen liganden ‘redox actief’ zijn en dus zouden moeten reageren via radicaal-type mechanismen.

Figuur 9. Reversibele waterstofatoom transfer tussen CoII en organische radicalen.

Streszczenie

155

Streszczenie Uważa się, że rodniki są zbyt reaktywne aby być selektywne. Dla wolnych rodników stwierdzenie to może być prawdziwe, lecz w sferze koordynacyjnej metalu przejściowego, chemo-, regio-, diastereo- oraz enancjoselektywne reakcje rodnikowe są możliwe i istnieje wiele poglądowych przykładów enzymatycznych, biosyntetycznych transformacji rodnikowych. Głównym celem tej pracy jest wykazanie, że reakcje rodnikowe są syntetycznie użyteczne również w chemii metaloorganicznej i mogą nawet prowadzić do selektywnych reakcji katalitycznych. Badania, które doprowadziły do tych wniosków poskutkowały także odkryciem kilku interesujących reakcji zamnkniętopowłokowych (Rozdzialy 2 i 3).

Selektywne, rodnikowe reakcje transferu atomu otwartopowłokowych związków metaloorganicznych zwykle wymagają tego, aby niesparowana gęstość elektronowa zlokalizowana była na reaktywnym ligandzie. Tak zwana redox non-innocence ligandów jest więc ważną właściwością w tych reakcjach. Koncepcja ta jest przedstawiona w Rozdziale 1, który zawiera również przegląd literaturowy chemii otwartopowłokowych kompleksów metaloorganicznych metali grupy 9 (Co, Rh, Ir) z karbonylkami, karbenami i alkenami (i z podobnymi ligandami). Jednoelektronowa aktywacja ligandów alkenowych, karbonylkowych i karbenowych daje ogromną różnorodność selektywnych reakcji rodnikowych. Ligandy te mają nisko położone puste orbitale, które z łatwością przyjmują niesparowany elektron z metalu w kompleksach otwartopowłokowych. Powoduje to, że reakcja rodnikowa zachodzi na atomach węgla (Rys. 1), co prowadzi m.in. do powstawania nowych wiązań C–H lub C–C poprzez transfer atomu wodoru, bądź sprzęganie redukcyjne, dając dostęp do produktów, które nie mogą być otrzymane poprzez reakcje zamnkniętopowłokowe.

Rozdział 2 opisuje syntezę i właściwości redoks kompleksów karbonylkowych rodu i

irydu [MI(CO)n(Me3tpa)]+ (M = Rh, Ir, n = 1, 2) stabilizowanych przez ligand Me3tpa. Dwukarbonylkowy kompleks irydu ulega nieodwracalnemu jednoelektronowemu utlenieniu, co prowadzi do powstania monokarbonylkowego kompleksu wodorkowego (Rys. 2). W badanych układach CO jest ligandem typu redox innocent, i utlenianie kompleksu irydu prowadzi do selektywnej reakcji rodnikowej na metalu (Rys. 2), w przeciwieństwie do wyników Waylanda, który opisał że addukty karbonylkowe RhII(por) zachowywały się jak rodniki centrowane na atomie węgla. Ponadto, wykazaliśmy dwuelektronową aktywację ligandu karbonylkowego dla zamkniętopowłokowych układów [MI(CO)n(Me3tpa)]. Reakcja dwukarbonylkowego kompleksu irydu z wodą (lub metanolem) powoduje powstawanie kompleksów

M Mn n+1

M Mn n+1C C

M Mn n+1

O

O

R1

R2

R1

R2 Rysunek 1. Właściwości ligandów redox non-innocent: alkeny, CO i karbeny

N

NN

NIr

CO

CO

N

N NN Ir

CO

H

+

I

+

III

2+

Ag+

Acetone

H2O (traces)

Rysunek 2. Reaktywność kompleksu [Ir(CO)2(Me3tpa)] na metalu, jako rezultat jednoelektronowego utlenienienia.

Streszczenie

156

hydroksykarbonylowo-wodorkowych (bądź metoksykarbonylowo-wodorkowych), co umożliwiło izolację irydowych analogów wszystkich kolejnych pośredników w reakcji water gas shift (Rys. 3). Reakcja zamkniętopowłokowego karbonylkowego kompleksu rodu z tlenem w obecności wody prowadzi do selektywnego powstania kompleksu węglanowego.

Rozdział 3 zawiera wyniki

badań nad wpływem siły donorowej trój- i cztero-kleszczowych pochodnych bispikolyloaminy na strukturę karbonylków rodu. Ligand bpa, który jest silniejszym donorem powoduje uprzywilejowane powstawanie mostkowych triskarbonylków nad terminalnym karbonylkiem, który tworzy się ze słabszymi donorami typu alkil-bpa. Co ciekawe, kompleks karbonylkowy rodu z czterokleszczowym ligandem trispikolyloaminą wykazuje dynamiczną równowagę pomiędzy formą terminalną mono- i formą mostkową biskarbonylkową (Rys. 4).

W Rozdziale 4 opisaliśmy badania nad hipotetyczną rolą rodników aminylowych w

reaktywności otwartopowłokowych kompleksów rodu i irydu. Nieoczekiwanie, w obecności zasady zaobserwowaliśmy jednoelektronową redukcję kompleksów typu [MII(dbcot)(bla)]2+ (dbcot = dibenzocyclooctadien, bla = bislutidyloamina) (Rys. 5). Eksperyment kontrolny z użyciem [RhII(dbcot)(Bz-bla)]2+ (Bn-bla = N-benzylo-bislutidyloamina) wykazał, że reakcje te zachodzą również bez udziału rodników aminylowych, aczkolwiek nie można wykluczyć szybkiej abstrakcji atomu wodoru z rozpuszczalnika przez rodnik aminylowy. Niestety, otrzymanie wykrywalnej ilości rodnikowych ligandów aminylowych poprzez deprotonację tych kompleksów było niemożliwe, komplikując badanie ich reaktywności. Kontrastuje to z łatwością tworzenia rodnikowych ligandów aminylowych, stabilizowanych poprzez kompleksację z rodem, opisaną przez Grützmachera i współpracowników. Rozważyliśmy również

N

N

N

NIr

H

O OH

CO

N

N

N

NIr

CO

CO

N

N

N

NIr

HH

CO

+

III

+

I

H2O/Acetone Acetone

+

III- CO2

Rysunek 3. Wyizolowane irydowe analogi kolejnych pośredników w reakcji water gas shift.

N

N N

NRh

CO

N

NN

NRh

N

N N

NRh

O

O

+

I

2+

I

I2

Rysunek 4. Dynamiczna równowaga pomiędzy monomerycznym [Rh(κ3-tpa)(CO)]+ i dimerycznym [Rh(κ4-tpa)(μ-CO)]2

2+.

Rysunek 5. Elektrochemia kompleksów rodu i irydu z ligandami dbcot i bislutidynami.

Streszczenie

157

możliwe mechanizmy redukcji z jednoczesnym utlenieniem rozpuszczalnika lub wody w środowisku zasadowym.

Rozdział 5 opisuje reaktywność paramagnetycznego kompleksu [IrII(eten)(Me3tpa)]2+

ze związkami diazowymi (będącymi prekursorami karbenów). W tym przypadku, karben wygenerowany na otwartopowłokowym IrII bez wątpienia zachowuje się jako ligand redoks-aktywny. Według nas jest to pierwsze wykazanie redox non-innocence karbenów (typu Fischera) metalu przejściowego grupy 9. W rozdziale tym przedstawiamy rodnikowe reakcje tworzenia wiązań C–C i C–H występujące na ‘rodniku karbenowym’ oraz badania ich mechanizmu za pomocą DFT. Na podstawie tych wyników podajemy opis struktury elektronowej karbenów aktywowanych jednoelektronowo. „Rodniki karbenowe” można opisać jako rodniki zlokalizowane na atomie węgla, związane z metalem przejściowym o strukturze zamkniętopowłokowej (Rys. 6).

W Rozdziale 6 pokazujemy, że „rodniki karbenowe” są nie tylko chemiczną

ciekawostką, ale odgrywają ważną, dotychczas nieznaną rolę w istniejących katalitycznych reakcjach cyklopropanacji, możliwej dzięki porfirynowym kompleksom kobaltu. Paramagnetyczne kompleksy CoII(por) są efektywnymi katalizatorami cyklopropanacji olefin, a najlepszym z nich jest chiralny kompleks kobaltu i porfiryny CoII(3,5-DitBu-ChenPhyrin) opisany przez Zhanga i współpracowników (Rys. 7). Jego reaktywność, stereokontrola, i możliwość katalizowania cyklopropanacji z użyciem (prawie) stechiometricznych ilości alkenów i przy braku dimeryzacji karbenów jest bezprecedensowa. Kolejną intrygująca cechą systemu CoII-porfiryna jest efektywność w

cyklopropanacji ubogich elektronowo olefin, takich jak akrylan metylu lub akrylonitryl. Reaktywność ta jest godna uwagi i znacząco różna od reaktywności typowego elektrofilowego pośrednika – karbenu typu Fischera – który tworzy się w układach z CuI oraz w opartych na Rh2 układach

N

N N

NIr

NCMe

H R

N

N N

NIr

NCMe

H R

2+

III

2+

II

Figure 6. Redox non-innocent właściwości karbenu koordynującego do [Ir(Me3tpa)(NCMe)]2+. Struktura rezonansowa (u góry) i gęstość spinowa obliczona metodą DFT (z prawej).

N

N N

NCo

HN

O

O

O

H

N

N N

N

Co

HNO

HNO

NHO

NHO

H

H H

H

(S)(S)

(S)(S)


Rysunek 7. CoII(3,5-DitBu-Chen-Phyrin) (z lewej) i wiązanie wodorowe do ‘rodnika karbenowego’ w kluczowym pośredniku katalitycznym (u góry).

Streszczenie

158

typu Doyle’a, co sugeruje znacząco inny charakter pośrednika transferującego karben w układzie opartym na kompleksach CoII(por).

Badaliśmy również mechanizm reakcji tego katalizatora przy użyciu spektroskopii EPR oraz obliczeń DFT.

Mechanizm cyklopropanacji jest ewidentnie mechanizmem rodnikowym, który wymaga utworzenia reaktywnego rodnika węglowego w sferze koordynacyjnej kobaltu. Pokazujemy, że kluczowy pośrednik kobalt-karben ma znaczący charakter rodnika węglowego, i de facto jest „rodnikiem karbenowym”, podobnym do irydowych rodników karbenowych opisanych w Rozdziale 5 (Rys. 8). Jego nukleofilowy i rodnikowy charakter ułatwia cyklopropanację ubogich elektronowo olefin, i ogranicza jego aktywność w dimeryzacji karbenów. Zgodnie z naszym stanem wiedzy jest to pierwszy udokumentowany przykład, w którym ligand redox-non innocent odgrywa kluczową rolę w metaloorganicznej, katalitycznej transformacji. Po addycji „rodnika karbenowego” do olefiny następuje niskobarierowy stan przejściowy zamknięcia pierścienia cyklopropanowego. Ostatni proces można opisać jako rodnikowe tworzenie wiązania C–C z jednoczesnym homolitycznym przerwaniem wiązania kobalt-węgiel. Stąd, obok zachowania redox non-innocent karbenu, również wewnętrzna słabość wiązania Co–C odgrywa zasadniczą rolę w tej reakcji.

N

N N

NCo

N

N N

NCo H

H COOR

N2

H COOR

R'H

N2

H COOR

R' H

N

N N

NCo

COORNN

N

N N

NCo

H COOR

N

N N

NCo

H COOR

HR'

Kolejną interesującą właściwością tego katalizatora jest to, że arylo-dicyklopropano-karboksamidowe grupy ligandu 3,5-DitBu-ChenPhyrin (Rys. 7) są silnymi donorami wiązań wodorowych. Obniża to barierę aktywacji tworzenia się „rodników karbenowych” z diazoestrów, prowadząc do szybszej reakcji oraz większej selektywności. Ustalenie, że mechanizm ten ma charakter rodnikowy pozwala stwierdzić, iż cyklopropanacja katalizowana przez CoII(por) jest obrazowym przykładem syntetycznej reakcji katalitycznej, operującej poprzez procesy rodnikowe, jednocześnie umożliwiającej wysoką selektywność cyklu.

W Rozdziale 7 umieściliśmy wyniki obliczeń DFT opisujących rolę metalorodników

kobaltu w katalitycznym transferze łańcucha (catalytic chain transfer, CCT) w reakcji polimeryzacji rodnikowej mediowanej przez kobalt. Obliczenia koncentrowały się na odwracalnym transferze atomu wodoru między porfiryną kobaltu a rodnikami organicznymi (Rys. 9). Obliczenia te dają wgląd w mechanizm CCT. Rezultaty tych

Figure 8. Wykres gęstości spinowej Co(por) ‘karbenu rodnikowego’ (u góry) i cykl katalityczny reakcji cyklopropanacji przy udziale rodnikowych karbenów (z prawej).

Streszczenie

159

badań sugerują ponadto, że transfer atomu wodoru może być istotny w wielu reakcjach, w których następuje insercja olefiny w słabe wiązanie M−H.

Głównym odkryciem przedstawionym w tej pracy jest to, że idea ligandów redox non-innocent ma zastosowanie nie tylko w reakcjach stechiometricznych, ale i w katalitycznych. Pokazaliśmy, że jednoelektronowa aktywacja dokoordynowanego ligandu powoduje dramatyczne zmiany w jego reaktywności, co może doprowadzić do reakcji z substratami, które w innym wypadku byłyby niereaktywne. Uważamy, że lepsze zrozumienie zachowania nienasyconych ligandów koordynujących do paramagnetycznych metali grupy 9 może prowadzić do nowych, użytecznych (katalitycznych) transformacji. Dlatego w Perspektywach spekulujemy nad możliwym zastosowaniu kompleksów rodnikowych karbenów w nowych (katalitycznych) reakcjach, które potencjalnie mogą prowadzić do wartościowych nowych produktów poprzez bezprecedensowe mechanizmy katalityczne. Przewidujemy, że reakcje karbenów rodnikowych ze sprzężonymi dienami, bądź reakcje winylokarbenów rodnikowych z olefinami mogą prowadzić do podstawionych cyklopentenów. Przypuszczamy również, że nitreny powinny być ligandami redoks-aktywnymi i w konsekwencji reagować według mechanizmów rodnikowych.

Rysunek 9. Odwracalny transfer atomu wodoru międzyn CoII i rodnikami organicznymi.

161

List of Publications 1. Ligands that Store & Release Electrons during Catalysis: Dzik, W. I.; van der

Vlugt, J. I.; Reek, J. N. H.; de Bruin, B. Angew. Chem. Int. Ed. in press. 2. Remote Supramolecular Control of Catalyst Selectivity in the Hydroformylation of

Alkenes: Dydio, P.; Dzik, W. I.; Lutz, M.; de Bruin, B.; Reek, J. N. H. Angew. Chem. Int. Ed. in press.

3. ‘Carbene Radicals’ in CoII(por)-Catalyzed Olefin Cyclopropanation: Dzik, W. I.;

Xu, X., Zhang, X. P., Reek, J. N. H., de Bruin, B. J. Am. Chem. Soc. 2010, 132, 10891-10902.

4. Carbonyl Complexes of Rhodium with N-Donor Ligands: Factors Determining the

Formation of Terminal versus Bridging Carbonyls: Dzik, W. I.; Creusen, C.; de Gelder, R.; Peters. T. P. J.; Smits, J. M. M., de Bruin, B. Organometallics 2010, 29 1629-1641.

5. Hydrogen-Atom Transfer in Reactions of Organic Radicals with [Co-II(por)]• (por

= Porphyrinato) and in Subsequent Addition of [Co(H)(por)] to Olefins: de Bruin, B.; Dzik, W. I.; Li, S.; Wayland, B. B. Chem. Eur. J. 2009, 17, 4312-4320.

6. Activation of Carbon Monoxide by (Me3tpa)Rh and (Me3tpa)Ir: Dzik, W. I.; Smits,

J. M. M; Reek, J. N. H.; de Bruin, B. Organometallics 2009, 28, 1631-1643. 7. Selective C-C coupling of Ir-ethene and Ir-carbenoid radicals: Dzik, W. I.; Reek, J.

N. H.; de Bruin, B. Chem. Eur. J. 2008, 14, 7594-7599. 8. Carbon-carbon bond activation of 2,2,6,6-tetramethyl-piperidine-1-oxyl by a Rh-II

metalloradical: A combined experimental and theoretical study: Chan, K. S.; Li, X. Z.; Dzik, W. I.; de Bruin, B. J. Am. Chem. Soc. 2008, 130, 2051-2061.

9. Rhodium-mediated stereoselective polymerization of "carbenes", Hetterscheid, D.

G. H.; Hendriksen, C.; Dzik, W. I.; Smits, J. M. M, van Eck, E. R. H.; Rowan, A. E.; Busico, V.; Vacatello, M.; Van Axel Castelli, V.; Segre, A.; Jellema, E.; Bloemberg, T. G.; de Bruin, B. J. Am. Chem. Soc. 2006, 128, 9746-9752.

Acknowledgements

163

Acknowledgements When I look back at the time when I just started my Amsterdam adventure four years ago, I see how much I have changed through these days and recall all the great moments that I have had the luck to experience. This would not be possible without the many great people I have met here.

It all started with an e-mail from Bas, who invited me here and ‘gave me the opportunity to work in the group’. But saying this would not be enough to describe how much I am indebted to you, Bas. You taught me how to do great science, had great enthusiasm about chemistry and the doors to your office were always open for me to come to discuss all my real and imaginary issues.

Joost, thank you for giving the last touch to my papers and my thesis, giving constructive criticism to my research, but most importantly, for assembling a group of great personalities – a group that one wants to be at.

Jarl Ivar, I have to admit that recently I noticed that you are actually a very nice bloke. Thanks for your valuable comments (also regarding the manuscript) and for being in my opposition. Franti(šku) it was a pleasure to work with you on various practica, and to chat during the borrels. Piet, thank you for the input during my (early) minimeetings. Bert, I enjoyed our talks during coffee breaks. I am still astonished with the idea of a broodje koek.

Vladice, it was a great pleasure to meet a true humanist in a catalytic group. We had many nice discussions about languages, history, science and religion. Thanks to you I saw many things from a different perspective and your friendship was a true enrichment of my life (except of the butthole surfers concert). God bless you in all your future enterprises.

Pawle (a raczej: oj Krzysiu, Krzysiu) dzięki za to, że zgodziłeś się zostać moim paranimfem i musisz teraz roznosić te książeczki. Wspaniale było Cię poznać i zobaczyć, że nie tylko jesteś koneserem dobrego polskiego kina, ale i świetnym kumplem i chemikiem (kolejność dowolna). Dzięki, za wspólną pracę, mam nadzieję, że niejeden papier jeszcze razem opublikujemy (se wezmę, spiszę).

Jeroen, the B9.38 outing was a great stunt, thanks for being my chemistry consultant and for sharing the office and the lab with you. It was a great pleasure, especially that your music preferences were so close to mine.

The other (great) members of the group deserve my special thanks too. Dennis (you gave the foundations of my research, I am extremely happy that you are in my committee, take good care of the Tuinkabouter), Guillaume (supergezellig climbing sessions, with discussions about chemistry and the world around it. I enjoyed it very much), Fred (the champion, although being your paranymph was a scary time for me), Pierre-Alain (bbqs at Fokke Simonzs were so nice), Franco (thanks for the movie nights and discussions after football), Fabrizio (always walking… you were the first person who made me laugh with a joke in a language I don’t understand and that was priceless. The Washington/NY trip was also a great adventure), Ivo (passenger Jacobs, I hope nobody will ever maltreat any cup of yours again), Tendai (I guess the only person in the group except of me who understands the concept of pajamas. Although you have authoritarian tendencies I accepted your political views after you had bribed me with ten billion dollars), Leszek (cóż za niespodziewane spotkanie po latach), Tehila (you were always so friendly for everybody), Johanneke (You were so kind with your eagerness to help me or anybody else in the group. Our spelletjesavonden will be also something I

Acknowledgements

164

remember), Ruta, Fatna, Erik (cheers for all the help in the lab), Fred II (you actually make good cakes), Fabrizio R., Piluka (amazing Dr. del Rio, if we had shared the office longer I would not vouch for myself), Rick (great cycling skills, thanks for the daily batch of cartoons), Michel (1st class football player, do you still have the glass ball from NCCC?), Sumesh (I regret I did not get to know you better), Irshad, Jana, Alma (my sub-group co-nerd), Nicole (the choice of the labouting location was excellent), Markus (Herr Doktor), Miguel (thanks for teaching me how to ‘motivate’ Spanish students), Annemarie (I adored your singing in the lab and your skills as a guide through Almere), Massimo, Sandra, Sara M., Sara A., Carien, Alistair, Taasje (remember the waterballetje and the CO alarm?), Elsbeth, Mark Kuil, Mark B., Rosalba (I particularily enjoyed our Franconian experience), Samir, Yasemin, Axel, Jitte, Rene, Quinten (the non-innocence comes again. Thanks for all the fun at the coffee corner, your precision in jokes and throwing garbage into a bin is legendary), Jurjen (you were such a good mate, I had great fun talking to you about everything – pity we agreed on most issues – and playing in one team with you), Sander (hallo meneertje! Thanks for the anecdotes and for being our group’s artist), Zof, Yann, Iris, Jelmer, Dinesh, Chretien, Annelie (I liked you a lot, but whenever I heard your music I had to turn it off or leave the lab), Evelien, Bart, Remko, Lidy, Zohar, Mandy-Nicole, Guido (dat meen je niet), Rutger, Gennady, Sander O., Gledison, Stefan M., Florian, Ruben, Eoin (you had a great sense of humor, and though you did not have a mullet or even a hat, you were still an excellent climber), Alessandra (I think your plastic animals ate some of the compounds that were in the glovebox).

Thanks to Jan Meine and Jan G. (NMR guys: It was always a pleasure to seek for your help) and Han (the massaman). Paul, thanks for all the fun during every EPR measurement. Jan & Maxime thanks for the beautiful structures you measured for me.

Many thanks to my students. It was very rewarding to guide you and to see how fast your pick up (both chemistry and biking skills) and a real pleasure to get to know you. Luis, you were the first one. You did a great job in the lab. The research you started was the base for Chapter 4 and without you my thesis would be incomplete. Enjoy life, and success with your ‘concrete’ job. Sara, you were a very hard working person and I appreciated it a lot. Although your work did not end up in this book, it made into a very nice manuscript, and I’m sure we’ll see it in print soon. Have fun, and I hope you will be a great teacher. Pascal, Andres and Bart, it was nice to guide you through your projects, good luck with finishing your studies.

Special thanks to the chemists from other groups. Deniz (you and your cat were very good neighbors, extra thanks for your help with housing and for you smile), Szymon (grill i piłka), Vân Anh (our train trip to Leiden, helping you with the glovebox was a real pleasure to me), Sergio (sharing the CV and IR), Matthijs, Alexandre (I still have your bike), Stefan & Peter (keeping my fingers crossed for your defenses) and Dorette.

Thanks to all the members the football team not mentioned above, especially to: Akshay, Alessia A., Alessia G., Alex, Andreas, Arnaud, Filippo, Gabriel, Gordon, Greg, Jaap, José, Kush, Mauro, Tibert, Thomas, Victor and Yves.

Thanks to Maureen and Petra for their administrational help and the library people for their excellent service. Hans, thanks for housing and dinners, although I still have some issues with the tap.

I thank Peter and Jolanda for nice evenings in Utrecht and the trips outside, including the one into the Friese blabber. Ouders van Erica, dank jullie voor de gezelligheid in ‘de stad in het Hoge Noorden’.

Acknowledgements

165

Many thanks to the people back in Poland. My homies: Szymen, Rufus and Dobo (dzięki chłopaki za to, że zawsze gdy tylko byłem w pobliżu, mogłem się z wami spotkać. To niesamowite, że mimo tego jak różne są nasze drogi, wciąż mamy sobie tyle do powiedzenia). My chemistry mentors: Bogdan Janas (dzięki Sorze za rozwinięcie moich zainteresowań) and Wojciech Grochala (za pomoc w pierwszych krokach w branży i za wesołe maile kontrolne).

Xenia, Edward and Zofia thank you for the support that me and my family have had from you since I was born. It was very moving to meet you again in Columbus.

Without the constant support of my family I would have never gone that far (both scientifically and geographically). Kochani Rodzice, dawaliście mi wsparcie we wszystkich moich decyzjach i mimo, że było wam z tym ciężko, zachęcaliście mnie do wyjazdu. Jestem wam za to bardzo wdzięczny (jak i za cały wasz process wychowawczy, który zaowocował tą oto książką). Kasiu, dziękuję za to, że zawsze dawałaś mi wsparcie i za to, że odwiedzaliście mnie z Tomciem (dzięki Tomek za wprowadzenie ekstra luzu do rodziny).

I thank Erica – a great girl that had the biggest influence on me during these years. Eer, you were the first person in the group that I saw, and instantly become the favorite one (even though once I though you stole my research). You were the most important discovery I made and it was a true blessing that we came across each other here. You are my best friend and without a little bit of you by my side my life would have much less colors. You were with me with all the happy moments here, and gave me support when things went less well. For that and all the love you give me, I thank you A-LOT.

uva-dare (digital academic repository) group 9 open-shell ...can play a dominant role in directing...

Documents