[1st Edition] 9780128050767, 9780128047101

Table of contents :
Content:
CopyrightPage iv
ContributorsPage vii
PrefacePage ixPedro J. Pérez
Chapter One - Lithium, Sodium, and Potassium Magnesiate Chemistry: A Structural OverviewPages 1-46A.J. Martínez-Martínez, C.T. O’Hara
Chapter Two - Dearomatization of Transition Metal-Coordinated N-Heterocyclic Ligands and Related ChemistryPages 47-114R. Arévalo, M. Espinal-Viguri, M.A. Huertos, J. Pérez, L. Riera
Chapter Three - Insight into Metal-Catalyzed Water Oxidation from a DFT PerspectivePages 115-173D. Balcells
Chapter Four - Golden Jubilee for Scorpionates: Recent Advances in Organometallic Chemistry and Their Role in CatalysisPages 175-260C. Pettinari, R. Pettinari, F. Marchetti
Chapter Five - Recent Advances in Transition Metal-Catalyzed Dinitrogen ActivationPages 261-377M.D. Walter
IndexPages 379-386

Citation preview

Academic Press is an imprint of Elsevier 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, USA 525 B Street, Suite 1800, San Diego, CA 92101-4495, USA The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK 125 London Wall, London, EC2Y 5AS, UK First edition 2016 Copyright © 2016 Elsevier Inc. All rights reserved No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. ISBN: 978-0-12-804710-1 ISSN: 0065-3055 For information on all Academic Press publications visit our website at https://www.elsevier.com

Publisher: Zoe Kruze Acquisition Editor: Poppy Garraway Editorial Project Manager: Shellie Bryant Production Project Manager: Surya Narayanan Jayachandran Designer: Alan Studholme Typeset by SPi Global, India

CONTRIBUTORS R. Are´valo Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo, Spain D. Balcells University of Oslo, Oslo, Norway M. Espinal-Viguri Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo, Spain M.A. Huertos Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo, Spain F. Marchetti School of Science and Technology, University of Camerino, Camerino, Italy A.J. Martı´nez-Martı´nez University of Strathclyde, Glasgow, Scotland, United Kingdom C.T. O’Hara University of Strathclyde, Glasgow, Scotland, United Kingdom J. Pe´rez Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo; Centro de Investigacio´n en Nanomateriales y Nanotecnologı´a (CINN), CSIC-Universidad de Oviedo-Principado de Asturias, El Entrego, Spain C. Pettinari School of Pharmacy, University of Camerino, Camerino, Italy R. Pettinari School of Pharmacy, University of Camerino, Camerino, Italy L. Riera Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo; Centro de Investigacio´n en Nanomateriales y Nanotecnologı´a (CINN), CSIC-Universidad de Oviedo-Principado de Asturias, El Entrego, Spain M.D. Walter Institut fu¨r Anorganische und Analytische Chemie, Technische Universita¨t Braunschweig, Braunschweig, Germany

vii

PREFACE This is the first volume of the Advances in Organometallic Chemistry series for 2016, in which we have compiled five excellent reviews on several aspects of this discipline. Main group chemistry is represented in Chapter 1, where O’Hara and coworkers have provided an update on the structural diversity of alkali metal magnesiates, a field of growing interest not only from the structural but also from its synthetic point of view. The variety of processes involving ligand dearomatization in the coordination sphere of mid-transition metals bearing NHC ligands has been reviewed by Pe´rez, Riera, and coworkers. The flavor of classical organometallic chemistry accompanies this contribution, where understanding reaction steps might be employed in the future in other synthetic or catalytic reactions. Balcells has provided an account of water oxidation with transition metal complexes, from a DFT perspective. The main catalytic systems, mechanistic proposals, and evidences are discussed in a must-read contribution for the groups involved in the field. The celebration of the Golden Jubilee of trispyrazolylborate ligands, first reported by Jerry Trofimenko in 1966, has reached our Serial thanks to Pettinari, which has taken over Trofimenko’s task to disseminate the importance of these ligands in organometallic chemistry. A review of their role in organometallic and homogeneous catalysis in the last few years is presented in Chapter 4. This volume is completed with Chapter 5, a review on the recent advances on dinitrogen activation by transition metal centers. After the rise and fall of dinitrogen chemistry several decades ago, in the last one it has emerged with new strategies and perspectives. A review on this emergence has been prepared by Walter, in an effort to contain all the relevant discoveries recently made in the field. I very much appreciate the authors for accepting the invitation to participate, particularly in these days in which invitations to many tasks and journals are so frequent. Last, but not least important, is my acknowledgment to Shellie Bryant and Surya Narayanan that have made an outstanding work from the Editorial. PEDRO J. PE´REZ

ix

CHAPTER ONE

Lithium, Sodium, and Potassium Magnesiate Chemistry: A Structural Overview A.J. Martínez-Martínez, C.T. O’Hara1 University of Strathclyde, Glasgow, Scotland, United Kingdom 1 Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Lithium Magnesiate Complexes 2.1 Alkyl/Aryl Lithium Magnesiate Complexes 2.2 Amido Lithium Magnesiate Complexes 2.3 Heteroleptic Lithium Magnesiate Complexes 3. Sodium Magnesiate Complexes 3.1 Donor-Free Homo- and Heteroleptic Sodium Magnesiate Complexes 3.2 Introducing Donors to Sodium Magnesiate Complexes 3.3 Inverse Crown Molecules 3.4 Miscellaneous Sodium Magnesiate Complexes 4. Potassium Magnesiate Complexes 4.1 Inverse Crown Molecules 4.2 Introducing Donors to Potassium Magnesiate Complexes 4.3 Miscellaneous Potassium Magnesiate Complexes 5. Summary References

1 2 2 6 9 14 14 18 28 31 32 32 34 40 42 42

1. INTRODUCTION The deprotonative metalation (deprotonation) of an aromatic ring (ie, the replacement of a hydrogen atom with a metal one) has been known since 1908 when Schorigin reported that a C–H bond of benzene could be cleaved by a mixture of sodium metal and diethylmercury, to yield phenylsodium.1,2 Monometallic compounds, particularly organolithium reagents have historically been employed in deprotonation reactions.3,4 Advances in Organometallic Chemistry, Volume 65 ISSN 0065-3055 http://dx.doi.org/10.1016/bs.adomc.2016.02.001

#

2016 Elsevier Inc. All rights reserved.

1

2

A.J. Martínez-Martínez and C.T. O’Hara

In recent years, bimetallic variants (one metal being an alkali metal, the other magnesium, zinc, aluminum, etc.) have come to the fore as a class of compounds capable of smoothly performing deprotonation reactions.5–12 These reagents often offer enhanced functional group tolerance, greater stability in common laboratory solvents, and also reactions can be performed at ambient temperature (rather than at
78 °C). The bimetallic compounds are often referred to as “ate” complexes, a term coined by Wittig in 1951 when he studied bimetallic compounds such as the lithium magnesiate LiMgPh3, lithium zincate LiZnPh3, and “higher-order” lithium zincate Li3Zn2Ph7.13 There was a window of almost five decades before chemists significantly exploited “ate” chemistry. Since 2000, the number of structural and synthetic studies using bimetallic reagents has increased dramatically, and due to their wide scope, they continue to be a hot topic in modern chemistry. Several reviews have been published in this area.6–12 In this chapter, an overview of the recent structural chemistry of alkali metal magnesiates (from 2007 to 2015) is presented focusing specifically at the metal pairs utilized.

2. LITHIUM MAGNESIATE COMPLEXES In this section, the surprisingly diverse structural chemistry of recently published lithium magnesiate complexes, containing carbon- and/or nitrogen-based anions, will be surveyed. Since 2007, several different structural motifs have been reported. In this section, these will be summarized according to the ligand sets within the lithium magnesiate framework.

2.1 Alkyl/Aryl Lithium Magnesiate Complexes Lithium magnesiates comprised completely of carbanionic ligands were among the first ate complexes reported. They are generally prepared by combining the two monometallic organometallic species in a hydrocarbon medium that also contains a Lewis base donor. Since 2007, contacted ion pair “lower-order” lithium (tris)alkyl magnesiates (and dimers of this motif ) and “higher-order” dilithium (tetra)alkyl magnesiates, and solvent-separated examples have been reported. Examples of each of these structural types will be discussed here. The monomeric tris(carbanion) motif is the simplest structural form of a lithium all-carbanionic magnesiate. To isolate this particular form, the use of a multidentate Lewis basic donor compound is generally required. Hevia and coworkers have reported the PMDETA (N 0 ,N 0 ,N 00 ,N 000 , N 000 -pentamethyldiethylenetriamine) solvated monomeric lithium magnesiate

Lithium, Sodium, and Potassium Magnesiate Chemistry

3

Fig. 1 Molecular structure of [(PMDETA)LiMg(CH2SiMe3)3] 1.

Fig. 2 Molecular structure of [(THF)LiMg(CH2SiMe3)3]1 2.

[(PMDETA)LiMg(CH2SiMe3)3] 1 (Fig. 1).14 It has an open-motif, whereby a single CH2SiMe3 alkyl bridge connects the metals. This structure is intermediate between a solvent-separated ion pair and a molecule that consists of a closed four-membered Li–C–Mg–C ring (vide infra). When the denticity of the donor is lowered, it is possible to completely change the structure of the isolated lithium magnesiate. For instance by using THF, a polymeric chain variant [(THF)LiMg(CH2SiMe3)3]1 2 is isolated (Fig. 2).14 The monomeric unit of 2 consists of a closed Li–C–Mg–C ring, and polymer propagation occurs via an intermolecular interaction between the CH2SiMe3 group not present in this ring and a Li atom. Another interesting and unusual feature of 2 is that the molecule of THF that is present binds to the magnesium center.

4

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 3 Section of the polymeric chain of [(1,4-dioxane)2LiMg(CH2SiMe3)3]1 3.

Fig. 4 Section of the polymeric chain of [(1,4-dioxane)Li2Mg2(CH2SiMe3)6]1 4.

When 1,4-dioxane is used in place of THF, two different lower-order magnesiates can be formed depending on the quantity of the donor that is employed—higher quantities of donor lead to a polymeric complex which incorporates two molecules of 1,4-dioxane per monomeric unit, [(1,4dioxane)2LiMg(CH2SiMe3)3]1 3 (Fig. 3).14 In 3, one 1,4-dioxane molecule binds solely to the lithium atom in a monodentate fashion (the other O atom does not participate in bonding). The polymeric arrangement is formed by a combination of Li-(1,4-dioxane)-Li and Mg-(1,4-dioxane)-Mg bridges to give a “head-to-head” and “tail-to-tail” repeating pattern. When a molar deficit of 1,4-dioxane is employed, the polymeric “tetranuclear” lower-order magnesiate [(1,4-dioxane)Li2Mg2(CH2SiMe3)6]1 4 is isolated.14 Each tetranuclear building block in 4 consists of three fused four-membered metal-carbon rings: two are LiC2Mg rings while the other is a Mg2C2 ring. The junctions occur at the Mg atoms (Fig. 4).

Lithium, Sodium, and Potassium Magnesiate Chemistry

5

Fig. 5 Molecular structure of [Li(THF)4]+[Mg(mesityl)3]
5.

The examples discussed thus far are classed as contacted ion pairs, as both distinct metals are contained within the same molecule. Since 2007, one example of a solvent-separated lithium tris(aryl) magnesiate (ie, the complex exists as distinct cationic and anionic moieties) has been reported. [Li (THF)4]+[Mg(mesityl)3]
, 5 (where mesityl is 2,4,6-trimethylphenyl) resembles many other trialkyl/aryl lithium magnesiates and consists of a tetrahedrally disposed tetra-THF-solvated lithium cation and a trigonal planar magnesium tris(aryl) anion (Fig. 5).15 Another common motif in organomagnesiate chemistry occurs when the compound is rich in alkali metal with respect to magnesium. In general, two factors can lead to this scenario: (1) and most obviously, if the organolithium to organomagnesium reagent ratio employed in the synthesis is 2:1 or higher; (2) if the spatial nature of the lower-order reagent (including steric bulk of anions and donor ligand) precludes the inclusion of a further molecule of “Li–R” (R is alkyl/aryl). Since 2007, five complexes that can be classed as higher-order lithium magnesiates have been reported. The first three are structurally similar and are the (trimethylsilyl)methylcontaining [(TMEDA)2Li2Mg(CH2SiMe3)4] 6 (Fig. 6A);14 the 1,4buta-di-ide [(TMEDA)2Li2Mg[CH2 (CH2)2CH2]2] 7 (Fig. 6B);16 and the heteroanionic 1,4-buta-di-ide, diphenyl-containing [(TMEDA)2Li2Mg(Ph)2 [CH2(CH2)2CH2]] 8 (Fig. 6C).16 Complex 7 was prepared by treating 1,4-dilithiobutane with THF-solvated magnesium dichloride; whereas 8 was produced by combining 1,4-dilithiobutane with dioxane-solvated diphenylmagnesium.

6

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 6 Molecular structures of higher-order magnesiates (A) [(TMEDA)2Li2Mg (CH2SiMe3)4] 6; (B) 1,4-buta-di-ide [(TMEDA)2Li2Mg[CH2(CH2)2CH2]2] 7; and (C) [(TMEDA)2Li2Mg(Ph)2[CH2(CH2)2CH2]] 8.

The remaining two higher-order magnesiates have a subtly different structure and can be described as “magnesiacyclopentadienes”.17 By reacting substituted 1,4-dilithio-1,3-butadienes with 0.5 molar equivalents of MgCl2 in the presence of TMEDA, the spiro-dilithio magnesiacyclopentadiene complexes ðTMEDAÞ2 Li2 Mg½CR1 2 ðCR2 2 Þ2 CR1 2 2 9 and 10 (for 9, R1 ¼ SiMe3; R2 ¼ Me; and for 10, R1 ¼ SiMe3; R2 ¼ Ph) are formed (Fig. 7).

2.2 Amido Lithium Magnesiate Complexes In keeping with the chemistry discussed thus far, tris(amido) lithium magnesiate complexes can be grouped into lower-order (contacted or solvent-separated ion pairs) and higher-order species. Since 2007, it appears that only one tris(amido) lower-order lithium magnesiate has been synthesized namely the dimeric unsolvated lithium magnesium guanidinate [Li2Mg2(hpp)6] 11 (Fig. 8) (where hpp is 1,3,4,6,7,8-hexahydro-2Hpyrimido[1,2-a]pyrimidide).18 The guanidinates anions adopt two different coordination modes—one bridging between two metal centers and the other between four metal centers. Three solvent-separated tris(amido) lithium magnesiates have been reported since 2007. All three are tris(HMDS) (1,1,1,3,3,3-hexamethyldisiliazide)

Lithium, Sodium, and Potassium Magnesiate Chemistry

7

Fig. 7 Molecular structures of (A) [(TMEDA)2Li2Mg[C(SiMe3)2(CMe2)2C(SiMe3)2]2]9 and (B) [(TMEDA)2Li2Mg[C(SiMe3)2(CPh2)2C(SiMe3)2]2] 10.

Fig. 8 Molecular structure of Li2Mg2(hpp)6 11.

complexes [Li{(
)-sparteine}2]+[Mg(HMDS)3]
12 (Fig. 9A);19 [Li {(R,R)-TMCDA}2]+[Mg(HMDS)3]
13 (Fig. 9B);19 and [Li(IPr)2]+ [Mg(HMDS)3]
14 (Fig. 9C)20 [where (R,R)-TMCDA and IPr are (R,R)-tetramethylcyclohexyldiamine and 1,3-bis(2,6-diisopropylphenyl) imidazolyl-2-ylidene, respectively] and have essentially identical Mg(HMDS)3 anions. Only one example [Li2Mg{(NDipp)2SiMe2)2] 15 (Fig. 10) of a higherorder heteroleptic amido magnesiate has been published between 2007 and 2015.21 It incorporates the bulky dianionic bis(amido)silane ligand [Me2Si (DippN)2]2
(where Dipp is diisopropylphenyl). The Mg atom is tetrahedrally disposed—η2 (N,N)-bound to two bis(amido)silane ligands—and the lithium atoms σ-bind to a N atom of each ligand. Further stabilization to the lithium atoms is provided by π-coordination to an arene-C atom (Fig. 10).

8

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 9 Molecular structure of (A) [Li{(
)-sparteine}2]+[Mg(HMDS)3]
12; (B) the cation of [Li{(R,R)-TMCDA}2]+[Mg(HMDS)3]
13; and (C) the cation of [Li(IPr)2]+[Mg(HMDS)3]
14.

Fig. 10 Molecular structure of [Li2Mg{(NDipp)2SiMe2}2] 15.

Lithium, Sodium, and Potassium Magnesiate Chemistry

9

2.3 Heteroleptic Lithium Magnesiate Complexes So far, only all carbanion or all amido lithium magnesiates have been discussed. In this section of the review, heteroleptic lithium magnesiates will be described and will begin by focusing on mixed carbanion/amido lithium magnesiates. Then magnesiates, which contain carbanions (or amido ligands) with other ligands, will be discussed. The structural chemistry for this set of molecules is diverse. The simplest example is the unsolvated lower-order monomeric complex [LiMg(HMDS)2tBu] 16 (Fig. 11).22 The HMDS ligands bridge between the two metals in the structure while the tBu is terminally bound to the magnesium atom. Further stabilization of the lithium atom is achieved by two agostic-type interactions from a pair of CH3-groups present on the HMDS ligands. Redshaw and coworkers have recently reported the synthesis and structure of a monomeric bimetallic calixarene-containing complex [(THF) LiMgnBuR*] 17 (where R* is 1,3-dipropoxy-p-tert-butylcalix[4]arendiide).23 The Mg atom in 17 is five coordinate, adopting a distorted square pyramidal arrangement with the n-butyl ligand sitting apically with respect to the four equatorially positioned oxygen atoms. The lithium atom has a trigonal planar coordination sphere and bonds to two anionic O centers and a THF molecule, and it sits within the calixarene cone. Complex 17 has been successfully utilized in the ring-opening polymerization of raclactide (Fig. 12). Two halide containing amido lithium magnesiates [(THF)3LiMg(TMP) Cl2] 18 (Fig. 13)24 and the dimeric [(THF)2LiMg(NiPr2)Cl2]2 19 (Fig. 14)25 have recently been reported. These are of particular importance to the welldeveloped synthetic area of turbo-Hauser metalation chemistry pioneered

Fig. 11 Molecular structure of LiMg(HMDS)2tBu 16.

10

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 12 Molecular structure of [(THF)LiMgnBuR*] 17.

Fig. 13 Molecular structure of [(THF)3LiMg(TMP)Cl2] 18.

Fig. 14 Molecular structure of [(THF)2LiMg(NiPr2)Cl2]2 19.

Lithium, Sodium, and Potassium Magnesiate Chemistry

11

Fig. 15 Molecular structure of [(THF)3Li2Mg{(rac)-BIPHEN}(nBu)2] 20. Complex 21 is isostructural except that nBu groups are replaced by CH2SiMe3 groups.

by Knochel.26 Complex 18 is the most heavily utilized turbo-reagent and is dinuclear. The chloride anions bridge the two metals, the TMP anion adopts a terminal position on the Mg cation and three molecules of THF complete the structure—two binding to the Li cation and one to the Mg. In contrast, 19 is tetranuclear. A key structural difference which perhaps contributes to significant synthetic differences between the two compounds is that the diisopropylamide groups adopt bridging positions between two Mg cations at the center of the structure resulting in the formation of a dimer rather than a monomer, presumably due to the reduced steric influence of diisopropylamide vs TMP. Akin to their homoleptic analogs, higher-order heteroleptic species have also been isolated. A series of higher-order magnesiates which contain the dianionic (rac)-BIPHEN ligand have been reported. These include [(THF)3Li2Mg{(rac)-BIPHEN}(nBu)2] 20 (Fig. 15),27 [(THF)3Li2Mg{(rac)BIPHEN}(CH2SiMe3)2] 21,27 [(THF)2Li2Mg{(rac)-BIPHEN}(tBu)2] 22,27 and [(THF)2Li2Mg{(rac)-BIPHEN}(2-pyridyl)2] 23 (Fig. 16).27 In 20–23, the biphenolate ligand stitches together the three metals forming a Li–O– Mg–O–Li zig–zag chain. The main structural framework is completed by the alkyl or pyridyl ligands adopting bridging positions between the metals. Interestingly, 20 and 21 contain three THF molecules while 22 and 23 only contain two. The compounds were prepared by co-complexation of the dilithium biphenolate with the respective dialkyl (or dipyridyl)magnesium

12

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 16 Molecular structure of [(THF)2Li2Mg{(rac)-BIPHEN}(2-pyridyl)2] 23. Complex 22 has a similar motif except that the pyridyl groups are replaced by tBu groups.

Fig. 17 Molecular structure of [(THF)LiMg3Me3(OC6H11)4] 24.

reagent. Also 23 can be prepared by reacting 20 with 2-bromopyridine showing that 20 is active in magnesium–halogen exchange reactions. Two magnesium-rich species, which adopt cubane-type motifs, have recently been isolated. The first is the tetranuclear lithium-trimagnesium alkyl alkoxide [(THF)LiMg3Me3(OC6H11)4] 24 (Fig. 17).28 The metal cations and alkoxide anions occupy the corners of the cube, and the methyl ligands are terminally bound to the Mg centers. The coordination sphere

Lithium, Sodium, and Potassium Magnesiate Chemistry

13

Fig. 18 Molecular structure of [(nBu3N)LiMg3tBu3{S(tBu)}4] 25.

Fig. 19 Molecular structure [(THP)2Li2Mg3(TMP)2(Fc*)2] 26.

of

trimagnesium-bridged

ferrocenophane

of the Li cation is completed by a molecule of THF. Complex 24 has been employed as a molecular single-source precursor for the preparation of MgO nanoparticles which contains lithium. The second cubane thiol-containing [(nBu3N)LiMg3tBu3{S(tBu)}4] 25 (Fig. 18) adopts a similar structural motif to 24 and was isolated by Schn€ ockel and coworkers while attempting to 29 access Mg(I) complexes. Another lithium magnesiate arises from the double magnesiation of N-methyl-1,3-propylenediaminoboryl ferrocene (Fc*-H2).30 The trimagnesium-bridged ferrocenophane [(THP)2Li2Mg3(TMP)2(Fc*)2] 26 (Fig. 19, where THP is tetrahydropyran) has a motif which has been previously observed.31

14

A.J. Martínez-Martínez and C.T. O’Hara

3. SODIUM MAGNESIATE COMPLEXES 3.1 Donor-Free Homo- and Heteroleptic Sodium Magnesiate Complexes Tri-organo-sodium magnesiates can be prepared as solvates using common donor molecules (TMEDA, PMDETA, THF, etc.) or as solvent-free complexes. The presence of polar alkali metals in their formulations is often required to increase their solubility in hydrocarbon solvents, often at the cost of altering their aggregation states in solution. If the anions within the magnesiate are judiciously chosen, polymeric (or highly oligomeric) aggregation states in the solid state can be achieved. The polymeric sodium magnesiate [NaMg(CH2SiMe3)3]1 27 (Fig. 20A) is an example of a homoleptic tri-basic alkyl deprotonating agent.32 Related species have been used in deprotonation reactions, for instance, its nBu analog [NaMg(nBu)3]33 has been used as an effective deprotonating reagent of a sterically demanding ketone (2,4,6-trimethylacetophenone) for preparing mixed metal enolate complexes,34 and more recently to deprotonate benzophenone imine to give sodium magnesiate complexes containing ketimino anions.35 The homoleptic sodium magnesiate 27 (Fig. 20) represents the first example of a structurally characterized solvent-free tris-alkyl version reported in the literature. Complex 27 exists as a polymeric ate—prepared by a co-complexation approach by mixing the monometallic alkyls [NaCH2SiMe3] and [Mg(CH2SiMe3)] in an n-hexane/toluene solvent mixture.32 The organo alkali metal reagent [NaCH2SiMe3] interacts with the diorgano magnesium complex [Mg(CH2SiMe3)] to formally give a “[NaMg(CH2SiMe3)3]” complex (Fig. 20A). The trigonal planar Mg atom is bound to three alkyl ligands, one bridges to the Na cation in the asymmetric unit whereas the other two bridging alkyls are linked to neighboring Na atoms. The absence of Lewis donor molecules is crucial in inducing polymerization by forcing the alkali metal Na to directly coordinate to a neighboring alkyl group. This situation results in a 12-atom [NaCMgC]3 fused ring which propagates as a honeycomb layered twodimensional infinite network (Fig. 20B) in which all CH2SiMe3 ligands are rendered equivalent. The bis(amido) alkyl sodium magnesiate [NaMg(HMDS)2(nBu)]1 28 (Fig. 20C) is also polymeric;36 however, it adopts a one-dimensional chainlike infinite polymer through an almost linear Na–C(nBu)-Mg bridge. Two

Lithium, Sodium, and Potassium Magnesiate Chemistry

15

Fig. 20 (A) Molecular structure of polymeric [NaMg(CH2SiMe3)3]1 27 showing the contents of the asymmetric unit. (B) Section of the two-dimensional sheet network of 27. (C) Section of the extended polymeric framework of [NaMg(HMDS)2(nBu)]1 28.

bridging HMDS ligands complete the trigonal planar coordination sphere of both Mg and Na cations. Returning to 27, it has been utilized in the promotion of catalytic hydroamination/trimerization reactions of isocyanates.37 It also reacts with diphenylamine in a 1:3 molar ratio (albeit in the presence of THF) to yield [(THF)NaMg(NPh2)3(THF)] 29 (Fig. 21). Complex 29 is a contacted ionpair whereby the cationic [Na(THF)]+ fragment exhibits π-interactions with two arenes groups (in a η5 and η2 fashion) from two distinct diphenyl amido PhN groups. The Mg binds to three di-diphenyl-amido ligands and one molecule of THF to complete its coordination sphere. Complex 29 acts as a precatalyst to selectively promote the hydroamination/trimerization reactions of isocyanates in good yields under mild conditions.37 When it is reacted with three molar equivalents of tert-butyl isocyanate, the novel tris(ureido)sodium magnesiate [(THF)3NaMg(ureido)3] 30 is formed resulting from the insertion of an heterocumulene molecule into each of

16

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 21 (A) Molecular structure of [(THF)NaMg(NPh2)3(THF)] 29 and (B) [(THF)3 NaMg (ureido)3] 30.

the Mg–N bonds of 29. In 30, each ureido ligand is fac-disposed and chelates to the octahedral Mg center via its O and N atoms forming a four-membered [Mg–O–C–N] ring, while the terminal Na atom is bonded to the three O atoms of the ureido ligands and to three THF molecules in an octahedral fashion. Complex 27 is also an ideal bimetallic precursor for novel solvent-free sodium magnesiate complexes which contain both alkyl and alkoxide ligands. When 27 is exposed to atmospheric oxygen in a controlled manner, the alkoxide-containing complex [Na2Mg2(OCH2SiMe3)2(CH2SiMe3)4]1 31 is obtained (Fig. 22A).38 It features a dimeric arrangement comprising two “NaMgR2(OR)” units giving rise to a face-fused double heterocubane structure with two missing corners. Alternatively, the complex can be described as a sodium magnesium inverse crown (see Section 3.3 for definition) consisting of a cationic eight-membered polymetallic [NaCMgC]2 ring with four bridging CH2SiMe3 groups between Na and Mg atoms, and two alkoxide OCH2SiMe3 guests. Each alkoxide group is bonded to two Mg and one Na atom. In absence of Lewis donor molecules, discrete inverse crown units propagate in the two-dimensional space by long secondary Na⋯Me electrostatic contacts between the two Na atoms and CH2SiMe3 groups from neighboring inverse crown molecules (Fig. 22B). Interestingly, around the same time that the structure of [NaMg (HMDS)2(nBu)] 27 was reported, Hill and co-workers39 studied the

Lithium, Sodium, and Potassium Magnesiate Chemistry

17

Fig. 22 (A) Molecular structure of [Na2Mg2(OCH2SiMe3)2(CH2SiMe3)4]1 31 showing the contents of the asymmetric unit. (B) Section of the two-dimensional network of 31.

Fig. 23 Molecular structure of the higher metal hydride cluster [Na6Mg6{N (SiMe3)2}8H10] 32.

reactivity of an in situ mixture of [NaMg(HMDS)2(nBu)] with PhSiH3. The resulting novel higher metal hydride cluster is the heterododecametallic complex [Na6Mg6{N(SiMe3)2}8H10] 32 (Fig. 23). Two distorted octahedral [MgH6] units share two hydride ligands forming a single [Mg2H10] unit. The remaining four Mg centers are coordinated to two HMDS and two hydride ligands in a tetrahedral fashion and six Na atoms occupy the terminal sites. The formation of 32 involves the distinct metathesis of both nBu and amide ligands present in 27. This reactivity indicates the under-represented utility of heteroleptic magnesiates for selective metathesis chemistry.

18

A.J. Martínez-Martínez and C.T. O’Hara

3.2 Introducing Donors to Sodium Magnesiate Complexes The mixed sodium magnesium compounds [Na2(HMDS)2Mg(nBu)2 (donor)]1 (donor is TMEDA and (R,R)-TMCDA for 33 and 34, respectively, Fig. 24) are isostructural and can be considered as the first examples of “inverse sodium magnesium ate” complexes. They can be rationally prepared by combining HMDS(H) with a mixture of nBuNa and nBu2Mg in the presence of the corresponding donor molecule in a 2:2:1:1 molar ratio. Normally, ate complexes are associated with bimetallic systems, whereby one of the metals has higher Lewis acidity (ie, Mg2+) than the other (ie, Na+), thus the former metal captures more Lewis basic anionic ligands.

Fig. 24 Sections of the linear polymeric inverse magnesiates (A) [Na2(HMDS)2Mg (nBu)2(TMEDA)]1 33 and (B) [Na2(HMDS)2Mg(nBu)2(R,R-TMCDA)]1 34.

Lithium, Sodium, and Potassium Magnesiate Chemistry

19

Fig. 25 Molecular structure of (A) [(TMEDA)NaMg(TMP)2(CH2SiMe3)] 35 and (B) [{(
)sparteine}NaMg(TMP)(nBu)] 36.

For 33 and 34, this situation is reversed; these polymers can be better described as the nBu2Mg moiety formally acting as a Lewis base to solvate the dimeric [NaHMDS]2 unit [ie, the (NaHMDS)2 dimer acts as a Lewis acidic entity], hence the new term “inverse magnesiate”. Both complexes are still polymeric despite the presence of TMEDA and (R,R)-TMCDA donors. The two complexes [(TMEDA)NaMg(TMP)2(CH2SiMe3)] 35 and [{(
)-sparteine}NaMg(TMP)2(nBu)] 36 are isostructural (Fig. 25).40 Complex 36 is an example of a chiral mixed-metal, mixed alkyl-amide sodium magnesiate and represents the first structural example whereby (
)-sparteine (a highly important ligand in asymmetric synthesis) is chelated to an alkali metal other than lithium. Both are discrete monomers consisting of fourmembered Na–N–Mg–C rings with one bridging TMP and alkyl ligand between Na and Mg, and one terminal TMP and bidentate chelating ligand, coordinated to Mg and Na, respectively. They contain the basic skeleton evident for many bimetallic synergic bases and indeed can be prepared by the typical co-complexation protocol in hydrocarbon solvent. Complex 35 has been utilized in the metalation of furan, tetrahydrofuran, thiophene, and tetrahydrothiophene. Several interesting deprotonation and cleave/capture mechanistic insights have been uncovered using this base. For instance, 35 reacts in a different fashion with thiophene and tetrahydrothiophene giving rise to different structural motifs. Toward the former, 35 behaves as a tri-basic reagent yielding [(TMEDA)Na (α-C4H3S)3Mg(TMEDA)] 37 which contains three α-deprotonated thiophenyl moieties (Fig. 26A).41 It exhibits three α-deprotonated thiophenyl molecules that are bonded to Mg in a σ-fashion and Na is π-coordinated to the three thiophenyl moieties. TMEDA ligands are coordinated to both

20

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 26 Molecular structure of (A) [(TMEDA)Na(α-C4H3S)3Mg(TMEDA)] 37 and (B) [(TMEDA)NaMg(TMP)2(α-C4H7S)] 38.

Na and Mg atoms, an exceptionally rare structural feature in the chemistry of sodium magnesiates. When 35 reacts with an equimolar quantity of tetrahydrothiophene, the bis amido complex [(TMEDA)NaMg(TMP)2(α-C4H7S)] 38 is obtained (Fig. 26B). Complex 38 is structurally related to 35, where an alkyl group has been replaced by the deprotonated tetrahydrothiophenyl unit and it represents the first structural example of a magnesiated tetrahydrothiophenyl molecule. Interestingly, the Na atom is also interacting in a π-fashion with the softer S atom from the tetrahydrothiphenyl ligand providing additional stabilization for the α-deprotonated substrate. When 35 reacts with furan, it mirrors the reactivity observed with thiophene acting as a dual alkyl-amido base; however, the unexpected dodecasodium hexamagnesium ate complex [{(TMEDA)3Na6Mg3(CH2SiMe3) (2,5-C4H3O)(2-C4H3O)5}2] 39 (Fig. 27A) is isolated.42 This structure is built upon a network containing ten α-deprotonated and six twofold α,α0 -deprotonated furan ligands. The core of the structure represents a unique structural motif in mixed metal chemistry containing twelve Na and six Mg sites, the highest nuclearity uncovered thus far for alkali-metalmediated magnesiation reactions. Perhaps the most useful feature of 35 is its ability to cleave and capture highly sensitive molecules. The bimetallic butadiene-diide containing complex [{(TMEDA)NaMg(TMP)2}2{1,4-C4H4}] 40 (Fig. 27B) was isolated from the reaction of 35 with THF, to induce an example of cleave and capture chemistry through the fragmentation of THF.43 This reaction yields 40 as a result of breaking two C–O bonds and four C–H bonds of THF to produce the dianionic buta-1,3-diene (C4 H4 2
) fragment which has been

Lithium, Sodium, and Potassium Magnesiate Chemistry

21

Fig. 27 (A) Monomeric unit of [{(TMEDA)3Na6Mg3(CH2SiMe3)(2,5-C4H3O)(2-C4H3O)5}2] 39 with TMEDA and CH2SiMe3 groups omitted for clarity. (B) Molecular structure of [{(TMEDA)NaMg(TMP)2}2{1,4-C4H4}] 40.

trapped by two terminal dinuclear [(TMEDA)NaMg(TMP)2]+ cationic residues of the original base [(TMEDA)NaMg(TMP)2 (CH2SiMe3)] 35. The bis amido alkyl complex [(donor)nNaMg(HMDS)2(alkyl
)] 41 (donor, diethyl ether; alkyl, tBu; n ¼ 1) and 42 (donor, TMEDA; alkyl, n Bu; n ¼ 2) are discrete monomeric complexes (Fig. 28).22 Complex 41 is prepared via a metathetical approach by reacting NaHMDS with the Grignard reagent tBuMgCl in a 1:1 molar ratio in the presence of Et2O in hydrocarbon solvent with concomitant NaCl elimination. Complex 42 is prepared by a different synthetic approach involving the deprotonative metalation of HMDS(H) by reacting nBuNa, nBu2Mg, in the presence of TMEDA in 2:1:1:2 molar ratio in hydrocarbon solution. For 41, its structure consists of a four-membered Na–N–Mg–N ring with both the Na and Mg

22

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 28 (A) Molecular structure of [(Et2O)NaMg(HMDS)2(tBu)] 41. (B) Molecular structure of [(TMEDA)2NaMg(HMDS)2(nBu)] 42.

Fig. 29 Molecular structure of [(TMEDA)NaMg(cis-DMP)3] 43.

atoms occupying distorted trigonal planar arrangements. Two bridging HMDS ligands are connecting Na to Mg, and a terminal tBu group is bound to Mg completing its coordination sphere. Complex 42 is best described as a loosely contacted ion pair structure as only a single nBu group bridges Na to Mg. The chelation of two molecules of TMEDA to Na gives rise to a square pyramidal rearrangement, hampering the coordination of a second bridging HMDS amido molecule to Na and preventing the formation of a typical four-membered Na–C-Mg–N ring. The tris-amido sodium magnesium ate complex [(TMEDA)NaMg(cisDMP)3] 43 (Fig. 29) was prepared by a mixed-metalation approach.44 Two cis-DMP ligands bridge the Mg and Na centers while one terminal amido ligand is coordinated to Mg completing its trigonal planar coordination sphere. One molecule of TMEDA ligand chelates to Na. The isolation

Lithium, Sodium, and Potassium Magnesiate Chemistry

23

of 43 has allowed structural comparisons with other related TMP and diisopropylamido magnesiates, and helped postulate that the chemistry of cis-DMP(H) more closely resembles the latter amide. 3.2.1 Solvent-Separated Sodium HMDS Amido Magnesiate Reagents Several solvent-separated sodium magnesiates have been isolated. For instance, 44–47 are all well-defined charge separated ion pair magnesiates which have the generic formula [(donor)2Na]+[Mg(HMDS)3]
(Fig. 30).19,20 The four magnesiates are constructed by a trigonal planar Mg center ligated to three bis(trimethylsilyl)amido ligands in a trigonal planar fashion [Mg(HMDS)3]
. The donor ligands are chelating Lewis basic molecules, TMEDA, (
)-sparteine, and R,R-TMCDA, in 44, 45, and 46, respectively. For 45 and 46, the coordination environment of the Na atom is distorted tetrahedral while an unusual square-planar-like coordination environment is found in 44. Interestingly in the case of 45, the cationic [Na{(
)-sparteine}2]+ constitutes the first example in which the alkali metal center is sequestered by two (
)-sparteine molecules. In 47, the Na atom is

Fig. 30 Molecular structure of (A) [(TMEDA)2Na]+[Mg(HMDS)3]
44, (B) [(R,R-TMCDA)2 Na]+[Mg(HMDS)3]
45, (C) [{(
)-sparteine}2Na]+[Mg(HMDS)3]
46, and (D) [(IPr)2Na]+ [Mg(HMDS)3]
47.

24

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 31 Molecular structure of [Na(THF)6]+[(THF)Mg(nBu){(DippN)2SiPh2}]
48.

coordinated by two N-heterocyclic carbene (NHC) ligands, 1,3-bis(2,6diisopropylphenyl)-imidazol-2-ylidene (IPr). NHCs are well-known two electron σ-donor ligands which steric properties can be easily tuned by modification of the N-bound imidazolyl organic residues. They favor unusual bonding modes and low coordination numbers in complexes containing metals from across the entire periodic table. However, their utilization in s-block systems is relatively recent, and crystallographic data of alkali metal containing examples is limited. Complex 47 is the first example in which a neutral NHC ligand is bound to sodium. The two IPr ligands are σ-bound through the sp2-hybridized-carbenic C atom to the Na in almost a linear array while the harder bis(trimethylsilyl)amido ligands are coordinated to the harder Mg metal, in keeping with 44–46. In general terms, both the anionic [Mg(HMDS)3]
and cationic [Na (donor)2]+ moieties are typical structural motifs for solvent-separated alkali metal containing bimetallic magnesium complexes. A special case of a solvent-separated sodium magnesiate complex is [Na (THF)6]+[(THF)Mg(nBu){(DippN)2SiPh2}]
48 (Fig. 31), where Dipp is 2,6-diisopropylphenyl,45 prepared by reacting [NaMgBu3]34,33 with the bis(silyl)amine Ph2Si(NHDipp)2 in THF. In contrast to the previous examples 44–47, the anionic [MgN2(nBu)(THF)]
moiety contains a Mg atom bonded to a bidentate bis amido ligand and to an alkyl group. The cationic moiety consists of a [Na(THF)6]+ unit. The reactivity of 48 has been studied in magnesiation reactions, and it has been discovered that this bulky magnesiate can induce complex magnesium-mediated transformations. For instance, simple organomagnesium reagents will deprotonate benzothiazole at the 2-position; however, 48 initiates a remarkable cascade process with benzothiazole at ambient

Lithium, Sodium, and Potassium Magnesiate Chemistry

25

Fig. 32 Molecular structure of the sodium magnesiate 49. THF molecules form the [Na (THF)2]+ and [Na(THF)3]+ units have been omitted for clarity.

temperature comprising a sequence of C–H deprotonations, C–C coupling, ring-opening, and nucleophilic addition reactions, forming the novel magnesiate 49 (Fig. 32). Structural studies unveil that the molecular structure contains two similar Mg centers solvated by THF and connected by two newly generated trianionic fragments [{C7H4NS}C{NC6H4S}]3
as a result of this cascade event. The contacted-ion pair magnesiate 49 is completed by two [Na(THF)2]+ and [Na(THF)3]+ units. Complex 48 also reacts as a single mono-alkyl base with Nmethylbenzimidazole to deprotonate the most acidic C2 site at ambient temperature in THF solution resulting in the solvent-separated ion pair derivative


NaðTHFÞ5

 + h 2

  i2
Ph2 SiðNDippÞ2 MgfðN
methylbenzimidazolylÞ
2g 2

50 (Fig. 33). The novel structure of 50 contains two Na cations solvated by only five THF molecules [Na(THF)5]+ and a dinuclear dianionic unit featuring two [{Ph2Si(NDipp)2}Mg{(N-methylbenzimidazolyl}]
units linked by two bridging N-methylbenzimidazolyl ligands via its N- and C-metallated atoms forming a six-membered [Mg–C–N]3 core ring. In addition, each Mg atom is bonded to a bulky bis(amido)silyl amide group [Ph2Si(NDipp)2] which chelate the Mg atom forming a four-membered [Mg–N–Si–N] terminal ring. Emphasizing the versatility and polybasic nature of 48, it can act as a di-basic alkyl-amido reagent, using its single nBu arm and one amido site of its bulky bis(amido)ligand to deprotonate certain substrates. For instance, the sodium magnesiate [{Na(THF)2}{Mg(C4H4N)2{(DippN) SiPh2(DippNH)}}] 51 (Fig. 33B) is formed as a result of its deprotonation reaction with pyrrole.46 Complex 51 is a contacted ion-pair tris(amido) magnesiate in which the sodium atom exhibits π-Na⋯C interactions with

Fig. 33 Molecular structure of (A) [Na(THF)5]+2[{{Ph2Si(NDipp)2}Mg{(N-methylbenzimidazolyl)-2}}2]2
50 (cations not shown), (B) [{Na(THF)2} {Mg(C4H4N)2{(DippN)SiPh2(DippNH)}}] 51, and (C) [Na(THF)6]+[{Ph2Si(NDipp)(NHDipp)}Mg(NHDipp)2(THF)]
52.

Lithium, Sodium, and Potassium Magnesiate Chemistry

27

pyrrole ligands (η5 fashion) while the Mg binds to three amido N atoms. Two molecules of THF complete the coordination sphere of Na. The Mg atom binds to two deprotonated pyrrole ligands and one monodeprotonated bis(amido)silyl ligand. Highlighting the complexity of these deprotonation reactions, when 49 reacts with the primary amine 2,6-diisopropylaniline, the formation of the solvent-separated sodium magnesium ion pair complex [Na(THF)6]+ [{Ph2Si(NDipp)(NHDipp)}Mg(NHDipp)2(THF)]
52 (Fig. 33C) occurred as the result of a double amination process involving both alkyl and amido basic groups of 49. Despite their similar tris(amido) constitution, the sodium magnesiates 51 and 52 exhibit different structural features. In 52, the Mg atom adopts a distorted four-coordinate tetrahedral geometry bonded to one amido(silyl)amine [Ph2Si(NDipp)(NHDipp)] and two N(H)Dipp amido groups (where Dipp is 2,6-diisopropylphenyl group) and a solvating molecule of THF. Complex 52 exhibits a solvent-separated structure where the Na is fully solvated by six molecules of THF in a distorted octahedral manner. The new N(H)Dipp amino group present in the amido(silyl)amine ligand of 52, which is generated by protonation of one of the chelating N atoms of the bis-amido(silyl) ligand of 49, does not coordinate to Mg. Sodium-rich higher-order sodium magnesiates such as the tetra-alkyl magnesiate [(TMEDA)2Na2Mg(nBu)4]33 can direct deprotonative metalation of 1-methylindole in a regioselective manner toward the 2-position. In keeping with the starting material, the product of this reaction is the sodium-rich tetraindol-2-yl magnesium complex [(TMEDA)2Na2Mg{(1-methylindolyl)2}4] 53 (Fig. 34).47 It can be prepared using a 4:1 molar ratio of 1-methylindole to base, mirroring the presence of four basic alkyl arms in the metallating reagent. The four basic n-butyl chains have been replaced by 2-metalated 1-methylindolyl groups giving rise to a tetrahedrally disposed

Fig. 34 Molecular structure of [(TMEDA)2Na2Mg{(1-methylindolyl)-2}4] 53.

28

A.J. Martínez-Martínez and C.T. O’Hara

Mg center. The two terminal Na atoms are linked to the five-membered ring of the indolyl systems via electrostatic cation π-interactions in a η2-manner with the C2 (deprotonated) and C3 (proton-bearing) atoms. Each Na atom is additionally solvated by a chelating TMEDA donor ligand completing their tetrahedral coordination spheres. This high order magnesium ate complex represents the first example of a structure of a C-magnesiate indolyl system.

3.3 Inverse Crown Molecules The donor-free sodium magnesiate complexes with general formula [Na4Mg2(TMP)6(arene-di-ide)] (arene-di-ide ¼ 2,5-C6H3OMe in 54; 3,5C6H3NMe2 in 55; and 3,5-C6H3Me in 56; Fig. 35) are representative examples of inverse crown complexes.48,49 Those complexes are coined inverse crowns in view of their topological but inverse relation to conventional crown ethers in

Fig. 35 Molecular structure of sodium magnesium inverse crowns (A) [Na4Mg2(TMP)6(2,5C6H3OMe)] 54, (B) [Na4Mg2(TMP)6(3,5-C6H3NMe2)] 55, and (C) [Na4Mg2(TMP)6(3,5C6H3Me)] 56.

Lithium, Sodium, and Potassium Magnesiate Chemistry

29

which the Lewis basic heteroatoms (ie, oxygen) of the host rings trap Lewis acid metal guests (ie, Li, Na, K, etc.). Sodium magnesiates 54 and 55 are organometallic intermediates toward the regioselective functionalization via deprotonative metalation of aromatic substrates by the solvent-free sodium magnesium ate complex [Na4Mg2(TMP)6(nBu)2]. They can ultimately be converted to organic products by reaction with appropriate electrophiles. The template base [Na4Mg2(TMP)6(nBu)2] also reacts with anisole to give regioselective 2,5-di-metalation of the arene; with dimethylaniline (an arene which offers steric protection to both ortho-sites) yields di-metalation in a 3,5-regioselective fashion. Regioselective metalation in 3,5-positions of dimethylaniline 55 constitutes the first example in which the metalation at both ortho-sites of an aromatic substituted with a traditionally ortho-directing group has been overridden, breaking the dogma of Directed ortho-Metalation (DoM) effects. Similarly, the solvent-free combination of nBuNa/TMP(H)/Mg (CH2SiMe3)2 in 2:3:1 molar ratios reacts with toluene to give 3,5-di-metalation of the arene, 56. The three complexes 54–56 exhibit similar structural features, two Mg and four sodium cations within a 12-atom metal-TMP host ring, where the Mg cations anchor the corresponding guest arene-di-ide with four Na cations π-bonding in pairs to each metallated C position. However, as expected, the structures of 2,5- (54) and 3,5-dimetallated arene (55 and 56) are subtly different due to the different metalation regioselectivities of the guest substrates. Another example of a sodium magnesium inverse crown is the naphthalen-1,4-diide containing [{Na4Mg2(TMP)4(TTHP)2(1,4-C10H6)}] 57 (where TTHP is 2,2,6-trimethyl-1,2,3,4-tetrahydropyridide) (Fig. 36).

Fig. 36 Molecular structure of sodium magnesium inverse crown [Na4Mg2(TMP)4 (TTHP)2(1,4-C10H6)] 57.

30

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 37 Molecular structure of sodium magnesium inverse crown [Na2Mg2(Ph2C] N)6(Ph2C]NH)2] 58.

It is obtained by reacting naphthalene with the aforementioned solvent-free sodium magnesium template base [Na4Mg2(TMP)6(nBu)2].50 Prior to this result, using conventional alkyllithium or alkali metal bimetallic bases naphthalene, had only been regioselective metalated at the 2-position51 or di-metalated at the 2,6-positions.52 The polymetallic 12-membered Na4Mg2N6 ring resembles that in 54; however, two of the TMP ligands have been transformed into TTHP (2,2,6-trimethyl-1,2,3,4-tetrahydropyridide) ligand, formally by the loss of methane.52a Another inverse crown molecule which has recently been reported contains benzophenone imine species [Na2Mg2(Ph2C]N)6(Ph2C]NH)2] 58 (Fig. 37).35 Its solid-state structure shows a bicyclic arrangement, centered on a planar [Mg2N2] four-membered ring with both Mg atoms bridged above and below the plane by N–Na–N linkers. All the N atoms of the ring system are ketimino anions, and exocyclic ligands are neutral ketamine molecules which bond to the Na cations. The eight-membered dicationic [(NaNMgN)2]2+ ring hosts two Ph2C]N
guest anions which sit above and below the center of the polymetallic ring and bridge the Mg cations.

Lithium, Sodium, and Potassium Magnesiate Chemistry

31

Fig. 38 Molecular structure of the sodium magnesium “open” inverse crown [{(TMEDA)NaMg(TMP)2}2(1,4-C6H4)] 59.

Subjecting benzene to a bis-amido mono-alkyl mixture of NaTMP/ BuMgTMP/TMEDA in a 1:2:2:2 molar ratios readily produces the open magnesiate complex [{(TMEDA)NaMg(TMP)2}2(1,4-C6H4)] 59 (Fig. 38). In this complex, benzene has been converted to a 1,4-dianion. Comparing this complex to compounds like 57, it seems that the addition of TMEDA has resulted in the formation of a molecule akin to an open inverse crown molecule (with the extrusion of neutral NaTMP). The TMEDA ligands chelate to the Na cations. t

3.4 Miscellaneous Sodium Magnesiate Complexes There are several other structural motifs which are prevalent in sodium magnesiate chemistry. The structure of [(TMEDA)2NaMg(CH2SiMe3)2{PhC (NSiMe3)2}] 60 (Fig. 39) displays two anionic alkyl bridging ligands between Mg and Na and a terminal bidentate benzamidinate ligand bound to the Mg cation.53 Two bidentate TMEDA ligands coordinate to the alkali metal to form a discrete monomeric bimetallic contacted ion pair structure. The reaction of [NaMg(NiPr2)3] with two molar equivalents of phenylacetylene in the presence of TMEDA yields the bimetallic complex [(TMEDA)2Na2Mg2(PhC^C)4(NiPr2)2] 61 (Fig. 40).54 This complex exists as a near-linear tetranuclear Na⋯Mg⋯Mg⋯Na chain stitched together by acetylido and amido bridges. It is a contacted ion pair of two terminal [Na(TMEDA)]+ cations and a heteroleptic dinuclear dianion [Mg2(C^CPh)4(NiPr2)2]2
.

32

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 39 Molecular structure of [(TMEDA)2NaMg(CH2SiMe3)2{PhC(NSiMe3)2}] 60.

Fig. 40 Molecular structure of [(TMEDA)2Na2Mg2(PhC^C)4(NiPr2)2] 52.

4. POTASSIUM MAGNESIATE COMPLEXES 4.1 Inverse Crown Molecules The final part of this chapter will focus on the recent chemistry which has been reported involving potassium magnesiate chemistry. The first examples to be discussed involve the characterization of solvent-free potassium magnesiates which contain two TMP and one n-butyl anions per potassium (or magnesium) cation [ie, “KMg(TMP)2nBu”]. Three different oligomeric forms have been reported including the polymeric [KMg(TMP)2nBu]1 62, the tetrameric [KMg(TMP)2nBu]4 63, and the hexameric [KMg (TMP)2nBu]6 64 examples (Fig. 41). The structures of 63 and 64 are architecturally similar to the inverse crowns discussed in Section 3.3; but they still retain the basic n-butyl arm, and as such they have been coined as pre-inverse

Lithium, Sodium, and Potassium Magnesiate Chemistry

33

Fig. 41 Molecular structure of (A) 62 showing the contents of the asymmetric unit, which corresponds to a single turn of the helical chain, (B) 63, (C) 64, and (D) 65.

crowns. Pre-inverse crown 64 has been shown to function as a base toward naphthalene to induce regioselective mono-deprotonation of the arene at the 2-position producing inverse crown [KMg(TMP)2C10H7]6 65.50 Complex 62 exists as an unusual helical polymer, the backbone of which is repeating [KNMgN] units. This chain is supported by K⋯CH2(nBu)⋯K interactions generating a series of four-atom four-element [KNMgC] rings, fused together along the Mg–C edge to another ring of identical composition. Each K cation participates in a shared vertex that links neighboring pairs of doubly fused tetranuclear ring systems, favoring the propagation of the polymeric chain. The Mg center exhibits a distorted trigonal planar arrangement bonded to TMP N atoms of both bridging amido ligand, and the C atom of the nBu anion. Tetrameric 63 and hexameric 64 consist of 16-atom and 24-atom polymetallic inverse crown-type rings, respectively.

34

A.J. Martínez-Martínez and C.T. O’Hara

4.2 Introducing Donors to Potassium Magnesiate Complexes The deprotonation of anisole with the heteroleptic potassium magnesiate [(PMDETA)KMg(TMP)2(CH2SiMe3)] 66 (Fig. 42) was found to serve as a perfect bimetallic system for “structurally tracking” alkali metal mediated ortho-deprotonation transformations.55 Starting from 66 and anisole, the first stage of the reaction produces an ortho-magnesiated anisole intermediate [(PMDETA)KMg(2-C6H4OMe)(TMP)(CH2SiMe3)] 67, but the ultimate product is the bis(amido) ortho-magnesiated anisole complex [(PMDETA)KMg(2-C6H4OMe)(TMP)2] 68 (ie, a TMP ligand has formally been reincorporated into the final product with the elimination of alkane). This structural study provided evidence that the heteroleptic base reacts kinetically through its TMP arm, but ultimately the alkyl group is the thermodynamically more basic ligand. The molecular structures of 66, 67, and 68 (Fig. 42) all contain the same [K-TMP-Mg] backbone unit, chelated by PMDETA at the K cations. Moving from 66 to 67, the terminally disposed ligand on Mg changes from TMP to the alkyl CH2SiMe3. An ortho-deprotonated anisole ligand fills the vacated bridging position in 67. Moving now from 67 to 68, the terminal site on Mg loses the CH2SiMe3 ligand by reaction with TMP(H) and concomitant release of Me4Si but gains a TMP anion.

Fig. 42 Molecular structure of (A) 66, (B) 67, and (C) 68.

Lithium, Sodium, and Potassium Magnesiate Chemistry

35

Fig. 43 Molecular structure of 69.

Mg adopts a distorted trigonal planar geometry within 66–68 and binds to the ortho-C-site of the aromatic system in the deprotonated intermediates 67 and 68. In addition, K engages long π-interaction with ipso/ortho-C atoms bonded to Mg in the anisolyl ligands. A novel potassium tris(amido) magnesiate [(PMDETA)KMg (NPh2)3(THF)] 69 (Fig. 43) was prepared56 by combining an equimolar mixture of benzylpotassium and di-n-butylmagnesium with three molar equivalents of diphenylamine in the presence of THF and PMDETA. The molecular structure of 69 comprises a monomeric dinuclear potassium magnesiate motif. The tetracoordinate (consisting of three amido groups and one THF molecule) Mg cation adopts a distorted tetrahedral arrangement. Two diphenyl amido ligands bridge Mg to K; however, perhaps surprisingly, not via the “hard” amido N anions. Reflecting the soft nature of the heavier K metal, it engages π-interactions (η6-bonding mode) with two phenyl rings, one from each bridging amido group. The second phenyl rings on the bridging diphenylamido ligands do not contribute to the stabilization of K. Two novel potassium tris(amido) magnesiates, namely, [{(
)-sparteine} K+{Mg(HMDS)3}
]n 70 and [{(R,R)-TMCDA}K+{Mg(HMDS)3}
]n 71 (Fig. 44) can be prepared from equimolar mixtures of KHMDS and nBu2Mg reacted with two further molar equivalents of HMDS(H) in a hydrocarbon medium with the corresponding addition of the chiral donor (
)-sparteine and (R,R)-TMCDA molecules, respectively.19 Focusing on 70, it has a polymeric structure. In the asymmetric unit, two [{(
)-sparteine}K+{Mg (HMDS)3}
] ion pairs are linked by an agostic-type K⋯Me interaction with a Me(SiMe2)N unit. Both [Mg(HMDS)3]
anions interact with the two K atoms within the asymmetric unit and a third neighboring

36

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 44 Molecular structure of (A) 70 and (B) 71.

K atom, acting as μ3-bridges, hence supporting the formation of a polymeric arrangement. This is the first example of a metal complex which incorporates the chiral diamine (
)-sparteine as part of a polymeric framework, and also the first example of a K complex containing this chiral diamine. For 71, mirroring the situation for 70, its molecular structure consists of a contacted ion-pair potassium magnesiate [{(R,R)-TMCDA}K+{Mg (HMDS)3}
]1. The asymmetric unit comprises two unique anions [Mg (HMDS)3]
-and cations [{(R,R)-TMCDA}K]+. Four K⋯Me agostic interactions result in the formation of a 12-membered ring motif. As for 70, there are no K-HMDS N-amide interactions. Both [Mg(HMDS)3]
units coordinate with two K atoms within the asymmetric unit and to a third neighboring K atom via long agostic-type interactions, propagating polymerization. In contrast to 70, the extended framework of 71 forms a linear arrangement of alternating small 12-membered and larger 16-membered fused cyclic aggregates. It is the first K adduct containing (R,R)-TMCDA as a chiral donor ligand. Bis(benzene) chromium can act as a donor toward a potassium magnesiate HMDS amide complex. [{K{(C6H6)2Cr}1.5(Mes)}]]+[Mg (HMDS)3]
1 72 was obtained from an attempt to deprotonate bis(benzene) chromium with a bimetallic KHMDS/MgHMDS2 system in the presence of mesitylene.57 As depicted in Fig. 45, compound 72 contains [Mg(HMDS)3]
anions and K+ cations that are coordinated to three bis(benzene)chromium and a mesitylene molecule. Mesitylene is coordinated to the K in a η6-manner, whereas the metallocenes are best described as η3coordinated. Complex 72 is therefore a solvent-separated ion pair complex, where mesitylene solvates the metal center in the extended framework. All three of the bis(benzene)chromium molecules bridge to neighboring K cations, hence the K centers act as trigonal nodes to build a two-dimensional

Lithium, Sodium, and Potassium Magnesiate Chemistry

37

Fig. 45 Molecular structure of (A) 72 and (B) its extended structure. Anions are not shown in (B).

Fig. 46 Molecular structure of 73.

framework with the [Mg(HMDS)3]
unit occupying the interstitial spaces between the layers of adjacent sheets. In the solvent-separated potassium magnesiate [(IPr)2K]+[MgHMDS3]
73 (Fig. 46),20 the K cation is coordinated to two IPr NHC ligands. Complex 73 represents a unique example whereby two neutral IPr ligands are σ-bound to a potassium in a near-linear arrangement forming the [(IPr)2K]+cation of 73. This is the first example of a K complex where the metal is solely coordinated to NHC donor ligands. Three donor solvated potassium tris(alkyl) magnesiates have been characterized since 2007. They are the polymeric lower-order [(C6H6)KMg (CH2SiMe3)3]1 74, and the higher-order [(PMDETA)2K2Mg(CH2SiMe3)4] 75 and [(TMEDA)2K2Mg(CH2SiMe3)4] 76 (Fig. 47).58 Complex 74 has a novel polymeric structure, which is formed by a combination of K–CH2, Mg–CH2 bonds, and medium-long K⋯Me electrostatic interactions. Its monomeric unit comprises a trigonal planar Mg bonded to three

38

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 47 Molecular structure of (A) 74, (B) section of the polymeric network of 74, (C) 75, and (D) 76.

trimethylsilylmethyl groups and a solvent-free K+ ion. Trimethylsilylmethyl groups link Mg to K, with K further engaging in interactions through a CH2 unit and to one Me group. In addition, K π-engages interactions with a molecule of benzene in a η6-manner and forms a long contact with a Me of neighboring trimethylsilylmethyl groups, propagating the polymeric two-dimensional network. The addition of polydentate N-donors, PMDETA, or TMEDA to 74 caused its deaggregation as well as inducing a redistribution process to yield the higher-order potassium magnesiates [(PMDETA)2K2Mg(CH2SiMe3)4] 75 and [(TMEDA)2K2Mg(CH2SiMe3)4] 76, respectively, with the concomitant elimination of neutral Mg(CH2SiMe3)2. The structures of 75 and 76 exhibit similar structural features. Both potassium magnesiates display a central distorted tetrahedral C4-coordinated Mg center flanked by two terminal ionic [(donor)2K]+ units. However, the different denticities of the N-ligands impose different coordination modes of the K+ ion resulting in subtly different metal arrangements. For 75, a linear K⋯Mg⋯K arrangement is observed, whereas for 76, a markedly non-linear K⋯Mg⋯K arrangement is evident. Two further higher-order potassium magnesiates have been reported, both containing a bulky bis(amido)silyl ligand [{(THF)2K}{Mg(N

Lithium, Sodium, and Potassium Magnesiate Chemistry

39

Fig. 48 Molecular structure of (A) 77 and (B) 78.

(Dipp)Si(Me)2N(Dipp)}{K(THF)}] 77 and the donor-free [K2Mg(N (Dipp)Si(Me)2N(Dipp)] 78 (Fig. 48).59 Complexes 77 and 78 are monomeric where the central magnesium cation is (N,N)-coordinated (η2) by two bulky di(amido) ligands. The K cations π-engage to the aromatic substituents of the amido ligand. The only difference between 77 and 78 are that the K cations in the former are also coordinated to THF molecules—one K cation to two, the other to one. The structures exhibit near-linear K⋯Mg⋯K arrangements. Reported by Hanusa, Okuda, and coworkers60 the potassium magnesiate allyl complexes [(THF)KMg(C3H5)] 79 and [(THF)2K2Mg(C3H5)4] 80 (Fig. 49) can be prepared from bis(allyl)magnesium [Mg(C3H5)2] and one or two molar equivalent of allyl potassium [K(C3H5)], respectively, in THF. These potassium magnesium complexes have been investigated as initiators for butadiene polymerization and ethylene oligomerization.

40

A.J. Martínez-Martínez and C.T. O’Hara

Fig. 49 Molecular structure of (A) 79 and (B) 80.

In 79, the Mg center is tetrahedrally coordinated by the THF ligand and by three η1-bonded allyl ligands forming [Mg(C3H5)3(THF)]
units. The K center is coordinated by allyl ligands of four different [Mg(C3H5)3(THF)]
units. The allyl ligands show η3-, η2-, and weak η1-interactions with K resulting in a distorted octahedral coordination geometry. This is the first example of one metal center interacting with six allyl ligands. In 80, the Mg center interacts with four allyl ligands in an η1-fashion resulting in a distorted tetrahedral coordination geometry and in contrast to 79, no Mg THF interactions are observed in the solid state. One of the crystallographically independent K atoms is coordinated by one THF ligand and by four allyl groups (η2 and three η3) resulting in distorted trigonal bipyramidal coordination mode with the η2-allyl and the THF ligand in the axial positions. Interactions between K+ ions and neighboring allyl ligands favor the propagation of the tridimensional network in both potassium magnesiate structures.

4.3 Miscellaneous Potassium Magnesiate Complexes Heterobimetallic potassium magnesium hydrides can be prepared by a selective σ-bond metathesis route from the corresponding s-block amido alkyls bearing the amido group HMDS. For instance, the potassium hydrido magnesiate 81 (Fig. 50) can be obtained by reacting [KMg(HMDS)2nBu] (in situ generated) with PhSiH3 as a source of hydride. This protocol provides a powerful methodology for the selective Mg–C/Si–H σ-bond metathesis.39 The molecular structure of 81 is a potassium magnesium hydrido species, which contains a Mg
H
Mg bridge and the eight-membered metal-amide ring of 81 is chair shaped. The Mg atoms are forced into a highly distorted tetrahedral coordination environment by coordination to two HMDS bridging ligand and a hydrido group, and the “naked” K+ ion participates

Lithium, Sodium, and Potassium Magnesiate Chemistry

41

Fig. 50 Molecular structure of 81.

Fig. 51 Molecular structure of 82.

in the stabilization of the hydrido ligand by engaging long K⋯H interactions. Additionally, this hydrido species can be considered as an inverse crown complex as alluded to by Mulvey and coworkers previously in a similar scenario.61 This complex constitutes the most recent addition of a hydrido species to the inverse crown family. Metal
metal bonding constitutes an important area of chemistry, which attracts much attention. A number of metal–metal bonds involving both pand d-block metals have been reported in recent years (silicon
silicon triple,62 and chromium
chromium quintuple63 bond). In 2004 Carmona and coworkers isolated the first stable compound containing a Zn
Zn bond.64,65 In this context, the complex [(THF)3K]2[LMg
MgL] 82,66 where L is ½ð2, 6
i Pr2 C6 H3 ÞNCðMeÞ2 2
, represents a Mg
Mg-bonded compound stabilized by a doubly reduced α-diimine ligand (Fig. 51).

42

A.J. Martínez-Martínez and C.T. O’Hara

It was prepared by reduction of a mixture of MgCl2 and the ligand L with K metal in THF. Its molecular structure comprises a dimeric structure with a Mg
Mg bond length of 2.9370(18) A˚, as a salient feature.

5. SUMMARY This chapter has demonstrated the recent progress made in understanding the solid-state structure of alkali metal magnesiates—key research as structure is inextricable linked to reactivity. The advantages that ate complexes have over conventional lithium reagents (ie., use of milder reaction conditions, better functional group tolerance, access to hitherto inaccessible synthetic chemistry, etc.) ensure that this area of research will blossom further in the coming years. This chapter focused only magnesiate systems. In addition, alkali metal zincate, aluminate, manganate, and cuprate systems are also the subjects of continual study and development, and it is highly likely that alkali metal ate chemistry will have an important future role in synthesis, and complement the massively important and longstanding role that single metal organometallic species such as organolithiums play in academia and industry.

REFERENCES 1. Schorigin P. Synthesis by sodium and halogen alkyls. Ber Dtsch Chem Ges. 1908;41: 2711–2717. 2. Schorigin P. A new synthesis of aromatic carboxylic acid from hydrocarbons. II. Announcement. Ber Dtsch Chem Ges. 1910;43:1938–1942. 3. Schlosser M. Organometallics in Synthesis Third Manual. Hoboken, NY: John Wiley & Sons, Inc.; 2013. 4. Hickey MR, Allwein SP, Nelson TD, et al. Process development and pilot plant synthesis of methyl 2-bromo-6-chlorobenzoate. Org Process Res Dev. 2005;9(6): 764–767. 5. Mulvey RE, Mongin F, Uchiyama M, Kondo Y. Deprotonative metalation using ate compounds: synergy, synthesis, and structure building. Angew Chem Int Ed. 2007; 46(21):3802–3824. 6. Mulvey RE. Avant-garde metalating agents: structural basis of alkali-metal-mediated metalation. Acc Chem Res. 2009;42(6):743–755. 7. O’Hara CT. Synergistic effects in the activation of small molecules by s-block elements. In: Fairlamb IJS, Lynam JM, eds. Organometallic Chemistry. vol. 37. Cambridge, UK: Royal Society of Chemistry; 2011:1–26. 8. Harrison-Marchand A, Mongin F. Mixed AggregAte (MAA): a single concept for all dipolar organometallic aggregates. 1. Structural data. Chem Rev. 2013;113(10): 7470–7562. 9. Mongin F, Harrison-Marchand A. Mixed AggregAte (MAA): a single concept for all dipolar organometallic aggregates. 2. Syntheses and reactivities of homo/heteroMAAs. Chem Rev. 2013;113(10):7563–7727.

Lithium, Sodium, and Potassium Magnesiate Chemistry

43

10. Mulvey RE, O’Hara CT. Mixed lithium complexes: structure and application in synthesis. In: Luisi R, Capriati V, eds. Lithium Compounds in Organic Synthesis. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA; 2014. 11. Mulvey RE, Robertson SD. FascinATES: mixed-metal ate compounds that function synergistically. In: Xi Z, ed. Organo-Di-Metallic Compounds. vol. 47. Heidelberg: Springer-Verlag; 2014:129–158. 12. Tilly D, Chevallier F, Mongin F, Gros PC. Bimetallic combinations for dehalogenative metalation involving organic compounds. Chem Rev. 2014;114(2):1207–1257. 13. Wittig G, Meyer FJ, Lange G. Uber das Verhalten von Diphenylmetallen als Komplexbildner. Justus Liebigs Ann Chem. 1951;571(3):167–201. 14. Baillie SE, Clegg W, Garcı´a-A´lvarez P, et al. Synthesis, structural elucidation, and diffusion-ordered NMR studies of homoleptic alkyllithium magnesiates: donor-controlled structural variations in mixed-metal chemistry. Organometallics. 2012;31(14): 5131–5142. 15. Langer J, Krieck S, Fischer R, G€ orls H, Walther D, Westerhausen M. 1,4-Dioxane adducts of Grignard reagents: synthesis, ether fragmentation reactions, and structural diversity of Grignard reagent/1,4-dioxane complexes. Organometallics. 2009;28(19): 5814–5820. 16. Fischer R, Suxdorf R, G€ orls H, Westerhausen M. Synthesis, crystal structures, and solution behavior of organomagnesium derivatives of alkane-1,4-diide as well as -1,5-diide. Organometallics. 2012;31(21):7579–7585. 17. Wei J, Liu L, Zhan M, Xu L, Zhang W-X, Xi Z. Magnesiacyclopentadienes as alkalineearth metallacyclopentadienes: facile synthesis, structural characterization, and synthetic application. Angew Chem Int Ed. 2014;53(22):5634–5638. 18. Coles MP, Hitchcock PB. Bicyclic guanidinates in mono- and di-valent metal complexes, including group 1/2 and group 1/12 heterometallic systems. Angew Chem Int Ed. 2013;66(10):1124–1130. 19. Garcia-Alvarez P, Kennedy AR, O’Hara CT, Reilly K, Robertson GM. Synthesis and structural chemistry of alkali metal tris(HMDS) magnesiates containing chiral diamine donor ligands. Dalton Trans. 2011;40(19):5332–5341. 20. Hill MS, Kociok-K€ ohn G, MacDougall DJ. N-Heterocyclic carbenes and charge separation in heterometallic s-block silylamides. Inorg Chem. 2011;50(11):5234–5241. 21. Yang D, Ding Y, Wu H, Zheng W. Synthesis and structural characterization of alkaline-earth-metal bis(amido)silane and lithium oxobis(aminolato)silane complexes. Inorg Chem. 2011;50(16):7698–7706. 22. Andrikopoulos PC, Armstrong DR, Kennedy AR, et al. Synthesis and structural characterisation of mixed alkali metal–magnesium mixed ligand alkyl-amido ate complexes. Inorg Chim Acta. 2007;360(4):1370–1375. 23. Walton MJ, Lancaster SJ, Redshaw C. Highly selective and immortal magnesium calixarene complexes for the ring-opening polymerization of rac-lactide. ChemCatChem. 2014;6(7): 1892–1898. ´ lvarez P, Graham DV, Hevia E, et al. Unmasking representative structures of 24. Garcı´a-A TMP-active Hauser and turbo-Hauser bases. Angew Chem Int Ed. 2008;47(42): 8079–8081. 25. Armstrong DR, Garcı´a-A´lvarez P, Kennedy AR, Mulvey RE, Parkinson JA. Diisopropylamide and TMP turbo-Grignard reagents: a structural rationale for their contrasting reactivities. Angew Chem Int Ed. 2010;49(18):3185–3188. 26. Haag B, Mosrin M, Ila H, Malakhov V, Knochel P. Regio- and chemoselective metalation of arenes and heteroarenes using hindered metal amide bases. Angew Chem Int Ed. 2011;50(42):9794–9824. 27. Francos J, Gros PC, Kennedy AR, O’Hara CT. Structural studies of (rac)-BIPHEN organomagnesiates and intermediates in the halogen–metal exchange of 2-bromopyridine. Organometallics. 2015;34(11):2550–2557.

44

A.J. Martínez-Martínez and C.T. O’Hara

28. Heitz S, Epping J-D, Aksu Y, Driess M. Molecular heterobimetallic approach to Li-containing MgO nanoparticles with variable Li-concentrations using lithiummethylmagnesium alkoxide clusters. Chem Mater. 2010;22(16):4563–4571. 29. Kruczy nski T, Pushkarevsky N, Henke P, et al. Hunting for the magnesium(I) species: formation, structure, and reactivity of some donor-free Grignard compounds. Angew Chem Int Ed. 2012;51(36):9025–9029. 30. Reichert A, Schmidt J, Bolte M, Wagner M, Lerner H-W. Magnesiation of N-methyl-1, 3-propylenediaminoboryl ferrocene. Z Anorg Allg Chem. 2013;639(7): 1083–1086. 31. Henderson KW, Kennedy AR, Mulvey RE, O’Hara CT, Rowlings RB. Trimagnesium-bridged trinuclear ferrocenophanes cocomplexed with solvated mononuclear alkali metal amide molecules. Chem Commun. 2001;(17):1678–1679. 32. Baillie SE, Clegg W, Garcia-Alvarez P, et al. Synthesis and characterization of an infinite sheet of metal–alkyl bonds: unfolding the elusive structure of an unsolvated alkali-metal trisalkylmagnesiate. Chem Commun. 2011;47(1):388–390. 33. Andrikopoulos PC, Armstrong DR, Hevia E, Kennedy AR, Mulvey RE, O’Hara CT. Stoichiometrically-controlled reactivity and supramolecular storage of butylmagnesiate anions. Chem Commun. 2005;(9):1131–1133. 34. Hevia E, Henderson KW, Kennedy AR, Mulvey RE. Synthesis and characterization of new mixed-metal sodium–magnesium enolates derived from 2,4,6-trimethylacetophenone. Organometallics. 2006;25(7):1778–1785. 35. Clegg W, Dale SH, Graham DV, et al. Structural variations in bimetallic sodium– magnesium and sodium–zinc ketimides, and a sodium–zinc alkide–alkoxide–amide: connections to ring-stacking, ring-laddering, and inverse crown concepts. Chem Commun. 2007;(16):1641–1643. 36. Francos J, Fleming BJ, Garcia-Alvarez P, et al. Complexity in seemingly simple sodium magnesiate systems. Dalton Trans. 2014;43(38):14424–14431. 37. Hernan-Gomez A, Bradley TD, Kennedy AR, Livingstone Z, Robertson SD, Hevia E. Developing catalytic applications of cooperative bimetallics: competitive hydroamination/trimerization reactions of isocyanates catalysed by sodium magnesiates. Chem Commun. 2013;49(77):8659–8661. 38. Baillie SE, Blair VL, Hevia E, Kennedy AR. A new polymeric alkyl/alkoxide magnesiumsodium inverse crown complex. Acta Crystallogr C. 2011;67(7):m249–m251. 39. Liptrot DJ, Hill MS, Mahon MF. Heterobimetallic s-block hydrides by σ-bond metathesis. Chem Eur J. 2014;20(32):9871–9874. 40. Kennedy AR, O’Hara CT. Isolation and characterisation of a (
)-sparteine coordinated mixedalkyl/amido sodium magnesiate, a chiral variant of an important utility ate base. Dalton Trans. 2008;(37):4975–4977. 41. Blair VL, Kennedy AR, Mulvey RE, O’Hara CT. Sodium-mediated magnesiation of thiophene and tetrahydrothiophene: structural contrasts with furan and tetrahydrofuran. Chem Eur J. 2010;16(29):8600–8604. 42. Blair VL, Kennedy AR, Klett J, Mulvey RE. Structural complexity of the magnesiation of furan: an octadecanuclear product with a subporphyrin-like Mg3(2,5-fur-di-yl)3 substructure. Chem Commun. 2008;(42):5426–5428. 43. Mulvey RE, Blair VL, Clegg W, Kennedy AR, Klett J, Russo L. Cleave and capture chemistry illustrated through bimetallic-induced fragmentation of tetrahydrofuran. Nat Chem. 2010;2(7):588–591. 44. Campbell R, Conway B, Fairweather GS, et al. cis-2,6-Dimethylpiperidide: a structural mimic for TMP (2,2,6,6-tetramethylpiperidide) or DA (diisopropylamide)? Dalton Trans. 2010;39(2):511–519. 45. Blair VL, Clegg W, Kennedy AR, Livingstone Z, Russo L, Hevia E. Magnesiummediated benzothiazole activation: a room-temperature cascade of C–H deprotonation,

Lithium, Sodium, and Potassium Magnesiate Chemistry

45

C–C coupling, ring-opening, and nucleophilic addition reactions. Angew Chem Int Ed. 2011;50(42):9857–9860. 46. Armstrong DR, Clegg W, Hernan-Gomez A, et al. Probing the metallating ability of a polybasic sodium alkylmagnesiate supported by a bulky bis(amido) ligand: deprotomagnesiation reactions of nitrogen-based aromatic substrates. Dalton Trans. 2014;43(11):4361–4369. 47. Conway B, Hevia E, Kennedy AR, Mulvey RE. Structurally-defined direct C-magnesiation and C-zincation of N-heterocyclic aromatic compounds using alkali-metalmediated metallation. Chem Commun. 2007;(27):2864–2866. 48. Martinez-Martinez AJ, Kennedy AR, Mulvey RE, O’Hara CT. Directed ortho-meta0 and meta-meta0 -dimetalations: a template base approach to deprotonation. Science. 2014;346(6211):834–837. 49. Blair VL, Carrella LM, Clegg W, et al. Tuning the basicity of synergic bimetallic reagents: switching the regioselectivity of the direct dimetalation of toluene from 2,5- to 3,5-positions. Angew Chem Int Ed. 2008;47(33):6208–6211. 50. Martinez-Martinez AJ, Armstrong DR, Conway B, et al. Pre-inverse-crowns: synthetic, structural and reactivity studies of alkali metal magnesiates primed for inverse crown formation. Chem Sci. 2014;5(2):771–781. 51. Gilman H, Bebb RL. Relative reactivities of organometallic compounds. XX. Metalation. J Am Chem Soc. 1939;61(1):109–112. 52. Clegg W, Dale SH, Hevia E, et al. Alkali-metal-mediated zincation of polycyclic aromatic hydrocarbons: synthesis and structures of mono- and dizincated naphthalenes. Angew Chem Int Ed. 2006;45(39):6548–6550. 52a. Kennedy AR, Leenhouts SM, Liggat JJ, et al. Dehydromethylation of alkali metal salts of the utility amide 2,2,6,6-tetramethylpiperidide (TMP). Chem Commun. 2014;50(73):10588–10591. 53. Forret R, Kennedy AR, Klett J, Mulvey RE, Robertson SD. Multistep self-assembly of heteroleptic magnesium and sodium–magnesium benzamidinate complexes. Organometallics. 2010;29(6):1436–1442. 54. Garcia-Alvarez J, Graham DV, Hevia E, Kennedy AR, Mulvey RE. Synthesis and characterisation of new bimetallic alkali metal–magnesium mixed diisopropylamideacetylides: structural variations in bimetallic lithium- and sodium-heteroleptic magnesiates. Dalton Trans. 2008;(11):1481–1486. 55. Clegg W, Conway B, Hevia E, McCall MD, Russo L, Mulvey RE. Closer insight into the reactivity of TMP-dialkyl zincates in directed ortho-zincation of anisole: experimental evidence of amido basicity and structural elucidation of Key reaction intermediates. J Am Chem Soc. 2009;131(6):2375–2384. 56. Fleming BJ, Garcı´a-A´lvarez P, Keating E, Kennedy AR, O’Hara CT. Synthesis and structural elucidation of a rare example of a tris(amido) potassium magnesiate. Inorg Chim Acta. 2012;384:154–157. 57. Morris JJ, Noll BC, Honeyman GW, et al. Organometallic polymers assembled from cation–π interactions: use of ferrocene as a ditopic linker within the homologous series [{(Me3Si)2NM}2(Cp2Fe)]1 (M ¼ Na, K, Rb, Cs; Cp ¼ cyclopentadienyl). Chem Eur J. 2007;13(16):4418–4432. 58. Baillie SE, Bluemke TD, Clegg W, et al. Potassium-alkyl magnesiates: synthesis, structures and Mg-H exchange applications of aromatic and heterocyclic substrates. Chem Commun. 2014;50(85):12859–12862. 59. Pi C, Wan L, Gu Y, et al. Synthesis of potassium–magnesium ate complexes with a bulky diamido ligand. Organometallics. 2009;28(17):5281–5284. 60. Lichtenberg C, Spaniol TP, Peckermann I, Hanusa TP, Okuda J. Cationic, neutral, and anionic allyl magnesium compounds: unprecedented ligand conformations and reactivity toward unsaturated hydrocarbons. J Am Chem Soc. 2013;135(2):811–821.

46

A.J. Martínez-Martínez and C.T. O’Hara

61. Andrikopoulos PC, Armstrong DR, Kennedy AR, Mulvey RE, O’Hara CT, Rowlings RB. Synthesis, structure and theoretical studies of the hydrido inverse crown [K2Mg2(NiPr2)4(μ-H)2  (toluene)2]: a rare example of a molecular magnesium hydride with a Mg-(μ-H)2-Mg double bridge. Eur J Inorg Chem. 2003;2003(18):3354–3362. 62. Sekiguchi A, Kinjo R, Ichinohe M. A stable compound containing a silicon-silicon triple bond. Science. 2004;305(5691):1755–1757. 63. Nguyen T, Sutton AD, Brynda M, Fettinger JC, Long GJ, Power PP. Synthesis of a stable compound with fivefold bonding between two chromium(I) centers. Science. 2005;310(5749):844–847. 64. Resa I, Carmona E, Gutierrez-Puebla E, Monge A. Decamethyldizincocene, a stable compound of Zn(I) with a Zn-Zn bond. Science. 2004;305(5687):1136–1138. 65. Carmona E, Galindo A. Direct bonds between metal atoms: Zn, Cd, and Hg compounds with metal–metal bonds. Angew Chem Int Ed. 2008;47(35):6526–6536. 66. Liu Y, Li S, Yang X-J, Yang P, Wu B. Magnesium–magnesium bond stabilized by a doubly reduced α-diimine: synthesis and structure of [K(THF)3]2[LMg–MgL] (L ¼[(2,6-iPr2C6H3)NC(Me)]22
). J Am Chem Soc. 2009;131(12):4210–4211.

CHAPTER TWO

Dearomatization of Transition Metal-Coordinated N-Heterocyclic Ligands and Related Chemistry R. Arévaloa, M. Espinal-Viguria, M.A. Huertosa, J. Péreza,b,*, L. Rieraa,b,* a

Departamento de Quı´mica Orga´nica e Inorga´nica, Universidad de Oviedo, Oviedo, Spain Centro de Investigacio´n en Nanomateriales y Nanotecnologı´a (CINN), CSIC-Universidad de Oviedo-Principado de Asturias, El Entrego, Spain *Corresponding authors: e-mail address: [email protected]; [email protected] b

Contents 1. Introduction 2. Addition of Main Group Organometallic Reagents to Pyridines and Related Heteroaromatics 3. Reaction of Pyridines with Early Transition Metal Complexes 4. The Elusive Nucleophilic Attack on Transition Metal-Coordinated 2,20 -Bipyridine and 1,10-Phenanthroline 5. Reactivity of Carbonyl Rhenium and Molybdenum Complexes 5.1 Reactions of the Phosphido Complex [Re(PPh2)(CO)3(phen)] with Activated Acetylenes and Olefins 5.2 Dearomatization by Deprotonation of Complexes with N-Alkylimidazole Ligands 5.3 Couplings and Dearomatization Initiated by Deprotonation of C(sp3)–H Bonds 5.4 Coupling Between Two Monodentate Heterocyclic Ligands 5.5 Ring Opening of Imidazole Rings 5.6 Couplings Between N-Alkylimidazoles and Imine Ligands 5.7 Coupling Between N-Alkylimidazoles and Nitriles or Isonitriles 5.8 Deprotonation of N-Alkylimidazole Ligands in Phosphane–Carbonyl Complexes 6. Metal-Mediated Tautomerization of N-Heterocycles to NHC Complexes 6.1 Tautomerization of Pyridines 6.2 Tautomerization of Imidazoles 6.3 Tautomerization of Other Azoles 7. Metal–Ligand Cooperation Based on Aromatization/Dearomatization Processes 8. Conclusions and Outlook Acknowledgments References

Advances in Organometallic Chemistry, Volume 65 ISSN 0065-3055 http://dx.doi.org/10.1016/bs.adomc.2016.03.002

#

2016 Elsevier Inc. All rights reserved.

48 51 56 60 61 61 64 68 71 79 79 82 86 88 89 92 101 101 104 106 106

47

48

R. Arévalo et al.

1. INTRODUCTION This chapter aims to review new aspects of transition metal-based reactivity of pyridine and imidazole ligands, many of which involve dearomatization of those heteroaromatic rings. Six-member pyridines (1, Fig. 1) and five-member imidazoles (2, Fig. 1) are some of the most important N-heterocycles.1 There are several reasons for this relevance. One is the prevalence of pyridine and imidazole rings in natural systems and in nonnatural molecules with physiological activity. Imidazole groups of histidine residues are some of the main donor groups toward metals in metalloproteins, including enzymes. This has been perhaps the main reason why imidazoles or imidazole-containing ligands have been the subject of many studies in coordination chemistry.2 Despite this wealth of work, few studies dealt with the reactivity of coordinated imidazoles beyond substitution reactions and proton transfer reaction (including hydrogen bond formation) to and from a noncoordinated imidazole nitrogen atom. N-alkylimidazolium salts (3, Fig. 1) have been thoroughly studied materials; for instance, as ionic liquids.3 One of the ways to access such salts is by alkylation of free N-alkylimidazoles. The deprotonation at C2 of N-alkylimidazolium cations is one of the most general ways to prepare N-heterocyclic carbenes (NHCs, 4 in Fig. 1), a type of ligands that are having a huge impact in contemporary chemistry.4 In particular, NHCs are employed as ligands in many catalytic processes, where their strong donor character, high steric profile, strong affinity for metal centers, leading to the high thermal stability of their complexes, and tunability, make them useful auxiliaries. Applications of NHC metal complexes outside the realm of

Fig. 1 Main types of N-heterocyclic species dealt with in this work.

Dearomatization of N-Heterocyclic Ligands

49

catalysis are currently being found as a result of intense research in this area.5 Despite the close relationship between N-alkylimidazoles and NHC ligands, few transformations linking the two types of compounds were known. Pyridines, being both good σ-donor and π-acceptor, are excellent ligands for all kinds of metal fragments. The ability of pyridines to accept electron density in their low-lying π-symmetric empty orbitals make them redox-active ligands, and is also the basis of the rich photophysic behavior of many metal complexes containing pyridine-based ligands, in particular rhenium carbonyl complexes. These properties have been particularly exploited for metal complexes of the strongly chelating 2,2-bipyridine (bipy, 5 in Fig. 1) and 1,10-phenanthroline (phen, 6 in Fig. 1) and their derivatives, which have been employed as ligands for a long time.6,7 The complexes of monodentate pyridines are often rather labile, a fact that considerably restricts their applications and the study of the reactivity of the coordinated pyridines beyond their substitution. In contrast, several types of pyridinebased polydentate ligands, including the relatively simple bipy and phen, form robust metal complexes that are very stable against substitution. Pyridyl-based ligands are typically resistant toward oxidation8 and very rarely have been found to undergo chemical modification of the pyridyl ring. However, it must be noted that this is not the result of an inert nature of pyridines. Pyridines are electron-poor aromatic heterocycles in which one of the CH groups of benzene has been replaced by the more electronegative nitrogen atom.1 Therefore, pyridines react with nucleophiles, which attack mainly the ortho and para positions, those with a higher positive charge. The product of such nucleophilic additions is a nonaromatic amide with a high tendency to recover aromaticity, typically by a subsequent elimination step, so that the overall addition–elimination reaction is formally a substitution. Indeed, the tendency to give substitution, as opposed to addition, products, is typical of aromatic substrates. If the pyridine does not contain a good leaving group, such as a halide, the second step involves a formal elimination of hydride, and the overall transformation is called an aromatic hydrogen substitution.9 Alternatively, aromaticity can be regained by oxidation, formally a hydride abstraction, often in fact a combination of electron transfer and proton abstraction steps.10 While those reactions are well known of free pyridines, including bipy and phen, there was virtually no precedent of nucleophilic attack to transition metal-coordinated pyridines. For monodentate pyridines, this can be attributed to their lability; however, this cannot explain the lack of examples of nucleophilic attack on coordinated bipy or phen ligands, especially given the vast number of their

50

R. Arévalo et al.

transition metal complexes and how extensively they have been employed in every phase of coordination chemistry. Neither can it be attributed to lack of interest; in fact, there has been a considerable interest in demonstrating nucleophilic attacks to transition metal-coordinated bipy and phen ligands following Gillard’s controversial suggestions that such attacks take place to the bipy and phen ligands of some cationic metal complexes in basic aqueous solution, as it will be discussed later. Metal-mediated transformations of pyridines are thus mainly the realm of main group highly electron-rich compounds such as alkyl derivatives of lithium, magnesium and zinc, aluminum hydrides, etc. This chemistry will be very succinctly covered with an emphasis on recent developments, as it will the related alkyl migration to pyridines which have been found in a number of f-block derivatives, a topic which has been recently reviewed. In particular, rollover cyclometalations, which are important processes that can be operative in bipy complexes, will not be mentioned, and the interested reader is referred to a recent account of this chemistry.11 Most of this chapter will focus on examples of dearomatization of pyridyl rings of coordinated bipy and phen. Common features of all these reactions are the very mild conditions under which they have been carried out and their intramolecular nature. In one instance, the reaction is initiated by the nucleophilic attack of a terminal phosphanido ligand on electron-poor alkenes and alkynes. In all the other examples, it is the result of the deprotonation of one of the CH groups of a ligand. Metal-coordinated N-alkylimidazoles, dimethylsulfide, trimethylphosphane, and monodentate pyridines have been demonstrated to generate, on deprotonation, internal nucleophiles, which then carry out the C–C coupling with proximal metalcoordinated heteroaromatic rings. Deprotonated N-alkylimidazoles have been coupled also with nonaromatic ligands such as imines, nitriles, isonitriles, and CO. Deprotonated monodentate pyridines have been coupled with other pyridine ligands, affording 2,2ʹ-bipyridine chelates within the metal template. A key feature in designing such coupling reactions is that the deprotonated ligand must be cis to the electrophilic ligand to which the attack is planned to occur, so that the nucleophile resulting from the deprotonation and its electrophilic counterpart are close in the space and can couple without leaving the metal coordination sphere. Many dearomatized products have been characterized in solution; in several examples, in addition, they have been isolated as crystalline solids and structurally characterized by single-crystal X-ray diffraction. A noteworthy feature of the C–C coupling events mentioned earlier is that they occur between

Dearomatization of N-Heterocyclic Ligands

51

partners that do not require functionalization other than coordination to the metal, therefore constituting a modular, metal-templated approach to the direct synthesis of metal complexes containing elaborate polydentate ligands from complexes of the monodentate ligands that serve as building blocks.

2. ADDITION OF MAIN GROUP ORGANOMETALLIC REAGENTS TO PYRIDINES AND RELATED HETEROAROMATICS The addition of main group-based organometallic or hydride reagents to pyridines and similar systems takes place through coordination of the heterocyclic molecule to the Lewis acid metal center followed by formation of dearomatized species, for which there is a small but growing body of structural information. The addition of nucleophiles to pyridines, pyridinium cations, and other electron-poor heteroaromatic substrates has been known for a long time.12–14 An example is the Chichibabin amination, in which typically sodium amide adds to the 2-position of a pyridine.15–17 Reactions of this kind are thought to proceed through an initial step in which the pyridine coordinates the main group metal cationic center, followed by the addition of the nucleophile (an alkyl or hydride) to the 2-position of the pyridine to afford a nonaromatic intermediate, the Meisenheimer adduct. From this species, rearomatization to the 2-substituted pyridine takes place through formal hydride elimination, as showed in Scheme 1. This kind of mechanism is generally operative in the chemistry of electron-poor arenes.18 Note, however, that the Chichibabin reaction has been carried out in a variety of conditions, including homogeneous and heterogeneous media, so different mechanisms can reasonably be expected. An elegant and synthetically useful fluorination of pyridines and diazines, inspired by the Chichibabin reaction, has been recently reported by Fier and Hartwig.19 Interestingly, in the (to our knowledge) only theoretical study of the mechanism of the Chichibabin reaction, Dransfield and coworkers found that the most

Scheme 1 Proposed mechanism of the Chichibabin reaction.

52

R. Arévalo et al.

favorable mechanism involves the elimination of dihydrogen, rather than sodium hydride as it is usually assumed.20 Alternative mechanisms involve heteroarynes or ring opening–ring closure sequences.21,22 Main group alkyl and, in some cases, hydride reagents, add to pyridines (the reaction is sometimes called carbometalation, or pyridine insertion into the M–C bond) affording either the dearomatized N-metalo-1,2-dihydropyridines or their 1,4-isomers. Usually the 1,2-isomers are produced under kinetic control and the 1,4-isomers, under thermodynamic control. Often mixtures of both isomers are obtained, although regioselectivity has been achieved under certain conditions. These main group complexes of dihydropyridines have been found to be the intermediates in the reactions between pyridines and organometallic reagents to afford alkylpyridines. In some instances, di- and trialkylpyridines have been prepared employing an excess of the organometallic reagent.23 Note that the classical syntheses of dihydropyridines (eg, Hantzsch synthesis) employ de novo construction of the rings by multicomponent reactions rather than starting from pyridines, and that some of the pyridine reduction processes are actually reductions of pyridinium cations (eg, Fowler’s sodium borohydride reduction of the N-acylpyridinium salts produced in situ by the reaction of pyridines with chloroformate esters). Neutral N–H or N–R dihydropyridines are employed both by nature (the NADH and NADPH dinucleotide coenzymes) and by synthetic chemists as selective reducing agents in transformations driven in large part by the tendency of dihydropyridines to achieve rearomatization. In the field of transition metal organometallic chemistry, dihydropyridines have been either formed on the coordination sphere of a metal to which they are bound through the C]C double bonds, or used as substrates. Ishitani studied the formation of intermediate complexes containing 1,4-dihydropyridines η2coordinated to ruthenium fragments in the reaction of pyridinium cations with ruthenium hydride complexes,24 Myers employed complexes of a strongly π-basic metal fragment with η2-coordinated pyridinium cations to access metal-complexed dihydropyridines,25 Rudler found dihydropyridines to be useful reagents in the chemistry of Fischer carbenes,26 and Liebeskind employed dihydropyridine π-complexes as intermediated in the synthesis of more reduced heterocycles, such as piperidines.27 The reaction of pyridines with organolithium reagents afforded mainly 2-substituted pyridines through alkyl addition to generate a dearomatized lithium complex of dihydropyridine. From this species, it has been though that elimination of highly insoluble lithium hydride would yield the final

Dearomatization of N-Heterocyclic Ligands

53

2-alkylpyridine product. Reaction mixtures usually contain some amount of the 4-alkylated isomer. Snaith and coworkers found that rather than undergoing an actual elimination of lithium hydride, N-lithium dihydropyridines hydrolize (eg, with traces of moisture) to N–H dihydropyridines, which subsequently oxidize (eg, on atmospheric exposure) to 2-alkylpyridines.28 Mulvey and coworkers reported evidence pointing to intramolecular hydride transfer from a dihydropyridide ligand to a pyridine ligand within a lithium complex (Scheme 2), and the crucial role of the stoichiometry in the reactions between pyridines and organolithium reagents.29 These results indicate the necessity of considering the entire coordination compound and to gain as much knowledge as possible of its structure both to understand and to synthetically exploit the reactivity of main group organometallics. Mulvey and coworkers recently isolated and structurally characterized several N-lithio-2-alkyl-dihydropyridines.30 Methods consisting of nucleophilic addition followed by oxidative rearomatization are some of the simpler and more expedient methods for pyridine functionalization. However, preactivation of the pyridine by its transformation to a pyridine N-oxide or an N-alkyl- or N-acyl pyridinium cation is often required. In 2013 and 2014, Knochel and coworkers reported nucleophilic additions to pyridines mediated by BF3.31,32 They found that the formation of pyridine-BF3 adducts has a dramatic effect on the reaction course. Indeed, some reactions do not occur without the presence of BF3 and, in other instances, the regioselectivity toward the product of addition to the 4-position is attributed to BF3 complexation. A similar effect of pyridine activation by its complexation has been found by Urabe and coworkers in the additions of benzyl Grignard reagents.33 In the reactions between pyridines, in particular bipy and phen, and other electron-poor N-heterocycles, and main group organometallic or hydride reagents, the formation of strongly colored solutions have been noted a long time ago, and bipy and phen have been used as visual indicators

Scheme 2 Proposed hydride transfer between ligands in a lithium complex.

54

R. Arévalo et al.

in the titrations of organolithium and Grignard reagents.34 However, attempts to isolate stable metal complexes from these reactions failed. Kaim and others showed that, instead of the previously assumed polar mechanism (carbanion transfer from the metal to the heterocycle), many reactions between the metal reagent (alkyllithium, organomagnesium, organozinc, aluminum hydride, etc.) and the heterocyclic substrate take place through a radical mechanism.35–37 Thus the initially formed adduct between the metal reagent and the heterocycle (a charge transfer adduct) undergoes an intramolecular single electron transfer to form a radical anion of the heterocyclic substrate and the radical cation of the metal reagent. One of the possible pathways from that system leads to the diamagnetic product of alkyl or hydride transfer to the heterocycle. Trapping of radicals and ESR spectroscopy provided evidence in support of such a radical pathway. Note, however, that most studies of the reactivity of main group alkyl or hydride reagents with electron-poor heteroaromatics do not include the probing of the possible intermediacy of radicals. Radical-based reactions of electron-poor heteroarenes can be synthetically useful.38 Note also that radicals can display a considerable nucleophilic character so that, for instance, radical addition to pyridines can be mainly directed to the 2- and 4-positions. Maron and coworkers reported in 2010 that bis(η3-allyl)calcium reacts with pyridine to regioselectively afford a species with two anionic, dearomatized 1,4-dihydropyridide ligands, each resulting from the addition of an allyl group to a coordinated pyridine.39 In excess pyridine, an octahedral complex could be isolated and structurally characterized by X-ray diffraction, which contains four intact pyridine ligands, and the two anionic ligands in trans positions. Its reaction with E-Cl electrophiles generates the neutral N-protected 1,4-dihydropyridines and calcium chloride. The mechanism of the dearomatization was studied by NMR and by computational methods. An adduct between the diallylcalcium moiety and pyridine initially forms, followed by allyl migration from calcium to the ortho position of two of the coordinated pyridines to afford the product of double 1,2-insertion. Next, the final double 1,4-insertion product is formed by a rate-determining cope rearrangement step. The Lansbury reagent, formed as the product of the reaction between pyridines with lithium tetrahydroaluminate (lithium alanate), has been employed as a useful reducing agent in organic chemistry for a long time. Hensen and coworkers demonstrated that Lansbury reagent has the

Dearomatization of N-Heterocyclic Ligands

55

[Li(pyridine)4][Al(1,4-dihydropyridide)4] composition and structurally determined its structure and those of similar derivatives.40 Budzelaar et al. studied the reactions of a pyridylbis(imine) pincer ligand with aluminum alkyls. Along with other species, they found products of alkyl addition to the pyridine ring. Radical and ionic mechanisms were considered. Regarding the latter, the authors found the best results in their computational modeling for the alkyl transfer from an anionic R3AlCl species to the pyridine ring of the pincer ligand coordinated to the cationic RAlCl+ fragment.41 In 2011, Okuda and coworkers reported a study of the reactivity of tris(η1-allyl)aluminum complexes toward pyridines. In some cases, they found formation of dearomatized products of the allyl addition to the ortho position of pyridine.42 Previous studies by other groups found either no reaction or just formation of stable pyridine adducts in the reaction between pyridines and other trialkylaluminum reagents. Okuda proposed a Claisen rearrangement in the case of allyl addition to explain the especial behavior of this alkyl group. They encountered a dramatic solvent effect in the reaction. Thus [Al(η1-allyl)3(py)], quantitatively formed as the product of the reaction between [Al(η1-allyl)3(THF)] and pyridine, was found to be stable in pentane for weeks; in contrast, in THF solvent, the dearomatized product of allyl attack to the pyridine ortho position was formed. Allyl attack is prevented by the presence of methyl groups not only at the ortho (steric blocking), but also on the meta or para positions of the pyridine ring, indicating that electron-rich pyridines are not electrophilic enough for undergoing the nucleophilic attack of the allyl group. Accordingly, 4-dimethylaminopyridine does not react, whereas 4-trifluoromethylpyridine affords mainly the product of ortho-allyl addition. Addition of diorganozinc reagents, which are relatively mild nucleophiles, to pyridines have been typically carried out employing transition metal catalysis. In 2015 Hevia and coworkers reported the regioselective addition of lithium arylzincate [LiZnPh3] and [Li2ZnPh4] reagents (prepared by cocomplexation of the appropriate amounts of to LiPh and ZnPh2) to the 9-position of acridine under microwave irradiation.43 They found a much better performance of the heterometallic reagents compared to simple diarylzinc reagents, which has been attributed to a combination of the anionic activation of the organozinc reagent (by forming ZnPh3  or ZnPh4 2 species) and an enhancement of the electrophilicity of the heterocycle by its coordination to the lithium centers. Note the relation with the Budzelaar’s

56

R. Arévalo et al.

Scheme 3 Dearomatization of acridine by reaction with lithium triphenyl zincate.

aluminum system mentioned earlier. Indeed, the authors structurally characterized the dearomatized product of the addition to the 9-position, in which the nitrogen atom is coordinated to LiðTHFÞ3 + , as shown in Scheme 3.

3. REACTION OF PYRIDINES WITH EARLY TRANSITION METAL COMPLEXES Early transition metal chemistry, including that of the f-block elements, shares some features with that of the main group metals due to a high polarity of the element–metal bonds. Early transition metal hydride and alkyl complexes, endowed with strongly electrophilic metal centers, often react with the CH groups in the α-position of N-heterocyclic ligands. For pyridines, such reaction produces a κ2-(C,N)-pyridyl complex, depicted in Fig. 2 (structure 7). Note that the pyridyl ring of this type of complex maintains the aromaticity. These pyridyl complexes display a rich reactivity such as insertions of unsaturated organic molecules (eg, olefins) in the M–C bond, eventually initiating catalytic cycles. Alternatively, if a second molecule of an N-heterocycle binds the same metal center, interligand C–C coupling can take place. For instance, the reaction of an unsaturated alkyl complex with two equivalents of pyridine can generate a bipy ligand on the metal coordination sphere. This chemistry has been excellently reviewed.44,45 Some early transition metal fragments are capable of binding pyridines in the very rare κ2-(C,N) mode. In contrast with the κ2-(C,N)-pyridyl complexes mentioned earlier, pyridine (as opposed to pyridyl) ligands coordinated in the κ2-(C,N) mode are dearomatized and are considered to possess metaloaziridine character (see structure 8 in Fig. 2). Pyridine and its analogs are some of the main contaminants present in petroleum crudes, and they should be removed to avoid the emission to the atmosphere of nitrogen oxides from fuel combustion. To this end,

Dearomatization of N-Heterocyclic Ligands

57

Fig. 2 κ2-(C,N)-pyridyl (7) and κ2-(C,N) pyridine (8) ligands.

large-scale hydrodenitrogenation (HDN) of petroleum crudes is carried out along with hydrodesulfurization (HDS) and hydrodeoxygenation processes. These hydrotreating processes employ sulfidized transition metal oxides as heterogeneous catalysts. Since the sulfur-containing organic pollutants would poison the catalysts of subsequent processes to which the petroleum will be subjected, these processes operate under conditions that have been optimized for HDS. Under these conditions, the HDN process, which is of obvious economic and environmental relevance, is not very efficient. The mechanism of HDN is incompletely understood and it has been hoped that homogeneous models could shed some light on its nature and eventually lead to improvements in the efficiency of the process. Hence, particular attention has been devoted to those few transition metal-based homogeneous systems that have been found to be able to mediate the ring opening of pyridines.46,47 In 1988, Wolczanski and coworkers reported the first example of a metal complex containing a pyridine ligand bonded in the κ2-(C,N) mode.48 The key to obtain this previously unknown coordination mode was the employment of the extremely reactive Ta(silox)3 (silox ¼ OSitBu3) complex, which was prepared by reduction of the previously known, stable complex TaCl2(silox)3 with excess sodium amalgam. Ta(silox)3 is a tricoordinated, highly reducing complex in which the axial coordination of normal twoelectron donors is discouraged by the presence of a metal-centered, twoelectron filled molecular orbital perpendicular to the molecular plane. This feature is of particular relevance for the reaction with pyridine, which is a potential two-electron σ-donor when coordinated in the most common κ1-(N) mode. The extraordinary reactivity of Ta(silox)3 is illustrated by the result of its reaction with the equimolar amount of CO, reported in 1986 by the same group: one could expect simply binding of carbon monoxide to afford a carbonyl complex, since many carbonyl complexes effectively stabilize low oxidation state metal fragments as a result of a significant back bonding. Instead, the reaction, an extreme form of an oxidative addition, breaks down the very strong carbon–oxygen bond of carbon

58

R. Arévalo et al.

monoxide, affording a mixture of O]Ta(silox)3 and (silox)3Ta]C]C] Ta(silox)3, both Ta(V) species with pseudotetrahedral geometries around the metal atoms.49 The presence of strong metal–carbon or metal–nitrogen double bonds in the products must be a significant contribution to the driving force of the reaction. The product of the reaction of Ta(silox)3 with pyridine is a metallaaziridine complex in which the metal-bonded C and N atoms display geometries consistent with sp3 hybridizations. The bond between these C and N atoms is a single bond; therefore, the pyridine has been potentially activated toward the subsequent C–N bond breaking. In 1997, Wolczanski and coworkers reported that the thermolysis of a niobium complex analogous to the tantalum species discussed earlier resulted in the ring opening of the κ2-(C,N)-coordinated pyridine, affording products containing Nb-nitrene and Nb-alkylidene bonds, as shown in Scheme 4.50 In 1987, one year before the first κ2-(C,N)-coordinated pyridine was reported, Cordone and Taube reported the first example of an η2-(C,C)coordinated arene: 2,6-lutidine.51 In this case, the metal fragment was Os(NH3)5, with which Taube, Harman, and coworkers conducted an extensive dearomatization chemistry.52 The pentaammineosmium fragment is strongly reducing; however, η2-coordination of pyridines through a carbon–carbon double bond is only observed when the nitrogen is sterically blocked, as in 2,6-lutidine, or quaternized (pyridinium cations).53 In general, when the aromatic substrate possesses functions capable of acting as two-electron donors, these are the ones that coordinate osmium, a fact attributed to the avoidance of the (destabilizing) disruption of the aromatic system. Wigley and coworkers reported the preparation of a κ2-(C,N)-coordinated pyridine complex and its subsequent ring opening by nucleophilic attack54–56 (Scheme 5). The complex does not result from the reaction of a metal precursor with a free pyridine, but from the reaction of a nitrile with a tantallacyclopentadiene, which in turn is formed as the product of the reduction of a Ta(V) precursor with sodium amalgam in the presence of two equivalents of a terminal acetylene. It must be noted, however, that the same authors obtained related complexes featuring κ2-(C,N)-coordinated quinoline and 6-methylquinoline ligands. The compound

Scheme 4 Pyridine ring opening mediated by a niobium complex reported by Wolczanski et al.

Dearomatization of N-Heterocyclic Ligands

59

Scheme 5 Synthesis of a κ2-(C,N)-coordinated pyridine complex and its ring opening by nucleophilic attack (Wigley et al.).

containing the κ2-(C,N)-coordinated pyridine reacts with nucleophiles such as hydride (its source being superhydride, LiBHEt3), Grignard, or organolithium reagents. The attack of these nucleophiles occurs initially at the metal, where they replace a chloride ligand, but subsequently, the hydride or alkyl group migrates to the metal-bonded carbon atom of the κ2-(C,N)coordinated pyridine, affording a metallacyclic product in which the ring opening of the ligated pyridine has taken place. In this product, the pyridine nitrogen has been incorporated into a metal-bonded nitrene group (see Scheme 5); as in the pyridine ring opening reported by Wolczanski (see earlier), surely the formation of the strong Ta]N bond is a significant contribution to the driving force of the reaction. Mindiola and coworkers, on the other hand, reported a completely different type of pyridine ring-opening reaction.57,58 The reaction of an unsaturated titanium alkylidine intermediate, generated by alkane elimination from an (alkyl)alkylidene complex, with pyridines results in ring-opening metathesis of one of the pyridine C]N bonds across the Ti–C triple bond, yielding the metallaazabicyclic products depicted in Scheme 6. The earlier discussed examples have been taken as a suggestion that the metal-mediated ring opening of pyridines under homogeneous mild conditions would require a complex with a κ2-(C,N)-coordinated pyridine as precursor and, therefore, a very particular type of metal fragments. The vast majority of metal complexes feature pyridines coordinated in the κ1-(N) mode, in which the pyridine ligand remains aromatic; it was though that such species would not be able to mediate pyridine ring opening.46

60

R. Arévalo et al.

Scheme 6 Titanium-mediated pyridine ring opening developed by Mindiola et al.

4. THE ELUSIVE NUCLEOPHILIC ATTACK ON TRANSITION METAL-COORDINATED 2,20 -BIPYRIDINE AND 1,10-PHENANTHROLINE Bipy and phen are some of the most prominent ligands in all areas of coordination chemistry. Because of their combination of good σ-donor and π-acceptor properties, and their tendency to form robust five-member chelate rings, these aromatic diimines form stable complexes with all metals in every oxidation state. In them, with very few exceptions, bipy and phen act as spectator ligands, stabilizing the complex without being themselves the center of the reactivity. Some 40 years ago, while investigating the behavior of cationic bipy and phen transition metal complexes in basic aqueous solution, Gillard and Lyons interpreted certain spectral features as due to species (termed covalent hydrates) formed by hydroxide anion (or water) addition to the diimine ligands.59 Gillard proposed that coordination to positively charged, Lewis-acidic metal centers activated the pyridine groups toward nucleophilic attack, and likened these complexes to the pyridinium cations produced by nitrogen quaternization.60,61 This proposal sparked a longstanding controversy,62–66 and work by other groups satisfactorily explained the evidence presented by Gillard in terms of hydroxide attack on the metal rather than on the bipy or phen ligands, thus discrediting Gillard’s hypothesis. In particular, Lay pointed out that, unlike alkyl cations, a transition metal fragment such as the ones Gillard was considering (which were from mid to late transition metals with relatively high electron counts) would be able to, besides withdrawing electron density via σ, release electron density into the empty π orbitals of the heterocycle,67 and Blackman and coworkers showed that formation of pentacoordinated complexes by hydroxide attack to [Pt(bipy)2]2+ complexes explained the early observations that prompted Gillard to propose attack to bipy.68 In 2002, Zhang et al. demonstrated the hydroxide attack on bipy and phen in the presence of cupric ions, suggesting that chemistry like that proposed by Gillard can be operative in certain

Dearomatization of N-Heterocyclic Ligands

61

systems.69 However, the fact that these reactions have needed to be carried out under hydrothermal conditions can be taken as an indication of the difficulty of functionalizing the metal-bonded pyridine rings, a transformation that involves their dearomatization. The subsequent isolation of an intermediate in these hydrothermal reactions containing both phen and OH ligands coordinated to the same metal atom, led Latham et al. to suggest initial attack by OH to the metal, followed by a metal to phen migration.70 A perusal of the literature indicated that not only the attack of the hydroxide anion, the one of the original Gillard’s proposal, to transition metal-bonded pyridines, but in fact any nucleophilic attack to these ligands, is extremely rare. A review of the chemistry carried out by Gillard and coworkers as well as by others following Gillard’s seminal suggestions have been recently published by one of the major players in the field.71

5. REACTIVITY OF CARBONYL RHENIUM AND MOLYBDENUM COMPLEXES 5.1 Reactions of the Phosphido Complex [Re(PPh2) (CO)3(phen)] with Activated Acetylenes and Olefins Pe´rez et al. studied the reactivity toward electrophiles of complexes containing basic ligands such as alkoxo,72,73 hydroxo,74–76 amido,77–81 alkylidenamido,82,83 etc. Alkoxo complexes [Re(OR)(CO)3(bipy)] reacted with dimethylacetylendicarboxylate (DMAD) to afford the product of formal insertion of DMAD into the Re–O bond, presumably through nucleophilic attack of the coordinated alkoxo to one of the acetylenic carbons.72 This would develop a negative charge at the other acetylenic carbon, which then would attack the metal, from which the oxygen would be displaced, as shown in Scheme 7A. This displacement would be aided by the positive charge developed on the metal-bonded oxygen, which would make it a good leaving group. For the reactions of DMAD with hydroxo or arylamido complexes, the target of the intramolecular nucleophilic attack is not the metal, but one of the proximal CO ligands.76,80 H migration from the hydroxo or arylamido group to the oxygen of the same carbonyl group helps to preserve the Re–O or Re–N bonds and to transform the CO into a hydroxycarbene group, which is incorporated into a five-member metallacycles, as shown in Scheme 7B. The same group synthesized the phosphido (phosphanido) complex [Re(PPh2)(CO)3(phen)] and carried out preliminary studies of its reactivity.77 In contrast with the reactions of alkoxo, hydroxo, or arylamido

62

R. Arévalo et al.

Scheme 7 Reactions of DMAD with alkoxo (A) and hydroxo or amido (B) complexes.

complexes discussed earlier, the reactions of the phosphido complex with either methyl propiolate or methyl acrylate afford products consistent with conjugated addition of the phosphido ligand to the acetylene or olefin to afford a zwitterionic adduct84 (reminiscent of those involved in several types of phosphane-catalyzed coupling reactions),85 followed by intramolecular attack of the other carbon to one of the ortho carbons of the phen ligand (Scheme 8). Evidence of attack to the phen ligand is found both in solution, where the NMR indicates a nonsymmetric phen, and in the solid state. The structure of the product, determined by X-ray diffraction, confirms the formation of a carbon–carbon bond and the nonplanarity of the phen moiety. The comparison of the bonding distances N(1)–C(11) ¼ 1.452(11), C(11)–C ˚ in the product with those (12) ¼ 1.540(14), and C(12)–C(13) ¼ 1.307(10) A ˚ , respecin the phosphido precursor (1.329(11), 1.387(15), and 1.354(17) A tively) indicates the dearomatization of the pyridine ring as a result of the intramolecular nucleophilic attack. The reaction of the phosphido complex with DMAD proceeds similarly, affording complex 9.86 However, whereas the product of the reaction with methyl propiolate has been found to be indefinitely stable in solution, the product of the analogous reaction with DMAD slowly evolves to its tautomer 10 as shown in Scheme 9. Compounds 9 and 10 have been characterized by IR, NMR, and X-ray diffraction. The transformation of 9 into 10 is at least partly driven by an increase in the conjugation. The reaction of 9

Dearomatization of N-Heterocyclic Ligands

63

Scheme 8 Reaction of [Re(PPh2)(CO)3(phen)] with methyl propiolate.

Scheme 9 Transformation of complex 9 into its tautomer 10.

with nBuLi (which, monitored by IR, showed the formation of a species with low νCO bands, consistent with deprotonation to form an anionic complex) followed by treatment with HOTf, instantaneously affords 10, suggesting that the hydrogen migrates as a proton, rather than as a hydrogen atom or as a hydride. The most striking feature of the reactions of the phosphido complex with the activated olefin and acetylenes is the rapid dearomatization of the phen ligand under very mild conditions. It seems likely that the reaction is somewhat facilitated by its intramolecular nature. Also noteworthy is the stability of the dearomatized products, which have made possible to discover the tautomerization discussed earlier and to crystallize the products by solvent slow diffusion. Compared with the alkaline salts resulting from the addition of, for instance, RLi reagents to pyridines, the dearomatized products of the reactions of the phosphido complex discussed earlier are neutral complexes, without the possibility of eliminating an alkaline hydride (which, having a large lattice energy, must contribute considerably to the driving force of the reaction). Additionally, the presence of the metallacycle hampers the regaining of planarity that would be required for rearomatization. It must be noted that the product of the attack to phen is the only product of the reaction, and no product of attack to one of the two proximal CO ligands

64

R. Arévalo et al.

was detected. This stands in contrast with the previously reported reactions of terminal phosphido complexes with DMAD, where formation of fivemember metallacycles resulted from the coupling of the phosphido ligand, DMAD, and one CO ligand.87–89 Note that the results in the reactions of the hydroxo and amido complexes with DMAD discussed earlier indicate that the proximal carbonyl ligands are susceptible to similar coupling reactions to form five-member metallacycles in pseudooctahedral rhenium tricarbonyl bipy complexes. Therefore, the unprecedented observed coupling involving the coordinated phenanthroline is not only viable under very mild conditions, but also preferred over the coupling involving one of the CO ligands.

5.2 Dearomatization by Deprotonation of Complexes with N-Alkylimidazole Ligands In 2008, Huertos et al. reported that the reaction of [Re(N-MesIm) (CO)3(bipy)]OTf (11) (N-MesIm ¼ N-mesitylimidazole; mesityl ¼ 2,4, 6-trimethylphenyl) with the strong base KN(SiMe3)2 affords a product consistent with deprotonation of the imidazole C2–H group, followed by attack of that carbon to the ortho (C6) carbon of the bipyridine ligand (complex 12, Scheme 10A).90 This reaction stands in contrast with the transformation of

Scheme 10 Imidazole-bipy C–C coupling (A) or transformation of N-RIm to NH–NHC complexes (B).

Dearomatization of N-Heterocyclic Ligands

65

an N-alkylimidazole (N-RIm) into an anionic, C-bonded imidazolyl ligand (trapped by protonation to afford a stable cationic NH,NHC complex) by deprotonation of a [Mn(N-RIm)(CO)3(bipy)]+ complex reported by Ruiz and Perandones in 2007 (Scheme 10B).91 Note that imidazoles have been widely employed as ligands in transition metal chemistry,2 and that metallation of free N-alkylimidazoles at C2 was a known reaction.92,93 However, the deprotonation of metal-coordinated N-alkylimidazoles has been virtually unexplored before. As in the reactions of the phosphido complex discussed earlier, the C–C coupling reaction resulting from the deprotonation of [Re(N-MesIm) (CO)3(bipy)]OTf affords a product in which one of the pyridine rings of the bipyridine ligands is dearomatized, and it takes place under very mild conditions. Reactions similar to the deprotonation of [Re(N-MesIm)(CO)3(bipy)] OTf were carried out with its N-methylimidazole (N-MeIm) analog as well as with the two similar compounds containing phen instead of bipy. The stability of the products varied greatly. The neutral product of [Re(NMesIm)(CO)3(bipy)]OTf deprotonation was sufficiently stable to permit isolation and full characterization, including X-ray crystallography. In solution, the dearomatization of the attacked ring is indicated by the upfield shift of several of the signals corresponding to its CH groups. The solid-state structure of this species confirms the formation of the new C–C bond between the central carbon (C2) of the imidazole ring and the C6 carbon atom of the bipyridine ligand, and the dearomatization of the attacked pyridine ring, geometrically shown by its loss of planarity (see Fig. 3). Unlike in the C–C coupling reactions of the phosphido complex discussed earlier, where six-member metallacycles were formed, now the product of the imidazole-bipy coupling features a five-member ring. As a result of the C–C forming reaction, the aromatic imine functionality in the reactant is transformed into an amido group in the product. The combination of amido ligands, with a high tendency to act as π-donors, with high oxidation state metal fragments possessing empty orbitals of appropriate symmetry, usually leads to very stable complexes.94 Although in principle, for an octahedral d6 complex, with the π-symmetric t2g orbitals completely filled, π-donation from the amido ligand should not be possible, the electron density from the amido lone pair can be donated to the strongly π-acceptor CO ligands via the π-symmetric metal orbitals.95 Such donation, which would stabilize the complex, would lead to a planar amido group. A planar at nitrogen geometry, characterized by a sum of the

66

R. Arévalo et al.

Fig. 3 Molecular structure of complex 12.

angles about the nitrogen atom being 360 degree, has been encountered in most structurally characterized amido complexes, including in arylamido rhenium tricarbonyl complexes, in which structural evidence supports the delocalization of the amido lone pair through the aryl substituents.77 In contrast, complex 12, the product of the deprotonation of [Re(N-MesIm) (CO)3(bipy)]OTf, which is a rare example of a stable amido complex of a low valent metal without aryl substituents at the amido nitrogen, features a pyramidal amido group, the sum of the angles around the amido nitrogen being 335 degree. This feature can be attributed to the constrained geometry resulting from the amido group being part of a rigid tridentate ligand. Besides of the structural evidence—the geometries about both the carbon and the nitrogen are consistent with sp3 hybridization in the product— this amido character is reflected in the reactivity. Thus, the neutral products resulting from the deprotonation of the [Re(N-RIm)(CO)3(N–N)]OTf (R ¼ Me, Mes; N–N ¼ bipy, phen) compounds react with MeOTf. For three of the compounds, the reactions afford stable, cationic complexes featuring amino groups resulting from methylation at nitrogen (Scheme 11). One of these derivatives (N-MesIm, phen) was fully characterized, including X-ray diffraction. Comparing the two series of compounds, namely, the neutral amido products with the cationic complexes that result from the

Dearomatization of N-Heterocyclic Ligands

67

Scheme 11 Methylation of the amido group of the dearomatized C–C coupled products affording stable amino complexes or a ring-opening product (13).

methylation of the amido compounds, the latter have been found to be significantly more stable. This is attributed to the high nucleophilic character of the amido ligand, a fact that has been previously noted in monodentate amido rhenium tricarbonyl complexes.77,78,80 In the cationic amino complexes resulting from at nitrogen methylation of the amido complexes, both NMR and X-ray data indicate the presence of a dearomatized pyridyl ring as in the amido precursor. The neutral species formed in the deprotonation of [Re(N-MeIm)(CO)3(bipy)]OTf reacts with excess methyl triflate under very mild conditions (15 min stirring at room temperature) affording compound 13 (see Scheme 11 and Fig. 4), which was characterized spectroscopically and by X-ray diffraction. The most interesting feature in the structure of 13 is that it reveals, besides the double methylation of its nitrogen atom, that nitrogen extrusion occurred in the pyridine ring to which the nucleophile attack took place. The product features a tridentate ligand with pyridine, imidazole, and dimethylamino donor

68

R. Arévalo et al.

Fig. 4 Molecular structure of the cation of compound 13.

groups, and a cyclopentadienyl bridgehead group formed as the product of the mentioned nitrogen extrusion from the attacked pyridine ring. The treatment of the neutral product of the initial deprotonation with the stoichiometric amount of MeOTf failed to yield a monomethylated product akin to the one resulting from methylation of 12; however, taken collectively, the results outlined earlier suggest that ring opening occurs via a second methylation of such a product. The fact that the monomethylated phen complex does not react with excess MeOTf suggest that the additional aromatic ring in the phen ligand somewhat hinders the rearrangement leading to nitrogen extrusion. The overall transformation that starts with the deprotonation of the cationic N-methylimidazole bipyridine complex and ends in the pyridine ring opening shows that pyridine ring opening under mild conditions can be achieved starting from a very stable metal complex in which the bipy ligand, usually very inert, is coordinated in its normal mode, ie, both pyridyl rings bind the metal only through their nitrogen atoms, in contrast with previous examples of homogeneous metal-mediated pyridine ring opening, which necessitated highly reactive early transition metal fragments capable of binding pyridines in the κ2-(C,N) mode.

5.3 Couplings and Dearomatization Initiated by Deprotonation of C(sp3)–H Bonds The C–C coupling reactions discussed earlier have been initiated by the deprotonation of the C(sp2)–H bond of an N-alkylimidazole ligand.

Dearomatization of N-Heterocyclic Ligands

69

It has been found that similar C–C couplings can be carried out through the deprotonation of aliphatic CH groups such as those of coordinated dimethylsulfide96 or trimethylphosphane,97 which triggers the C–C coupling between the resulting methylene group and one of the pyridyl groups of bipy and phen coligands. The product of the reaction of [Re(CO)3(bipy)(S(CH3)2)]OTf with a slight excess of KN(SiMe3)2 showed insufficient thermal stability to allow crystallization and thus was fully characterized only in solution by means of NMR. To aid in this characterization, the labeled analog [Re(CO)3(bipy)(S(13CH3)2)]OTf was similarly synthesized and deprotonated. The results of the NMR studies indicated that the methylene group resulting from the deprotonation of one of the methyl groups of S(CH3)2 attacked the C2-position of bipy, in contrast with the attack at C6 found in the C–C coupling resulting from N-alkylimidazole deprotonation.90 The product has been found to be the mixture of the diasteromers showed in Scheme 12. The reaction of the mixture of diasteromers with the equimolar amount of trimethylphosphane afforded a single compound, resulting from the displacement (from the rhenium coordination sphere) of the sulfide donor group by the phosphane (Scheme 12). In an analogous fashion, the diasteromeric mixtures reacted with bis(dimethylphosphino)methane (dmpm), affording a single product in which the sulfide group has been replaced by a monodentate dmpm. No product containing a bidentate dmpm ligand could be detected, despite the high nucleophilicity of this diphosphane, showing that the low stability of the diasteromeric mixture resulting from the deprotonation and C–C coupling does not arise from lability of the ligands (eg, decarbonylation).

Scheme 12 C–C coupling and pyridine dearomatization initiated by the deprotonation of a dimethylsulfide ligand.

70

R. Arévalo et al.

This low stability was thus attributed to the dearomatization of one of the pyridyl rings, and thus a similar C–C coupling was sought which would involve a nonaromatic diimine in the hope that it would afford a more stable product. Thus, compound [Re(CO)3(2,6-iPr-BIAN)(S(CH3)2)]BAr0 4 (2,6-iPr-BIAN ¼ 1,2-bis[(2,6-diisopropylphenyl)imino]acenaphthene) was prepared and characterized both spectroscopically and by X-ray diffraction, and it was allowed to react with KN(SiMe3)2. Indeed, the stable, single product of the reaction was found to be the neutral complex resulting from the coupling between the methylene group of an H3CSCH2 fragment, produced in the deprotonation of the coordinated dimethylsulfide, and one of the imine carbons of the diimine chelate. A factor that can make the BIANderived amido complexes more stable compared with their bipy counterparts is the steric protection lent by the bulky aryl group in the former. The deprotonation of [Re(CO)3(bipy)(P(CH3)3)]OTf with KN(SiMe3)2 afforded a mixture of two products (Scheme 13). One of them was the neutral cyano complex [Re(CO)2(CN)(bipy)(P(CH3)3)], formed from attack of the amide on one of the CO ligands, a known reaction that involves the elimination of hexamethyldisiloxane, O(SiMe3)2.98 The second, minor product, was the expected result of the deprotonation of one of the phosphane methyl groups and subsequent intramolecular attack on the bipy C6-position. Note the difference with the similar reaction of the dimethylsulfide complex discussed earlier, in which the attack was to the bipy C2-position instead.96 As in the reaction of the sulfide complex just described, the product of C–C coupling was found to be too unstable for isolation, but could be characterized by NMR at low temperature. Stable 2,6-iPr-BIAN analogs were isolated and characterized, including by X-ray diffraction, for P(CH3)3, P(CH3)2Ph, and P(CH3)Ph2 complexes (see Fig. 5 for the P(CH3)3 derivative). The deprotonation has been found to take place exclusively at the methyl—not at the phenyl—substituents. The amido groups in the C–C

Scheme 13 Formation of cyano and dearomatized C–C coupled products initiated by deprotonation of a trimethylphosphane ligand.

Dearomatization of N-Heterocyclic Ligands

71

Fig. 5 Molecular structure of the BIAN C–C coupled product obtained by deprotonation of a trimethylphosphane ligand.

coupling products were found to be planar at nitrogen, pointing to amido lone pair delocalization. At first sight such delocalization could be expected to occur onto the aryl amido substituents. However, coplanarity would be the optimal orientation of the aryl group with respect to the amido ligand in order to maximize the overlap between π-orbitals and allows the electronic delocalization onto the aryl group, and actually the amido group was found to be very far from coplanar with the aryl substituent (the dihedral angle between amido and aryl planes has been found to be 79 degree for the trimethylphosphane derivative). Presumably, this large deviation from coplanarity is due to the steric hindrance imposed by the bulky isopropyl groups, as it is typical of BIAN complexes. Therefore, the planarity at the amido nitrogen in the sulfide and phosphane derivatives results at least in part from π-donation from the amido group to the strongly π-acceptors CO ligands via the metal π-orbitals.95

5.4 Coupling Between Two Monodentate Heterocyclic Ligands Most small molecules, and certainly most metal-complexing ligands, are constituted by a few building blocks; therefore, a modular synthesis employing cross-coupling reactions between appropriate precursors could serve as a general synthetic approach.99 However, those appropriate precursors must be reactive enough, so conventional cross-coupling procedures require first the preparation of functionalized versions of the individual building blocks. For instance, in Pd-catalyzed C–C coupling schemes, one of the partners must be an electron-rich compound, such as a main group organometallic, organoborane, etc., while the other must be an

72

R. Arévalo et al.

Scheme 14 General scheme of most late-metal catalyzed C–C coupling reactions.

electrophile possessing a good leaving group, such a triflate or iodide (Scheme 14). That adds further complication to the whole synthesis and generates unwanted by-products. Some subunits, notably the 2-pyridyl group, are particularly problematic.99 Pyridyl groups are prevalent motifs in several types on natural and unnatural products, and in particular in polydentate ligands. Several families of coordination compounds, and in particular rhenium tricarbonyl complexes, which have been the focus of most of our work, are potentially useful due to their catalytic, luminescent, or anticancer properties.100,101 The several examples of inter-ligand C–C coupling within the metal coordination sphere discussed earlier suggested to us the possibility of employing similar schemes to assemble monodentate ligands, including pyridines, without the need of any previous functionalization other than metal coordination. The synthesis of polydentate ligands, especially nonsymmetric ones, is often a challenging task. In currently employed methodologies, the free ligand is first synthesized and purified employing conventional organic synthesis procedures, which are frequently tedious, multistep methodologies affording a low overall yield and require separation of by-products. Then, the ligand is reacted with an appropriate metal source to afford the targeted metal complex. An alternative scheme can be envisaged in which small, nonfunctionalized building blocks, like pyridines, imidazoles, etc., are first coordinated to a metal center by simple substitution reactions. Preassembly of those subunits would be taken care of by the geometric preferences of the metal fragment. For instance, the strong preference for a fac disposition of the three CO ligands in octahedral metal complexes ensures that any two of the remaining coordination sites are mutually cis and thus optimally placed for a C–C coupling. Additional advantages of the metal coordination are an enhanced acidity of the CH groups to be deprotonated and an enhanced reactivity of the electrophilic counterpart, both leading to more reactive systems and thus to milder conditions, which in turn would provide more selective reactions. Rhenium tricarbonyl complexes containing two N-alkyimidazole ligands and one pyridine were synthesized by the reaction of [Re(CO)3(N-RIm)2(OTf )] (which in turn is easily prepared from the reaction of [Re(CO)5(OTf )] and N-alkylimidazoles) with either pyridine or 4-picoline and the equimolar amount of NaBAr0 4.102,103 The new

Dearomatization of N-Heterocyclic Ligands

73

precursors [Re(CO)3(N-RIm)2(py)]BAr0 4 (R ¼ Me, Mes) are readily prepared in a half-gram scale, are relatively stable and can be kept for several weeks in Schlenk flasks under a nitrogen atmosphere. A key for the success of the reactions discussed later is that in solution, [Re(CO)3(N-RIm)2(py)]+ complexes—and, in general, rhenium(I) tricarbonyl octahedral complexes—do not undergo ligand scrambling. Their reactions with KN(SiMe3)2 in THF at 78°C immediately afford solutions with νCO bands at lower wavenumber values, consistent with deprotonation. The addition of silver triflate afforded cationic products (Scheme 15). The solid-state structures of two of these products were crystallographically determined, confirming the presence of a fac-Re(CO)3 fragment bonded to an intact N-alkylimidazole ligand and to a bidentate ligand formed by the coupling between the C2-positions of an N-alkylimidazole ligand and the C2 carbon of pyridine (Fig. 6). This is consistent with the deprotonation of the C2–H group of the N-alkylimidazole, followed by the intramolecular nucleophilic attack of the deprotonated group on the C2 carbon (the most electrophilic position) of the monodentate pyridine. The resulting bidentate ligand initially formed as the product of the C–C coupling would feature a dearomatized pyridyl ring, like those formed in the nucleophilic attacks to the pyridine rings of the bidentate bipy or phen ligands discussed earlier. Subsequently, that ring would be rearomatized by reaction with the one-electron oxidant silver triflate. Such rearomatizations could not be carried out with the products of coupling between a deprotonated N-alkylimidazole and a bidentate bipy or phen ligand discussed earlier, the likely reason being that the rigid fac-tridentate ligand formed as the product of the C–C coupling cannot become planar, as it would be required for rearomatization.

Scheme 15 Imidazole–pyridine cross coupling at a rhenium center.

74

R. Arévalo et al.

Fig. 6 Molecular structure of a pyridylimidazole complex obtained by C–C coupling of imidazole and pyridine ligands.

The fact that two equivalents of silver triflate are needed suggests the participation of two one-electron oxidation events, in agreement with formal hydride removal. In the absence of mechanistic studies, it is proposed that the neutral, dearomatized product of C–C coupling is oxidized by one equivalent of silver triflate to afford a radical cation, a species that would be deprotonated to afford a neutral radical; this would then be oxidized by the second equivalent of AgOTf yielding the observed cationic complex. Since only one equivalent of KN(SiMe3)2 was employed, the putative radical cation would be deprotonated by hexamethyldisilazane, HN(SiMe3)2, which would have been formed as the product of the initial deprotonation by KN(SiMe3)2. The course of the reaction has been found to be the same for the N-methylimidazole and N-mesitylimidazole ligands. No other organometallic compounds could be detected as reaction products. Note that free pyridylimidazoles, which are often employed as ligands, are usually prepared by de novo construction of the pyrazole ring. Compounds containing [Re(CO)3(N-RIm)(dmap)2]BAr0 4 (R ¼ Me, Mes, dmap ¼ 4-dimethylaminopyridine) cationic complexes were subjected to the same deprotonation/oxidation sequence, affording single products from coupling between the C2-positions of the N-alkylimidazole and the pyridine (Fig. 7). These reactions selectively afford the products of crosscoupling between one imidazole and one pyridine, without homocoupling (complexes containing biimidazole or bipyridine ligands) products, regardless of the particular

Dearomatization of N-Heterocyclic Ligands

75

Fig. 7 Solid-state structure of the complex obtained by the cross coupling of dmap and N-MesIm starting from [Re(CO)3(N-MesIm)(dmap)2]BAr0 4.

composition of the cationic precursors: two imidazoles and one pyridine, or two pyridines and one imidazole. This is attributed to the higher acidity of the C2–H group of the N-alkylimidazole ligand and the higher electrophilic character of the pyridine ligand. Note that even the relatively electron-rich dmap ligand is electrophilic enough to undergo the C–C coupling reaction. The presence of an intact N-alkylimidazole ligand in the cationic complexes formed in the reactions of pyridine–imidazole coupling and rearomatization suggested to us the possibility of its deprotonation. The internal nucleophile presumably resulting from deprotonation of the C2-position of the monodentate N-RIm ligand could attack either the pyridine or the imidazole ring of the nonsymmetric bidentate ligand. Thus, the reaction of tricarbonyl rhenium (imidazole)(pyridylimidazole) compounds with KN(SiMe3)2 afforded neutral (IR) products too unstable for isolation and even for NMR characterization. These species were in situ protonated with HOTf, affording stable products, which could be fully characterized (Scheme 16). The determination of its solid-state structure by X-ray diffraction established the protonation at the nitrogen of the dearomatized ring, and the presence of a fac-capping tridentate ligand resulting from the coupling between the C2-position of the N-RIm ligand and the ortho position of the pyridyl ring of the bidentate ligand. The metrical data of the pyridyl ring indicate its dearomatization, in agreement with the NMR solution data. When an analogous reaction was conducted starting with the precursor containing an N-MesIm monodentate ligand, the product of its deprotonation was not stable enough for isolation, but could be characterized in solution,

Scheme 16 C–C coupling between a monodentate N-RIm ligand and the pyridine ring of a pyridylimidazole chelate.

Dearomatization of N-Heterocyclic Ligands

77

and its spectroscopic data suggested deprotonation of the monodentate N-RIm ligand and attack to the 2-pyridyl ring, leading to its dearomatization. The reaction of this neutral species with triflic acid afforded a stable product (which was also crystallized and fully characterized) resulting from protonation at nitrogen similar to the one previously discussed. These results indicate that the intramolecular attack by the deprotonated N-alkylimidazole takes place selectively on the pyridine ring, as no product of attack to the imidazole ring could be detected. Note that the full sequence of reactions leading to these products with a tridentate ligand from the starting complexes containing three monodentate N-heteroaromatic ligands (two N-alkylimidazoles and one pyridine) involve two consecutive C–C coupling events between the C2 carbons of the two N-alkylimidazoles and the two ortho positions of the pyridine ring. The C–C coupling methodology outlined earlier was extended (in spite of the lower acidity of pyridines compared with N-alkylimidazoles) to the coupling between two pyridine ligands to afford complexes containing a bipy ligand. Thus, when [Re(CO)3(dmap)3]OTf (which was readily prepared from [Re(OTf )(CO)5] and dmap) was treated with KN(SiMe3)2 and then with AgOTf, the single organometallic product was found to be [Re(CO)3(2,20 -bipy-4,40 -NMe2)(dmap)]OTf, which was fully characterized, including X-ray diffraction. Compounds [Re(CO)3(py-R)3]OTf (py-R ¼ pyridine or 4-methoxypyridine), prepared by the reaction of the appropriate pyridine with the labile [Re(CO)3(DMSO)3]OTf compound, afforded [Re(CO)3(bipy-R)(OTf )] products resulting from the coupling of two of the coordinated pyridines to afford a bipy ligand when treated first with KN(SiMe3)2 and then with HOTf (which has been found to be a cleaner oxidant in this case than AgOTf ). The product containing the parent bipy ligand, a known complex, was identified by comparison (IR and NMR) with a authentic sample, and the product of C–C coupling between two 4-methoxypyridine ligands was fully characterized, including X-ray diffraction. These results demonstrated that deprotonation of a pyridine ligand coordinated in the by far most commonly encountered κ1-(N) mode is feasible. This methodology for inter-pyridine coupling could be successfully extended to two different pyridine ligands, affording complexes with nonsymmetric bipy ligands. Thus, [Re(CO)3(dmap)2(py)]BAr0 4 and [Re(CO)3(dmap)2(py-4-OMe)]BAr0 4 were prepared from addition of either pyridine or 4-methoxypyridine to [Re(CO)3(dmap)2(OTf )] in the presence of the equimolar amount of NaBAr0 4. The deprotonation of these

78

R. Arévalo et al.

mixed pyridine compounds with KN(SiMe3)2 followed by oxidation with AgOTf afforded [Re(CO)3(dmap)(N-N0 )]BAr0 4 compound featuring nonsymmetric bipys resulting from cross coupling between different pyridine ligands (Scheme 17). As in the formation of compounds with pyridylimidazole complexes (see earlier), no formation of symmetric, homocoupled (from coupling of two equal pyridines) could be spectroscopically detected. The formation of bipyridines by C–C coupling between two pyridyl groups on the coordination sphere of group III complexes has been reported by Diaconescu et al.104,105 A difference with the rhenium tricarbonyl system is that whereas Diaonescu starts with the metallation of one pyridine to afford a κ2(C,N) pyridyl ligand (see Scheme 18), two neutral pyridine ligands are first coordinated to rhenium in the κ1(N) normal coordination mode, and then the resulting stable cationic complex was allowed to react with an external base.

Scheme 17 Synthesis of a nonsymmetric bipy ligand by C–C coupling followed by oxidation.

Scheme 18 Coupling between two pyridine ligands on a scandium complex by Diaconescu et al.

Dearomatization of N-Heterocyclic Ligands

79

5.5 Ring Opening of Imidazole Rings The study of the deprotonation of coordinated N-akylimidazoles was extended to cationic fac-Re(CO)3 complexes containing three such ligands.106,107 This reaction produces neutral imidazolyl complexes as it will be discussed in Section 6.2 for [Re(CO)3(N-MeIm)3]OTf. In contrast, if at least one of the imidazole ligands is an N-arylimidazole, neutral complexes resulting from intramolecular attack of the deprotonated N-RIm ligand to the C2 carbon of one of the other N-RIm ligands, which as a consequence undergoes ring opening, are produced (see Scheme 19). In the products, the 1,3-diaza fragment resulting from imidazole ring-opening display a cisoid geometry as a result of an intramolecular hydrogen bond, shown in dashed line in Scheme 19. Protonation or methylation of these products occurs at the noncoordinated nitrogen of the ring opened moiety, affording transoid products. Carver and Diaconescu have reported the ring opening of N-methylimidazole upon reaction of scandium or yttrium benzyl complexes with three equivalents of MeIm for 5 h at 70°C (Scheme 20).108 The authors propose that the reaction occurs via C–H activation of one of the imidazoles (to afford an imidazole–imidazolyl intermediate which could be characterized for an yttrium analog). Electrophilic addition at the imine-type nitrogen to generate an imidazolium cation, followed by the attack of a nucleophile (usually hydroxide anion) at C2 has been proposed as the mechanism for a number of imidazole cleavage processes, from the venerable Bamberger reaction109 to the recent tandem ring opening of imidazoles with electron-deficient acetylenes and water.110 Of these reactions, particularly mild conditions (45–60°C) have been reported for the latter. N-alkylimidazoles metalated (eg, lithiated) at C2 have been found to display considerable stability toward ring opening in comparison with other metalated N-heterocyles, allowing their employment in reactions with organic electrophiles.93 Is somewhat surprising that, although N-alkylimidazoles have been widely employed as ligands, deprotonation of metal-coordinated N-alkylimidazoles, and the reaction of metalated N-alkylimidazoles with transition metal complexes remain very little explored.111

5.6 Couplings Between N-Alkylimidazoles and Imine Ligands In contrast with the reactivity displayed by the rhenium complexes discussed earlier, the deprotonation of [Mo(η3-methallyl)(CO)2(bipy)(N-RIm)]+

Scheme 19 Imidazole C–C coupling and ring opening at Re(I) complexes.

Dearomatization of N-Heterocyclic Ligands

81

Scheme 20 Proposed mechanism for scandium-mediated imidazole coupling and ring opening by Diaconescu et al.

complexes yielded the neutral imidazolyl complexes, discussed in Section 6.2, and no product of C–C coupling could be detected. Density functional theory (DFT) calculations, however, indicated that the energy differences between the pathway leading to the observed imidazolyl product and that leading to attack on bipy is only 3.5 kcal mol1.112 That led the authors to synthesize several compounds of the type [Mo(η3-methallyl) (CO)2(pyridylimino)(MeIm)]OTf and study their deprotonation with KN(SiMe3)2, expecting that the nonaromatic imine function would be more reactive than bipy and hence could be the target of the deprotonated imidazole.113 Indeed, attack by the deprotonated imidazole C2 carbon to the nonaromatic imine carbon, as shown in Scheme 21 has been found to be the sole product.

82

R. Arévalo et al.

Scheme 21 Imidazole–imine C–C coupling and protonation of the resulting amido group to afford cationic amino complexes.

DFT calculations showed that the products of C–C coupling were the favored ones both kinetically (they were formed in a process without computable barrier) and thermodynamically. Some of the neutral C–C products were found to be stable enough to permit isolations; for those too unstable, protonation at the nitrogen originating from the nonaromatic imine with HOTf yielded stable cationic complexes that were isolated as triflate salts and could be fully characterized. Since allylmolbdenum dicarbonyl complexes are not as substitutionally inert as the Re(I) complexes discussed earlier, possible intermolecular pathways involving the transfer of the monodentate imidazole (or deprotonated imidazole) ligands between metal centers were considered as a possibility (note that imidazolate bridges have been found by Braunstein and coworkers).111 They, however, were ruled out by a crossover experiment in which a mixture of two compounds having different pyridylimine and N-alkylimidazole ligands was treated with KN(SiMe3)2 afforded only the products of intramolecular C–C coupling (Scheme 22).

5.7 Coupling Between N-Alkylimidazoles and Nitriles or Isonitriles Nucleophilic attacks on the sp-hybridized carbon atom of nitrile and isonitrile ligands are well known, so in principle the coupling between these unsaturated ligands and C2-deprotonated N-alkylimidazole coligands could offer a potential pathway to access new anisobidentate chelates within the coordination sphere of a metal.114 Two possible difficulties were considered, one for each type of monodentate unsaturated ligand: nitriles are labile; therefore, attack to the metal by either the base or the deprotonated N-RIm ligand could compete with attack at the coordinated nitrile; as for isonitriles, the planned attack at the metal-bonded carbon atom would afford a four-member ring, perhaps of limited stability; note that such a

Dearomatization of N-Heterocyclic Ligands

83

Scheme 22 Crossover experiment showing the intramolecular nature of the N-RImimine coupling reactions.

product would be comparable (carbonyl and isonitrile ligands are isoelectronic) to the computationally encountered (but not stable enough for spectroscopic detection) intermediate in the reaction leading to imidazolyl complexes, which is formed via C–C coupling between a CO and the C2-deprotonated N-RIm ligand. Nitrile and isonitrile compounds were prepared by the reaction of complexes [Re(CO)3(N-RIm)2(OTf )] with pivalonitrile or tert-butylisocyanide and the salt NaBAr0 4 in dichloromethane.115 Employing the latter as a triflate abstractor is based on the low solubility of sodium triflate in dichloromethane and is needed to achieve the displacement of triflate from the cationic complex by the poor donor nitrile, or by the low-nucleophilic isonitrile. The [Re(CO)3(N-RIm)2(tBuCN)]BAr0 4 and [Re(CO)3(N-RIm)2 (tBuNC)]BAr0 4 compounds, which could be isolated and characterized, instantaneously reacted with KN(SiMe3)2 in THF at low temperature to afford, as indicated by the change in the color and the IR spectrum of the solution. For the nitrile complexes, the limited stability of the neutral (as judged by their νCO IR bands) products precluded their isolation and even in situ NMR characterization. Reaction of these neutral species with MeOTf or HOTf afforded cationic complexes, which were isolated and fully characterized (see Scheme 23). Their NMR data are in agreement with the initial deprotonation of the N-alkylimidazole ligand at its C2–H position and subsequent attack of the “internal nucleophile” to the α carbon of the

84

R. Arévalo et al.

Scheme 23 Rhenium-mediated couplings between imidazole and nitrile (top) or isonitrile (bottom) ligands.

nitrile. The result of the C–C coupling reaction would be the formation a five-member chelate ring featuring an alkylideneamido (N-metaloimine,82 ketimide116) donor group. A rhenium tricarbonyl complex has been found to be one of the very few highly reactive—yet isolable—alkylideneamido complexes,82 in contrast with the chemical inertness displayed by these ligands when coordinated to most other fragments.116 A similar high reactivity may be the origin of the instability of the products of the deprotonation of [Re(CO)3(N-RIm)2(tBuCN)]BAr0 4 (Scheme 23). Methylation of the alkylideneamido nitrogen would produce stable, cationic imino complexes. Treatment of the solutions containing the neutral unstable intermediate with triflic acid afforded similar products, resulting from protonation at nitrogen. The structure of one of these protonated derivatives, solved by X-ray diffraction (see Fig. 8), helped to establish the reactivity pattern sketched earlier. The products of the deprotonation of the isonitrile complexes were found to be stable both in solution and in the solid state and could be completely characterized, including the X-ray structural determination of one of the derivatives (Fig. 9). They result from the coupling between C2(imidazole) and the isonitrile carbon, resulting in the formation of

Dearomatization of N-Heterocyclic Ligands

85

Fig. 8 Molecular structure of the protonated final C–C coupling product between N-MesIm and tBuCN.

Fig. 9 Molecular structure of the C–C coupling product resulting of the deprotonation of the isocyanide compound [Re(CO)3(N-MesIm)2(tBuNC)]BAr0 4.

four-member chelates, consisting of an imidazole and a κ1(N)-iminoacyl donor groups (Scheme 23). Note that complexes with κ1(N)-iminoacyl ligands are relatively rare, as most iminoacyls are κ2(N,C)-coordinated to early transition metals. Reaction of these complexes occurred at the

86

R. Arévalo et al.

iminoacyl nitrogen, resulting in the formation of cationic aminocarbenes (see Scheme 23). It is interesting to note that the conceptually simple synthesis of nonheterocyclic carbenes by successive additions of a nucleophile and an electrophile to isonitriles remains little explored. The discussed results show that the intramolecular C–C coupling initiated by C2-deprotonation of N-alkylimidazole ligands can lead to the formation of four-member rings and to the attack on usually labile nitriles. In addition, the reactions are selective, affording the described products, and not the foreseeable products of attack on CO ligands or N-alkylimidazole coligands.

5.8 Deprotonation of N-Alkylimidazole Ligands in Phosphane–Carbonyl Complexes In view of the previously discussed results, which show that even notoriously inert coligands such as bipy and phen can be the target of deprotonated N-alkylimidazole ligands, the same group set out to explore the deprotonation of N-alkylimidazole ligands in carbonyl complexes containing only phosphane coligands.117 Since the latter cannot be the site of nucleophilic attack, the possibility of obtaining products of attack on the CO ligands was anticipated. As mentioned earlier, such an attack has been computationally found to yield intermediates in the formation of imidazolyl complexes, but stable products of the intramolecular attack of deprotonated N-RIm ligands on carbonyl coligands were not observed previously. The deprotonation of a coordinated N-methylimidazole ligand in cationic Re(I) tricarbonyl phosphane complexes [Re(CO)3(MeIm)2(PR3)] BAr0 4 employing the amide KN(SiMe3)2 showed a dramatic dependence on the nature of the phosphane. Thus, whereas for trimethyphosphane complexes, imidazolyl products were obtained, when the phosphane is either triphenylphosphane or methyldiphenylphosphane, the deprotonation followed by methylation with excess MeOTf reaction afforded the mixture of products shown in Scheme 24. Compound 15, which is the major product of the reaction, contains a binuclear cationic complex resulting from a double activation of a N-methylimidazole ligand. Its structure (Fig. 10 shows the X-ray structure of the cation of the triphenylphosphane derivative) suggests that the C2 carbon of the deprotonated imidazole attacks one of the cis CO ligands, forming a four-member metallacycles. Methylation of the oxygen of the involved carbonyl generates a methoxycarbene moiety. The same imidazolate ring binds the second rhenium fragment as an abnormal NHC through its C5 position, indicating a double imidazole deprotonation. This second rhenium center

Dearomatization of N-Heterocyclic Ligands

87

Scheme 24 Reactivity of [Re(CO)3(MeIm)2(PR3)]BAr0 4 with the strong base KN(SiMe3)2.

Fig. 10 Molecular structure of the cation present in compound 15.

completes its coordination sphere with one phosphane, two carbonyls and an N,N0 -bidentate ligand that results from the formal coupling of a molecule of CO (from the carbonyl ligand missing at this rhenium center), one imidazole and one imidazolyl ligand. The oxygen atom of the CO molecule involved in this coupling has been methylated, so the carbon atom originating from CO is now sp3-hybridized. The minor product (16 in Scheme 24) contains a cationic complex with two cis carbonyls, two trans phosphanes, and a chelate bis(C-imidazolyl)ketone ligand similar to the one just described for the binuclear complex, but without methylation at oxygen.

88

R. Arévalo et al.

Scheme 25 Transformation of a N-MeIm ligand in a NH–NHC by a deprotonation/ protonation sequence.

When compound 14 was deprotonated using n-butyllithium instead of KN(SiMe3)2, the product was a neutral imidazolyl complex which turned out to be too unstable for isolation. Its reaction with triflic acid afforded a stable, cationic complex featuring an heterocyclic carbene (NH)–NHC ligand (Scheme 25), which could be crystallized as its triflate salt and characterized in the solid state by single-crystal X-ray diffraction, and in solution by IR and NMR. This difference in reaction outcome as a function of the base employed suggests a relative stabilization of the deprotonated N-methylimidazole as a result of the interaction of its C2 carbon atom with lithium, a smaller, more polarizing cation than potassium. The formation of the products 15 and 16 obviously involved complex, multistep reactions, and proposing a meaningful mechanistic sequence would require additional work. Nevertheless, the outlined results demonstrate several points, including the deprotonation of N-alkylimidazole at C5 and subsequent formation of “abnormal” NHCs, the formation of alkoxycarbenes via intramolecular attack of a C2-deprotonated N-alkylimidazole ligand to a CO coligand, the CO activation by the attack of two C2-deprotonated N-alkylimidazole ligands, and the dramatic effect of the employed strong base on the nature of the products.

6. METAL-MEDIATED TAUTOMERIZATION OF N-HETEROCYCLES TO NHC COMPLEXES As we have mentioned earlier, the coordination of organic molecules to transition metals can be used as a pathway to activate them. In certain instances, a dramatic modification of its nature is found to be the consequence of such coordination. A well-established example of this concept is the metal-induced acetylene to vinylidene tautomerization (Fig. 11). This process has been studied, both theoretically and experimentally, and an activation barrier of 76 kcal mol1 has been found, the vinylidene species being

Dearomatization of N-Heterocyclic Ligands

89

Fig. 11 (A) Acetylene–vinylidene isomerization. (B) Metal-induced acetylene to vinylidene tautomerization.

44 kcal mol1 less stable than the acetylene isomer.118 In the presence of a number of metal fragments the process becomes thermodynamically favored and kinetically available, affording the corresponding vinylidene metal complexes, the relative energy of the two tautomers being inverted.119

6.1 Tautomerization of Pyridines The possible isomerization of pyridine to its 2-carbene tautomer120 was first proposed by Hammick as early as in 1937,121 but it took over 60 years to prove its existence. This type of tautomerization is frequently the key step in important biological processes, where the energetically less stable tautomer is often responsible for the biological activity.122 The extremely unstable pyridyl-2-ylidene was generated in the gas phase by Schwarz et al., and characterized by means of mass spectroscopy.123 Pyridin-2-ylidenes are more than 40 kcal mol1 less stable than pyridines, but coordination to transition metal fragments can strongly stabilize them. In 1987, Taube published the first pyridine carbene complex synthesized in solution, a 2,6-lutidinium ylide coordinated to Os(II) through the para carbon (Fig. 12).51 In the last 10 years, the number of pyridylidene complexes has increased considerably, probably due to the use of different strategies to provide an additional stability to the carbenic species. In 2006, Carmona et al. achieved the tautomerization of 2-substituted pyridines into the carbene tautomers mediated by the [TpMe2Ir(C6H5)2N2] (TpMe2 ¼ hydrotris(3,5-dimethylpyrazolyl)borate) fragment (Scheme 26).124 The steric hindrance provided by the substituents at the 2-position precluded the formation of the (κ1,N)-coordinated pyridine complexes while the formation of the NHCs was not affected by this issue, as the substituent in the 2-position is away from the coordinated carbon atom.

90

R. Arévalo et al.

Fig. 12 First metal–pyridylidene complex synthesized by Taube et al.

Scheme 26 Formation of Ir(III) pyridylidene complexes from 2-substituted pyridines.

Fig. 13 Stabilization of N-heterocyclic carbene complexes by intramolecular hydrogen bonds between the N–H group and chloride (A) or hydride (B) ligands reported by Esteruelas et al.

Concurrently, in the research group of Esteruelas stable carbene tautomers of quinoline and 8-methylquinoline were synthesized by coordination to osmium- and ruthenium-chloro complexes.125 It is interesting to note that an increased stability of the resulting NHC complexes is provided by an intramolecular NH⋯Cl interaction between the NH group and a chloride ligand coordinated to the same metal atom (see Fig. 13A). The bulkiness of the substituent and the coligands, along with an intramolecular hydrogen bond are also crucial in the formation of a carbene tautomer of 2-ethylpyridine bonded to a hydride osmium(II) complex (Fig. 13B).126

Dearomatization of N-Heterocyclic Ligands

91

There is just one example of a nonsubstituted pyridylidene complex obtained as a 1:1 mixture together with the N-adduct (Scheme 27).127 This outstanding transformation was mediated by an iridium complex featuring two Ir–CH2 bonds that resulted from the metallation of two methyl groups of mesityl substituents of the trispyrazolylborate ligand (TpMs ¼ hydrotris(5mesitylpyrazolyl)borate). The presence of an Ir–CH2 bond within a chelate is essential for the tautomerization process. In contrast, there are some examples in which parent pyridylidenes have been implicated in the functionalization of pyridines.128–132 In this context Bergman and coworkers found that the formation of the Rh–NHC complex 18 via C–H bond activation of the heterocycle 17 (see Scheme 28) was the key step in the Rh-catalyzed coupling of N-heterocycles and olefins.129 The 1,2-hydrogen shift involved in this process was studied in detail, using experimental and computational methods, and an intramolecular hydrogen transfer pathway through Rh–H intermediates was found to be the more plausible mechanism. Tautomerization of 2-substituted pyridines to NH–NHCs induced by Me2 Tp Ir(III) fragments (mentioned earlier) has been extended to the broadly used bipy and phen ligands.133 A new and unexpected coordination mode was found as monohapto NHCs, in which an intramolecular hydrogen bonding between the NH unit of the coordinated ring and the imine-type N atom of the other ring adds stability to these species (Fig. 14, complexes 19 and 20). A similar thermal activation was published for terpyridine which is able to act as a mono- or bidentante NHC toward one or two metal

Scheme 27 Tautomerization of nonsubstituted pyridine achieved by Carmona et al.

Scheme 28 C–H activation of N-heterocycle 17 by a Rh(I) complex.

92

R. Arévalo et al.

Fig. 14 Pyridylidene tautomers of 2,20 -bipyridine (complexes 19 and 20). Tautomerization of 1,9-phenanthroline aided by the chelation effect (structure 21).

atoms.134 A couple of examples have been published where chelation is the driving force to successfully achieve the tautomerization of a pyridine moiety.135,136 An example of this strategy is shown in Fig. 14 (complex 21) in which an Ir(I) fragment allows the formation of a stable metallacycle upon α-CH activation of 1,9-phenanthroline.

6.2 Tautomerization of Imidazoles In spite of the higher stability of imidazole-2-ylidenes and their metal complexes, tautomerization of imidazoles to the corresponding NHCs are also scarce processes. The relative stability of imidazole N- or C-metal bound isomers was computationally studied by Crabtree and Einsentein, who found a strong dependence on the nature of the metal fragment.137 Sundberg et al. published in 1974 the first example of N- to C-tautomerization of an imidazole ligand mediated by a Ru(II) complex (Scheme 29), but an acidic media was needed and very low yields were obtained.138 The only other example of tautomerization of a nonchelated imidazole ligand to its NHC form was proposed by Bergman and Ellman in their studies of Rh(I) catalyzed C–C coupling between alkenes and benzimidazole.139 The NHC is formed in situ via C–H activation of the heterocycle (Scheme 30) being one of the first examples in which a NHC plays a nonancillary role in a catalytic transformation. The additional stability gained because of the formation of a bidentate chelate ligand has been used to achieve imidazole to NHC tautomerization for a N-phosphane-functionalized imidazole in the presence of iridium140 and ruthenium141 metal fragments (structure 22 in Fig. 15). An analogous activation occurred when 2-pyridylbenzimidazole is refluxed in THF with

Dearomatization of N-Heterocyclic Ligands

93

Scheme 29 First example of a metal-mediated imidazole to NH–NHC tautomerization.

Scheme 30 Isomerization of a benzimidazole derivative to its NHC isomer promoted by Rh(I).

Fig. 15 Examples of tautomerization of N-alkylimidazole derivatives to NH–NHCs assisted by chelation.

[RuCp*Cl]4 to afford the five-membered chelate complex 23 depicted in Fig. 15,142 or in the reaction of an N-arylimine-functionalized imidazole with [Ir(cod)(μ-Cl)]2 aided by a halogen abstractor to afford complex 24 (Fig. 15).143 The reverse tautomerization, ie, from NH–NHC to imidazole ligands mediated by these two metals (Ir and Ru) have also been reported by Whittlesey144 and Li.145 The groups of Siebert146 and Erker147 studied the reactivity toward a strong base of imidazole-borane adducts, showing that for the BH3 moiety the anionic 3-boraneimidazol-2-ylidene could be isolated (structure 25 in Scheme 31), whereas migration of BR3 (R ¼ Et, C6F5) groups to the carbene carbon atom was observed for trialkyl- or triarylborane reagents (complex 26 in Scheme 31).

94

R. Arévalo et al.

Scheme 31 Reactivity of imidazole–borane adducts studied by Siebert and Erker.

Fig. 16 Analogy between an imidazolium salt (A) and a cationic imidazole metal complex (B).

Scheme 32 Tautomerization of a N-methylimidazole ligand promoted by a deprotonation/protonation sequence.

In the last years, the research interest of our group has been focused on the deprotonation of N-alkylimidazole ligands coordinated to organometallic fragments. A cationic metal complex bearing a N-RIm ligand can be regarded as a N-metallated imidazolium salt (Fig. 16), and therefore the deprotonation of the central imidazole CH group would afford a NHC in an analogous manner to the most general method used to prepare imidazol-2-ylidenes from imidazolium salts. Our results show that the species resulting from such deprotonation reaction are very reactive, mainly because of their high nucleophilic behavior, and their evolution strongly depends on the nature of the metal fragment and its coligands, as well as on the substituent of the imidazole. The reaction of [Re(CO)3(N-MeIm)3]OTf (27) with the equimolar amount of KN(SiMe3) afforded a neutral imidazol-2-yl product, 28 (Scheme 32).106,107 The formation of 28 implies, once the deprotonation

Dearomatization of N-Heterocyclic Ligands

95

has occurred, the isomerization of the heterocyclic ligand from the N- to the C-rhenium-bonded species. Examples of this type of complexes, which feature a C-bonded imidazol-2-yl ligand with a nonsubstituted nitrogen, are very rare.117,140–142,146–148 In addition, these species tend to be elusive and have been previously proposed for the tautomerization of N-alkylimidazole to NHC ligands.91 Compound 28 reacts smoothly with electrophiles, such as HOTf, to afford the product of protonation on the noncoordinated nitrogen, compound 29 (Scheme 32). The overall transformation from compound 27 to 29 can be seen as a tautomerization of an imidazole to a NH–NHC ligand promoted by a strong base, complex 28 being a stable intermediate. This reaction sequence is reminiscent of the mechanism proposed by Ruiz for the transformation of an imidazole into an NHC ligand at a Mn(I) center,91 although in that case the authors could not isolate the neutral complex (analog to 28), probably due to its lower stability. In compound 28, the heterocyclic ligand is an imidazol-2-yl, but its structural and spectroscopic properties are very close to those of the rhenium NHC compound 29 (see Fig. 17). Thus the Re–C1 distances are ˚ in 28 and 2.191(5) A ˚ in 29) and the 13C undistinguishable (2.190(7) A NMR chemical shifts for the Re-bonded carbon are 182.4 (28) and 178.7 ppm (29). The rest of the metrical data for the N-heterocyclic ligand shows a high electronic delocalization for both complexes. Also noteworthy is the similarly high trans influence, reflecting the high donor ability of both C-bonded N-heterocyclic ligands as judged from the Re–C11 bond

Fig. 17 NMR and structural similarities between imidazole-2-yl and NH–NHC ligands coordinated to {Re(CO)3(N-MeIm)2} fragment.

96

R. Arévalo et al.

˚ for 28 and 1.945(6) A˚ for 29), compared with the other distances (1.958(7) A ˚ for 28 and 1.916(6) A˚ for 29). The Re–Ccarbonyl bond distances (1.901(5) A presence of a hydrogen atom on N2 in 28 is the only difference between the metallic complexes present in 28 and 29, and it was confirmed by a topological analysis of the Laplacian of the electron density (— 2ρ). This study indicated the presence of an in-plane lone pair at N2 in complex 28 and of an N–H bond in 29, as well as the electron delocalization mentioned earlier. In contrast, complexes [Re(CO)3(N-MeIm)x(N-MesIm)3x]+ (x ¼ 1, 2), with at least one N-arylimidazole ligand, led, after deprotonation reaction, to the C–C coupling and ring-opening products,106 as it has been mentioned in Section 5.5. On the other hand, the related neutral complex [Re(OTf )(CO)3(NMeIm)2] (30) reacted instantaneously with KN(SiMe3)3 to yield compound 31 (see Scheme 33).149 The new complex contained a fac-{Re(CO)3} fragment bonded to one N-MeIm ligand, one imidazol-2-yl-ligand and one NH–NHC ligand, the last two resulting from N- to C-coordination change of two N-MeIm ligands. The N–H group of the NH–NHC ligand acts as hydrogen bond donor toward the noncoordinated nitrogen of the imidazol-2-yl ligand, contributing to the coplanarity of the two heterocyclic ligands (as shown in Fig. 18 for the analogous mesityl derivative). A molecular mirror plane is evident from the NMR spectra of 31, indicating the fast (even at low temperature) H+ transfer between the two nitrogens, ie, the complex can be described as featuring two imidazol-2-yl ligands that share a proton. The Re-bonded carbon of these ligands occurs at 180.7 ppm in the 13C NMR spectrum, and the two Re–C bond distances are undistinguishable ˚ ), showing the close similarity (Re–C2 2.214(10) and Re–C22 2.207(9) A between imidazol-2-yl and NH–NHC ligands. The yield of 31 notably increased when its preparation was conducted in the presence of the

Scheme 33 Tautomerization of two N-MeIm ligands promoted by a single equivalent of base.

Dearomatization of N-Heterocyclic Ligands

97

Fig. 18 Molecular structure of the mesityl derivative analogous to compound 31.

Scheme 34 Proposed (based on DFT computations) mechanism for the formation of imidazolyl–NHC complex 32 from the neutral bis(imidazole) derivative 30.

equimolar amount of N-methylimidazole (N-MeIm), as expected since, in its absence, part of the bis(imidazole) precursor 30 must have acted as a sacrificial source of N-MeIm. DFT calculations about the reaction mechanism showed that the most favorable one is that shown in Scheme 34. The Gibbs free energy in THF solution (in parentheses) is referred to that of the deprotonated species [Re(OTf )(CO)3(N-MeIm)2] (structure I, Scheme 34) and N-MeIm. The reaction starts with loss of triflate from I to give intermediate IIa in which

98

R. Arévalo et al.

the Re atom is simultaneously interacting with the noncoordinated N atom and the C-2 atom of the imidazolyl ligand. Intermediate IIa undergoes a rotation of the imidazole ring around the N–Re bond through TS–IIa/IIIa to give the intermediate IIIa. TS–IIIa/IVa connects IIIa with intermediate IVa wherein the two heterocyclic ligands are C-bound to the Re atom. Finally, addition of N-MeIm to IVa leads to the formation of a rhenium imidazol-2-yl(carbene) complex Va without any transition state. The formation of the rhenium imidazolyl–carbene complex would imply a Gibbs energy barrier in solution of 21.5 kcal mol1, consistent with the fast formation of the product experimentally observed. A key feature of the proposed mechanism is the intermediacy of η2-N,C-imidazolyl complexes, which make possible ligand dissociation without going through high-energy fivecoordinate species. Stable η2-N,C-imidazolyl complexes have been disclosed by Diaconescu et al. in scandium and uranium chemistry.150 The reaction of 31 with trifluoromethanesulfonic acid (HOTf ) afforded 32, the triflate salt of the bis(NH–NHC) complex resulting from protonation at nitrogen (Scheme 33). The overall formation of 32 from 30 involves, besides the substitution of OTf by the entering imidazole, the formation of two new Re–C bonds at the expense of the two Re–N bonds. As we have discussed earlier, deprotonation of a coordinated N-MeIm ligand in [Re(CO)3(N-MeIm)3]OTf followed by protonation of the resulting imidazol-2-yl ligand affords a NHC complex (see Scheme 32). The formation of the bis(carbene) complex from the bis(imidazole) precursor is not just twice that process, because the addition of only one equivalent of base triggers the Re–N to Re–C rearrangement of two imidazole ligands. It is also noteworthy that in the deprotonation reactions of (a) the tris(N-MesIm) compound or (b) the triflato complex 30 in the presence of free N-MesIm, the components of the reactant mixture are the same, whereas different isomers are obtained as products. The formation of the ring-opening product in the former case or the imidazolyl-carbene in the latter shows again the extreme sensitivity of the reaction course to the exact nature and number of the ligands. The wide variety of products obtained from the reactivity of the rhenium compounds prompted us to extend our studies of deprotonation of N-alkylimidazole ligands to molybdenum organometallic species.112 The chosen cationic compounds of formula [Mo(η3-C4H7)(bipy)(CO)2 (N-RIm)]OTf feature, like those of rhenium, pseudooctahedral structures in which the imidazole and each ring of the bipy ligand are in cis positions. The addition of a strong base to [Mo(η3-C4H7)(bipy)(CO)2(N-RIm)]+

Dearomatization of N-Heterocyclic Ligands

99

complexes, followed by protonation or alkylation of the resulting neutral product, leads to the formation of Mo–NHC compounds (Scheme 35). This reactivity constitutes another example in which the addition of a strong base, followed by an acid promotes the tautomerization of an N-alkylimidazole ligand to NH–NHC. In agreement with the experimental results, a DFT study showed that the most favorable reaction mechanism for the Mo(II) complexes consisted in an initial attack of the imidazole deprotonated carbon atom onto a cis-CO ligand, to afford in a second step imidazol-2-yl species (Fig. 19). This mechanism is reminiscent of the one found for [Mn(bipy)(CO)3(N-RIm)]OTf compounds,151 suggesting that this new and unexpected “carbonyl mechanism” can be quite general for related transformations. As it is shown in Fig. 19, pathway A, that led to the C–C coupling product, is favored over pathway B, which affords formation of the imidazole-2-yl complex, by only 3.5 kcal mol1. In addition to this, the formation of Mo-IIb (7.0 kcal mol1) implies a Gibbs energy barrier of only 4.8 kcal mol1, which is clearly lower than the one required for obtaining Mo-IIa (8.3 kcal mol1). The better kinetic accessibility of Mo-IIb in conjunction with the fact that its evolution to Mo-IIIb shows a Gibbs energy barrier in solution (11.8 kcal mol1) clearly lower than that for the reversion to Mo-IIa (15.3 kcal mol1) via Mo-I, and that the Mo-IIIb product is considerably stable (13.4 kcal mol1), make B the most favorable pathway, affording the formation of imidazol-2-yl products instead of the C–C coupling species, as it is encountered experimentally. In 2013, Darensbourg evidenced the occurrence of a base-promoted conversion of a N-alkylimidazole ligand to the corresponding NH–NHC mediated by a dinitrosyl iron complex.152

Scheme 35 Synthesis of imidazol-2-yl and NHC complexes from N-alkylimidazole ligands coordinated to {Mo(η3-C4H7)(bipy)(CO)2} fragment.

Mo-TSIa

Mo-TSIb

Mo-TSIIb

8.3 kcal mol−1 4.8 kcal mol−1 4.8 kcal mol−1

−8.1 kcal mol−1

−7.0 kcal mol−1

−13.4 kcal mol−1

Mo-IIb Mo-IIa Mo-I

Mo-IIIb

Route A Route B

Fig. 19 Schematic view of the reaction mechanisms found for the deprotonation reaction of [Mo(η3-C4H7)(bipy)(CO)2(N-MesIm)]+ at the B3LYP/6-31G(d) (LANL2DZ for Mo) level of theory.

Dearomatization of N-Heterocyclic Ligands

101

6.3 Tautomerization of Other Azoles Compared with imidazol-2-ylidenes less attention has been paid to N,Xheterocyclic carbenes (X ¼ O, S), mainly due to the instability of the free ligands which easily undergo dimerization processes.153 A probably better alternative would be to force the tautomerization of the coordinated azole ligand upon a deprotonation/protonation (or alkylation) sequence as that described earlier for N-alkyilimidazole ligands. This method is illustrated in Scheme 36 for {Mn(azole)(CO)3(bipy)}+ (azole ¼ oxazole, thiazole) complexes to afford the corresponding HCs.154 Prior to this work, deprotonation of 4-methylthiazole155 or benzothiazole156 ligands coordinated to {Cr(CO)5} followed by methylation had led to the formation of the carbene complexes, but lack of selectivity had been observed as the dimethylated azole complexes were also obtained.

7. METAL–LIGAND COOPERATION BASED ON AROMATIZATION/DEAROMATIZATION PROCESSES The chemical behavior of transition metal complexes is dramatically dependent on the steric and electronic properties of the ligands coordinated to the metal center. Modification of the ligands properties has been used as a tool to improve the desired properties of metal complexes depending on the particular processes, ie, increasing its electrophilic/nucleophilic, basic/acid character, etc. A particular area in which this concept has been broadly used is in homogeneous catalysis. In the vast majority of reactions catalyzed by metal complexes, the catalytic activity is primarily based on the metal while the ligands are considered as spectators, as they do not participate in the chemical transformations directly. In the last decade, the development of bifunctional catalysts has gained interest showing that the metal center and its coordinated ligands can jointly cooperate in the catalytic cycle.157

Scheme 36 Tautomerization of oxazole and thiazole to the corresponding carbene derivatives at a Mn(I) fragment.

102

R. Arévalo et al.

A particular case of this type of metal–ligand cooperation (MLC), studied by Milstein et al., is based on dearomatization/aromatization processes on the ligand. These studies have been comprehensively covered in several recent reviews,158–160 but due to their relevance and fundamental importance they have been, although briefly, also included herein. These authors found that pyridine-based PNP and PNN (structures 33 and 34, Fig. 20), and bipyridine-based (structure 35, Fig. 20) pincer ligands can stabilize coordinatively unsaturated complexes. The facile deprotonation of the pincer backbone at one of the methylene spacer groups affords the dearomatization of the pyridine ring, which must be stabilized by being part of the pincer ligand. The dearomatized complexes can activate various chemical bonds including H–H, C–H, O–H, and N–H bonds. These activations are carried out by cooperation between the metal and the ligand, regaining aromatization of the ligand by protonation of the benzylic carbon atom (Scheme 37). Several pincer complexes of this type, mainly based on Ru158 and Fe,159 which are able to participate in MLC processes, have been described showing their ability as versatile catalysts for hydrogenation, dehydrogenation, and related reactions. It has to be highlighted that the first application of dearomatization–reprotonation was the employment of complex 36 for the dehydrogenation of alcohols, wherein the combined metal and dearomatized ligand act in cooperative manner to convert, for example, primary alcohols to esters or secondary alcohols to ketones, with concomitant extrusion of H2 (Scheme 38).161 Milstein and coworkers reported that, in a similar manner to aliphatic alcohols, water could also be activated using complex 36 with the dearomatized PNN backbone (Scheme 39).162 Upon reaction with water at room temperature, aromatization takes place to quantitatively form the trans-hydrido–hydroxo complex 37. Thermal activation of a second

Fig. 20 PNP and PNN pincer ligands studied by Milstein et al.

Dearomatization of N-Heterocyclic Ligands

103

Scheme 37 Activation of H–Y chemical bonds by cooperation of the metal and the ligand through a dearomatization/rearomatization process.

Scheme 38 Dehydrogenation of alcohols to esters or ketones with a dearomatized pincer complex.

molecule of water releases molecular hydrogen and leads to the formation of cis-hydroxo complex 38. DFT studies indicate that this process involves H2 liberation from 37 by coupling of the hydride ligand with a proton from the side arm, followed by addition of H2O to the generated dearomatized intermediate. Significantly, irradiation of complex 38 led to the regeneration of compound 37 concomitant with evolution of O2. This sequence would involve a photochemically induced reductive elimination of H2O2 from dihydroxo complex 38, thereby generating a Ru(0) intermediate (39, not observed) that undergoes intramolecular proton transfer to regenerate the

104

R. Arévalo et al.

Scheme 39 Proposed reaction mechanism for the generation of O2 and H2 from H2O mediated by a Ru(PNN) complex.

starting compound. Therefore, this is a new approach toward a complete cycle for the generation of dihydrogen and dioxygen from water promoted by a soluble metal complex.

8. CONCLUSIONS AND OUTLOOK Pyridines react with several types of main group organometallic compounds to give metal–pyridine adducts, proposed on the basis of computational studies or by extrapolation of the behavior encountered in nonreactive systems. They evolve to metal complexes of anionic dihydropyridines, which have been isolated only in a few cases, and which finally yield substituted pyridines. Only in the last few years, some groups have started to employ N-coordination of the pyridine to an external Lewis acid (eg, BF3) combined with nucleophilic addition of an organometallic reagent, or the reaction of the pyridine with a bimetalic reagent containing different Lewis acid and nucleophilic metal sites (eg, lithiun aryl zincates),

Dearomatization of N-Heterocyclic Ligands

105

obtaining selective, synthetically useful methodologies. In contrast, the pyridine rings in transition metal complexes containing pyridyl-based ligands—which are among the most useful and widely studied within coordination chemistry—are typically very inert. In the 1970s, this notion was challenged by Gillard, who proposed that apparent anomalies in the behavior of cationic bipy and phen transition metal complexes in basic aqueous media could be the result of hydroxide attack to one of the pyridine carbon atoms. However, subsequent studies casted doubt on this proposal. Our group demonstrated several instances of C–C coupling via intramolecular nucleophilic attack to the pyridyl rings on Re-coordinated bipy and phen ligands under very mild conditions, most of them initiated by deprotonation of C–H groups of several types of commonly employed monodentate ligands. The products of such reactions feature metal-bonded, dearomatized dihydropyridyl groups. The products of coupling between two monodentate N-alkylimidazole or pyridine ligands could be rearomatized by oxidation to afford pyridylimidazole or bipyridine ligands, including nonsymmetric chelates. The reaction of one of the dearomatized products of the bipy-Nalkylimidazole coupling with excess MeOTf led to pyridine ring opening, a transformation previously limited to three examples of very reactive early transition metals. C–C coupling schemes were extended to ligands other than pyridines, allowing the modular construction of nonsymmetric polydentate ligands on the metal coordination sphere. In some cases, the deprotonation of coordinated N-alkylimidazoles afforded very rare imidazole-2-yl ligands. Their reaction with electrophiles afford NHC ligands; in particular, their reaction with acids yield NH,NHC ligands. This kind of reaction can serve to directly access metal complexes of NHC ligands directly from stable, easily accessible N-alkylimidazole complexes. Computational studies proved to be very useful in shedding light on the kinetics and thermodynamics of the competing pathways that lead in some systems to C–C coupling and in others to imidazolyl formation. Most of these reactions have been carried out employing highly stable, easily available rhenium tricarbonyl complexes, whose chemistry has been previously almost exclusively dominated by substitution reactions. A growing number of types of transition metal compounds are finding applications as catalysts, materials, and drugs. Therefore, the synthesis of transition metal complexes is becoming a proper goal beyond their classical employment as auxiliaries toward the synthesis of metal-free organic molecules. The traditional synthetic approach is based on the separate synthesis of the ligands employing conventional organic procedures, which in the case of

106

R. Arévalo et al.

nonsymmetric, polydentate ligands can be far from straightforward, followed by their binding to the metal through substitution reactions. In some cases, this scheme could be advantageously replaced by the modular synthesis of the polydentate ligands from easily available monodentate ligands, which would be first coordinated to the metal and then coupled, taking advantage of the ligand activation and geometric preorganization by their metal coordination.

ACKNOWLEDGMENTS Computational studies carried out by Dr. Ramo´n Lo´pez (Universidad de Oviedo) and Dr. Jesu´s Dı´az (Universidad de Extremadura) have been instrumental to our incipient understanding of the challenging mechanistic issues. The authors gratefully acknowledge funding from Ministerio de Economı´a y Competitividad and FEDER (grants CTQ2012-37379-C02-01 and CTQ2012-37379C02-02, and FPI graduate studentships to M.E.V. and M.A.H.) and Principado de Asturias (grant FC-15-GRUPIN14-103, administered by FICYT) and Ministerio de Educacio´n Cultura y Deporte (FPU graduate studentship to R.A.).

REFERENCES 1. Balaban AT, Oniciu DC, Katritzky AR. Aromaticity as a cornerstone of heterocyclic chemistry. Chem Rev. 2004;104:2777–2812. 2. Sundberg RJ, Martin RB. Interactions of histidine and other imidazole derivatives with transition metal ions in chemical and biological systems. Chem Rev. 1974;74:471–517. 3. Welton T. Room-temperature ionic liquids. Solvents for synthesis and catalysis. Chem Rev. 1999;99:2071–2083. 4. Hopkinson MN, Richter C, Schedler M, Glorius F. An overview of N-heterocyclic carbenes. Nature. 2014;510:485–496. 5. Zhao G, Lin H. Metal complexes with aromatic N-containing ligands as potential agents in cancer treatment. Curr Med Chem Anticancer Agents. 2005;5:137–147. 6. Kaes C, Katz A, Hosseini MW. Bipyridine: the most widely used ligand. A review of molecules comprising at least two 2,20 -bipyridine units. Chem Rev. 2000;100: 3553–3590. 7. Sammes PG, Yahioglu G. 1,10-Phenanthroline: a versatile ligand. Chem Soc Rev. 1994;23:327–334. 8. Shopov DY, Rudshteyn B, Campos J, Batista VS, Crabtree RH, Brudvig GW. Stable iridium(IV) complexes of an oxidation-resistant pyridine-alkoxide ligand: highly divergent redox properties depending on the isomeric form adopted. J Am Chem Soc. 2015;137:7243–7250. 9. Chupakhin ON, Charushin VN, van der Plas HC. Nucleophilic Aromatic Substitution of Hydrogen. 1st ed. San Diego, CA: Academic Press Inc.; 1994. 10. Shaw AP, Ryland BL, Franklin MJ, Norton JR, Chen JY-C, Hall ML. Using a twostep hydride transfer to achieve 1,4-reduction in the catalytic hydrogenation of an acyl pyridinium cation. J Org Chem. 2008;73:9668–9674. 11. Butschke B, Schwarz H. “Rollover” cyclometalation-early history, recent developments, mechanistic insights and application aspects. Chem Sci. 2012;3:308–326.

Dearomatization of N-Heterocyclic Ligands

107

12. Mertel HE, Mizzoni RH, Shaw EN, Tenenbaum LE, Yale HL. Pyridine and its derivatives. In: Weissberger A, Klingsberg E, eds. The Chemistry of Heterocyclic Compounds. New York, NY: Interscience Publishers Inc; 1961. 13. Bull JA, Mousseau JJ, Pelletier G, Charette AB. Synthesis of pyridine and dihydropyridine derivatives by regio- and stereoselective addition to N-activated pyridines. Chem Rev. 2012;112:2642–2713. 14. Donohoe TJ, Connolly MJ, Walton L. Regioselective nucleophilic addition to pyridinium salts: a new route to substituted dihydropyridones. Org Lett. 2009;11: 5562–5565. 15. McGill CK, Rappa A. Advances in the Chichibabin reaction. Adv Heterocycl Chem. 1988;44:1–79. 16. Ma Y, Breslin S, Keresztes I, Lobkovsky E, Collum D. Synthesis of a 7-azaindole by chichibabin cyclization: reversible base-mediated dimerization of 3-picolines. J Org Chem. 2008;73:9610–9618. 17. Mastalir M, Pittenauer E, Allmaier G, Kirchner K. 2,6-Diamination of substituted pyridines via heterogeneous chichibabin reaction. Tetrahedron Lett. 2016;57:333–336. 18. Terrier F. Modern Nucleophilic Aromatic Substitution. 1st ed. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co.; 2013. 19. Fier PS, Hartwig JF. Selective C–H fluorination of pyridines and diazines inspired by a classic amination reaction. Science. 2013;342:956–960. 20. Rudnitskaya A, To¨ro¨k B, Dransfield T. A B3LYP/6-31 + G(d) study of the reaction pathways and conformational preference in a model Chichibabin reaction. Comput Theor Chem. 2011;963:191–199. 21. Van der Plas HC. The SN(ANRORC) mechanism: a new mechanism for nucleophilic substitution. Acc Chem Res. 1978;11:462–468. 22. Breuker K, Van der Plas HC. Occurrence of an SN(ANRORC) mechanism in the chichibabin amination of 4-phenylpyrimidine. J Org Chem. 1979;44:4677–4680. 23. Francis RF, Davis W, Wisener JT. Reaction intermediates in the alkylation of pyridine with tert-butyllithium. J Org Chem. 1974;39:59–62. 24. Matsubara Y, Kosaka T, Koga K, et al. Formation of η2-coordinated dihydropyridine– ruthenium(II) complexes by hydride transfer from ruthenium(II) to pyridinium cations. Organometallics. 2013;32:6162–6165. 25. Harrison DP, Zottig VE, Kosturko GW, et al. Stereo- and regioselective nucleophilic addition to dihapto-coordinated pyridine complexes. Organometallics. 2009;28: 5682–5690. 26. Rudler H, Martı´n-Vaca B, Nicolas M, Audouin M, Vaissermann J. Formation and selective trapping of 2,5-dihydropyridine and its isopropylidene derivative by tungsten(0) alkylidene complexes. X-ray structure of the 5-isopropylidene-2,5dihydropyridine adduct. Organometallics. 1998;17:361–366. 27. Moretto AF, Liebeskind LS. Stereo- and regiocontrolled construction of trisubstituted piperidines using a TpMo(CO)2(dihydropyridine) scaffold (Tp ¼ hydridotrispyrazolylborate). J Org Chem. 2000;65:7445–7455. 28. Barr D, Snaith R, Mulvey RE, Reed D. Characterization of the bis(pyridine) complex of the butylllithium-pyridine adduct, 2-Bun(C5H5N)Li  (C5H5N)2; its decomposition to 2-butylpyridine via a dihydropyridine. Polyhedron. 1988;7:665–668. 29. Clegg W, Dunbar L, Horsburgh L, Mulvey RE. Stoichiometric dependence of the long-established reaction of butyllithium with pyridine: a hidden secondary reaction that produces a pyridine adduct of a lithiodihydropyridine. Angew Chem Int Ed Engl. 1996;35:753–755. 30. Armstrong DR, Harris CMM, Kennedy AR, et al. Developing lithium chemistry of 1,2-dihydropyridines: from kinetic intermediates to isolable characterized compounds. Chem Eur J. 2015;21:14410–14420.

108

R. Arévalo et al.

31. Chen Q, du Jourdin M, Knochel P. Transition-metal-free BF3-mediated regioselective direct alkylation and arylation of functionalized pyridines using Grignard or organozinc reagents. J Am Chem Soc. 2013;135:4958–4961. 32. Chen Q, Leo´n T, Knochel P. Transition-metal-free BF3-mediated oxidative and nonoxidative cross-coupling of pyridines. Angew Chem Int Ed. 2014;53:8746–8750. 33. Mizumori T, Hata T, Urabe H. Alkylation of pyridines at their 4-positions with styrenes plus yttrium reagent or benzyl Grignard reagents. Chem Eur J. 2015;21:422–426. 34. Watson SC, Eastham JF. Colored indicators for simple direct titration of magnesium and lithium reagents. J Organomet Chem. 1967;9:165–168. 35. Kaim W. Radical-forming electron-transfer reactions involving main-group organometallics. Acc Chem Res. 1985;18:160–166. 36. Kaim W. Electron transfer to complex ligands. Radical anions and organomagnesium radical complexes of 2,20 -bipyridines and 1,10-phenanthrolines. J Am Chem Soc. 1982;104:3833–3837. 37. Ashby EC, Goel AB. Evidence for a single electron transfer mechanism in the reduction of aromatic nitrogen heterocycles with main group metal hydrides and dihydropyridyl metal complexes. Tetrahedron Lett. 1981;22:4783–4786. 38. O’Hara F, Blackmond DG, Baran PS. Radical-based regioselective C–H functionalization of electron-deficient heteroarenes: scope, tunability, and predictability. J Am Chem Soc. 2013;135:12122–12134. 39. Jochmann P, Dols TS, Spaniol TP, Perrin L, Maron L, Okuda J. Insertion of pyridine into the calcium allyl bond: regioselective 1,4-dihydropyridine formation and C–H bond activation. Angew Chem Int Ed. 2010;49:7795–7798. 40. Hensen K, Lemke A, Stumpf T, et al. Synthesis and structural characterization of (1,4dihydropyrid-1-yl)aluminum complexes. Inorg Chem. 1999;38:4700–4704. 41. Knijnenburg Q, Smits JMM, Budzelaar PHM. Reaction of the diimine pyridine ligand with aluminum alkyls: an unexpectedly complex reaction. Organometallics. 2006;25: 1036–1046. 42. Lichtenberg C, Spaniol TP, Okuda J. Reactivity of tris(allyl)aluminum toward pyridine: coordination versus carbometalation. Organometallics. 2011;30:4409–4417. 43. Herna´n-Go´mez A, Herd E, Uzelac M, et al. Zincate-mediated arylation reactions of acridine: pre- and postarylation structural insights. Organometallics. 2015;34:2614–2623. 44. Diaconescu PL. Reactions of early transition metal-carbon bonds with N-heterocycles. Curr Org Chem. 2008;12:1388–1405. 45. Johnson KRD, Hayes PG. Cyclometalative C–H bond activation in rare earth and actinide metal complexes. Chem Soc Rev. 2013;42:1947–1960. 46. Bianchini C, Meli A, Vizza F. Modelling the hydrodenitrogenation of aromatic N-heterocycles in the homogeneous phase. Eur J Inorg Chem. 2001;43–68. 47. Angelici RJ. An overview of modeling studies in HDS, HDN and HDO catalysis. Polyhedron. 1997;16:3073–3088. 48. Neithamer DR, Pa´rka´nyi L, Mitchell JF, Wolczanski PT. η2-(N,C)-pyridine and μ–η2(1,2): η2(4,5)-benzene complexes of (silox)3Ta (silox ¼ t-Bu3SiO-). J Am Chem Soc. 1988;110:4421–4423. 49. LaPointe RE, Wolczanski PT, Mitchell JF. Carbon monoxide cleavage by (silox)3Ta (silox ¼ t-Bu3SiO-). J Am Chem Soc. 1986;108:6382–6384. 50. Kleckley TS, Bennett JL, Wolczanski PT, Lobkovsky EB. Pyridine C¼N bond cleavage mediated by (silox)3Nb (silox ¼ t-Bu3SiO). J Am Chem Soc. 1997;119:247–248. 51. Cordone R, Taube H. Pentaammineosmium(II)-η2-2,6-lutidine: an intermediate for C–H activation. J Am Chem Soc. 1987;109:8101–8102. 52. Harman WD. The activation of aromatic molecules with pentaammineosmium(II). Chem Rev. 1997;97:1953–1978.

Dearomatization of N-Heterocyclic Ligands

109

53. Cordone R, Harman WD, Taube H. Carbon–hydrogen bond activation in novel η2bound cationic heterocycle complexes of pentaammineosmium(II). J Am Chem Soc. 1989;111:2896–2900. 54. Gray SD, Smith DP, Bruck MA, Wigley DE. Regioselective C–N bond cleavage in an η2(N,C)-coordinated pyridine and an η1(N) ! η2(N,C) bonding rearrangement in coordinated quinoline: models for hydrodenitrogenation catalysis. J Am Chem Soc. 1992;114:5462–5463. 55. Gray SD, Weller KJ, Bruck MA, Briggs PM, Wigley DE. Carbon–nitrogen bond cleavage in an η2(N,C)-pyridine complex induced by intramolecular metal-to-ligand alkyl migration: models for hydrodenitrogenation catalysis. J Am Chem Soc. 1995;117:10678–10693. 56. Weller KJ, Filippov I, Briggs PM, Wigley DE. Pyridine degradation intermediates as models for hydrodenitrogenation catalysis: preparation and properties of a metallapyridine complex. Organometallics. 1998;17:322–329. 57. Bailey BC, Fan H, Huffman JC, Baik M-H, Mindiola DJ. Room temperature ringopening metathesis of pyridines by a transient Ti  C linkage. J Am Chem Soc. 2006;128:6798–6799. 58. Fout AR, Bailey BC, Tomaszewski J, Mindiola DJ. Cyclic denitrogenation of N-heterocycles applying a homogeneous titanium reagent. J Am Chem Soc. 2007;129:12640–12641. 59. Gillard RD, Lyons JR. Equilibria in complexes of heterocyclic ligands. J Chem Soc Chem Commun. 1973;585–586. 60. Gillard RD. An explanation for anomalies among complexes of N-heterocyclic ligands. Inorg Chim Acta. 1974;11:L21–L22. 61. Gillard RD. Equilibria in complexes of N-heterocyclic molecules. Part III (1,2). An explanation for classical anomalies among complexes of 1,10-phenanthrolines and 2,20 -bipyridyls. Coord Chem Rev. 1975;16:67–94. 62. Nord G. The base hydrolysis of bis(2,20 -bipyridyl)platinum(II) perchlorate. Acta Chem Scand A. 1975;A29:270–272. 63. Nord G. Some properties of 2,20 -bipyridine,1,10-phenanthroline and their metal complexes. Comments Inorg Chem. 1985;4:193–212. 64. Constable EC. Nucleophilic attack upon coordinated heterocycles; definitive evidence for the enhanced electrophilicity of coordinated pyridines. Inorg Chim Acta. 1984;82:53–57. 65. Constable EC, Housecroft CE. Ligand reactivity in coordination compounds; a molecular orbital investigation of the quaternisation-coordination analogy. Transition Met Chem. 1988;13:19–21. 66. Tzalis D, Tor Y. The organic chemistry of coordination compounds: unprecedented substitution reactions of functionalized polypyridine complexes. Angew Chem Int Ed Engl. 1997;36:2666–2668. 67. Lay PA. Does coordination activate heterocycles toward nucleophilic attack? The importance of π interactions between metal ion and ligand. Inorg Chem. 1984;23:4775–4777. 68. McInnes CS, Clare BR, Redmond WR, Clark CR, Blackman AG. The nature of the [Pt(bipy)2]2+ ion in aqueous alkaline solution: a new look at an old problem. Dalton Trans. 2003;2215–2218. 69. Zhang X-M, Tong M-L, Chen X-M. Hydroxylation of N-heterocycle ligands observed in two unusual mixed-valence CuI/CuII complexes. Angew Chem Int Ed. 2002;41:1029–1031. 70. Szpakolski KB, Latham K, Rix CJ, White JM, Moubaraki B, Murray KS. Synthetic and structural studies on copper 1H-[1,10]-phenanthrolin-2-one coordination complexes:

110

71. 72. 73. 74.

75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87.

88. 89.

R. Arévalo et al.

isolation of a novel intermediate during 1,10-phenanthroline hydroxylation. Chem Eur J. 2010;16:1691–1696. Constable EC. Forty years on—covalent hydration of transition metal complexes revisited. Polyhedron. 2016;103:295–306. Hevia E, Pe´rez J, Riera L, Riera V, Miguel D. Reactive alkoxide complexes of groups 6 and 7 metals. Organometallics. 2002;21:1750–1752. Hevia E, Pe´rez J, Riera L, et al. Insertion of unsaturated organic electrophiles into molybdenum-akoxide and rhenium-alkoxide bonds of neutral, stable carbonyl complexes. Chem Eur J. 2002;8:4510–4521. Morales D, Navarro Clemente ME, Pe´rez J, Riera L, Riera V, Miguel D. The combination of organometallic {Mo(η3-allyl)(CO)2(phen)} fragments and hard aquo and hydroxo ligands: controlled synthesis and structural characterization. Organometallics. 2002;21:4934–4938. Gerbino DC, Hevia E, Morales D, et al. A new reactivity pattern of low-valent transition-metal hydroxo complexes: straightforward synthesis of hydrosulfido complexes via reaction with carbon disulfide. Chem Commun. 2003;328–329. Cuesta L, Gerbino DC, Hevia E, et al. Reactivity of molybdenum and rhenium hydroxocarbonyl complexes toward organic electrophiles. Chem Eur J. 2004;10:1765–1777. Hevia E, Pe´rez J, Riera V, Miguel D. New octahedral rhenium (I) tricarbonyl amido complexes. Organometallics. 2002;21:1966–1974. Hevia E, Pe´rez J, Riera V, Miguel D. Different sites of insertion in the reaction of isocyanates with [Re(N(R)Ar)(CO)3(bipy)] (R ¼ H or Me): N–H vs. Re–N. Chem Commun. 2002;1814–1815. Morales D, Pe´rez J, Riera L, Riera V, Miguel D. Formation of a 1-azaallenylidene ligand by reaction of an amido complex with tetracyanoethylene. Inorg Chem. 2002;41:4111–4113. Hevia E, Pe´rez J, Riera V, Miguel D. Reactivity of the amido complex [Re(NHpTol) (CO)3(bipy)] toward neutral organic electrophiles. Organometallics. 2003;22:257–263. Morales D, Pe´rez J, Riera L, et al. Molybdenum amido complexes with single Mo–N bonds: synthesis, structure and reactivity. Chem Eur J. 2003;9:4132–4143. Hevia E, Pe´rez J, Riera V, Miguel D. Insertion and cycloaddition reactivity of a transition-metal N-metalloimine. Angew Chem Int Ed. 2002;41:3858–3860. Hevia E, Pe´rez J, Riera V, et al. Synthesis of β-lactams from a N-rhenaimine: effect of the transition metal on the energetic profile of the Staudinger reaction. J Am Chem Soc. 2003;125:3706–3707. Cuesta L, Hevia E, Morales D, et al. Activation of a 1,10-phenanthroline ligand on a rhenium tricarbonyl complex. Chem Commun. 2005;116–117. Methot JL, Roush WR. Nucleophilic phosphine organocatalysis. Adv Synth Catal. 2004;346:1035–1050. Cuesta L, Hevia E, Morales D, et al. Activation of ancillary ligands in the reactions of DMAD with phosphide and alkylidenamido rhenium complexes. Organometallics. 2005;24:1772–1775. Adams H, Bailey NA, Blenkiron P, Morris MJ. Metallacycle formation through the linking of an alkyne with phosphido and acetyl groups at an iron centre: X-ray structure of [CpFe(CO){Ph2PCH(CO2Me)C(CO2Me) ¼ C(Me)O}] (Cp¼η5-C5H5). J Organomet Chem. 1993;460:73–81. Ashby MT, Enemark JH. Cycloaddition of alkenes and alkynes to CpFe(CO)2PR2 to give (R)2PFeC(¼O)C¼C heterometallacycles. Organometallics. 1986;6:1323–1327. Serrano AL, Casado MA, Ciriano MA, de Bruin B, Lo´pez JA, Tejel C. Nucleophilicity and PC bond formation reactions of a terminal phosphanido iridium complex. Inorg Chem. 2016;55:828–839.

Dearomatization of N-Heterocyclic Ligands

111

90. Huertos MA, Pe´rez J, Riera L. Pyridine ring opening at room temperature at a rhenium tricarbonyl bipyridine complex. J Am Chem Soc. 2008;130:5662–5663. 91. Ruiz J, Perandones BF. Base-promoted tautomerization of imidazole ligands to N-heterocyclic carbenes and subsequent transmetalation reaction. J Am Chem Soc. 2007;129:9298–9299. 92. Hilf C, Bosold F, Harms K, et al. Carbene structure of stable acyl (formyl) anion equivalents. Chem Ber/Recueil. 1997;130:1201–1212. 93. Hilf C, Bosold F, Harms K, Marsch M, Boche G. The equilibrium between 2-lithium-oxazole(-thiazole,-imidazole) derivatives and their acyclic isomers—a structural investigation. Chem Ber/Recueil. 1997;130:1213–1221. 94. Gade LH. Transition metal complexes with polydentate amido ligands: novel structural building blocks and chemical reagents. J Organomet Chem. 2002;661:85–94. 95. Caulton KG. The influence of π-stabilized unsaturation and filled/filled repulsions in transition metal chemistry. New J Chem. 1994;18:25–41. 96. Are´valo R, Pe´rez J, Riera L. Intramolecular nucleophilic addition to the 2 position of coordinated 2,20 -bipyridine by a deprotonated dimethyl sulfide ligand. Inorg Chem. 2013;52:6785–6787. 97. Are´valo R, Pe´rez J, Riera L. Deprotonation of coordinated phosphanes in a rhenium complex: C–C coupling with diimine coligands. Chem Eur J. 2015;21:3546–3549. 98. Fehlhammer WP, Fritz M. Emergence of a CHN and cyano complex based organometallic chemistry. Chem Rev. 1993;93:1243–1280. 99. Dick GR, Woerly EM, Burke MD. A general solution for the 2-pyridyl problem. Angew Chem Int Ed. 2012;51:2667–2672. 100. Sato S, Ishitani O. Photochemical reactions of fac-rhenium(I) tricarbonyl complexes and their application for synthesis. Coord Chem Rev. 2015;282–283:50–59 and references therein. 101. Leonidova A, Gasser G. Underestimated potential of organometallic rhenium complexes as anticancer agents. ACS Chem Biol. 2014;9:2180–2193. 102. Espinal Viguri M, Huertos MA, Pe´rez J, Riera L, Ara I. Re-mediated C–C coupling of pyridines and imidazoles. J Am Chem Soc. 2012;134:20326–20329. 103. Espinal Viguri M, Pe´rez J, Riera L. C–C coupling of N-heterocycles at the fac-Re(CO)3 fragment: synthesis of pyridylimidazole and bipyridine ligands. Chem Eur J. 2014;20:5732–5740. 104. Miller KL, Williams BN, Benitez D, et al. Dearomatization reactions of N-heterocycles mediated by group 3 complexes. J Am Chem Soc. 2010;132:342–355. 105. Carver CT, Benitez D, Miller KL, et al. Reactions of group III biheterocyclic complexes. J Am Chem Soc. 2009;131:10269–10278. 106. Huertos MA, Pe´rez J, Riera L, Mene´ndez-Vela´zquez A. From N-alkylimidazole ligands at a rhenium center: ring opening or formation of NHC complexes. J Am Chem Soc. 2008;130:13530–13531. 107. Huertos MA, Pe´rez J, Riera L, Dı´az J, Lo´pez R. Effect of the nature of the substituent in N-alkylimidazole ligands on the outcome of deprotonation: ring opening versus the formation of N-heterocyclic carbene complexes. Chem Eur J. 2010;16:8495–8507. 108. Carver CT, Diaconescu PL. Ring-opening reactions of aromatic N-heterocycles by scandium and yttrium alkyl complexes. J Am Chem Soc. 2008;130:7558–7559. 109. Grace ME, Loosemore MJ, Semmel ML, Pratt RF. Kinetics and mechanism of the Bamberger cleavage of imidazole and of histidine derivatives by diethyl pyrocarbonate in aqueous solution. J Am Chem Soc. 1980;102:6784–6789. 110. Trofimov BA, Andriyankova LV, Nikitina LP, et al. Stereoselective tandem ring opening of imidazoles with electron-deficient acetylenes and water: synthesis of functionalized (Z,Z)-1,4-diaza-2,5-dienes. Org Lett. 2013;15:2322–2324.

112

R. Arévalo et al.

111. He F, Ruhlmann L, Gisselbrech J-P, et al. Dinuclear iridium and rhodium complexes with bridging arylimidazolide-N3,C2 ligands: synthetic, structural, reactivity, electrochemical and spectroscopic studies. Dalton Trans. 2015;44:17030–17044. 112. Brill M, Dı´az J, Huertos MA, Lo´pez R, Pe´rez J, Riera L. Imidazole to NHC rearrangements at molybdenum centers: an experimental and theoretical study. Chem Eur J. 2011;17:8584–8595. 113. Cebollada A, Espinal Viguri M, Pe´rez J, Dı´az J, Lo´pez R, Riera L. Influence of the N–N coligand: C–C coupling instead of formation of imidazol-2-yl complexes at {Mo(η3-allyl)(CO)2} fragments. Theoretical and experimental studies. Inorg Chem. 2015;54:2580–2590. 114. Espinal Viguri M, Huertos MA, Pe´rez J, Riera L. Imidazole-nitrile or imidazoleisonitrile C–C coupling on rhenium tricarbonyl complexes. Chem Eur J. 2013;19:12974–12977. 115. Hevia E, Pe´rez J, Riera V, Miguel D, Kassel S, Rheingold A. New synthetic routes to cationic rhenium tricarbonyl bipyridine complexes with labile ligands. Inorg Chem. 2002;41:4673–4679. 116. Lu E, Lewis W, Blake AJ, Liddle ST. The ketimide ligand is not just an inert spectator: heteroallene insertion reactivity of an actinide–ketimide linkage in a thorium carbene amide ketimide complex. Angew Chem Int Ed. 2014;53:9356–9359. 117. Huertos MA, Pe´rez J, Riera L. Double activation of an N-alkylimidazole. Chem Eur J. 2012;18:9530–9533. 118. Gallo MM, Hamilton TP, Schaefer III HF. Vinylidene: the final chapter? J Am Chem Soc. 1990;112:8714–8719. 119. Bruce MI. Organometallic chemistry of vinylidene and related unsaturated carbenes. Chem Rev. 1991;91:197–257. 120. Kunz D. Synthetic routes to N-heterocyclic carbene complexes: pyridine-carbene tautomerizations. Angew Chem Int Ed. 2007;46:3405–3408. 121. Dyson P, PLl Hammick. Experiments on the mechanism of decarboxylation. Part I. Decomposition of quinaldinic and isoquinaldinic acids in the presence of compounds containing carbonyl groups. J Chem Soc. 1937;1724–1725. 122. Raczy nska ED, Kosi nska W, Os´mialowski B, Gawinecki R. Tautomeric equilibria in relation to pi-electron delocalization. Chem Rev. 2005;105:3561–3612. 123. Lavorato D, Terlouw JK, Dargel TK, Koch W, McGibbon GA, Schwarz H. Observation of the Hammick intermediate: reduction of the pyridine-2-ylid ion in the gas phase. J Am Chem Soc. 1996;118:11898–11904. 124. Alvarez E, Conejero S, Paneque M, et al. Iridium(III)-induced isomerization of 2-substituted pyridines to N-heterocyclic carbenes. J Am Chem Soc. 2006;128:13060–13061. ´ lvarez FJ, On˜ate E. Stabilization of NH tautomers of quin125. Esteruelas MA, Ferna´ndez-A olines by osmium and ruthenium. J Am Chem Soc. 2006;128:13044–13045. 126. Buil ML, Esteruelas MA, Garce´s K, Oliva´n M, On˜ate E. Understanding the formation of N–H tautomers from α-substituted pyridines: tautomerization of 2-ethylpyridine promoted by osmium. J Am Chem Soc. 2007;129:10998–10999. ´ lvarez E, Conejero S, Lara P, et al. Rearrangement of pyridine to its 2-carbene tau127. A tomer mediated by iridium. J Am Chem Soc. 2007;129:14130–14131. 128. Hayton TW, Boncella JM, Scott BL, Abboud KA, Mills RC. Coupling of an aldehyde or ketone to pyridine mediated by a tungsten imido complex. Inorg Chem. 2005;44:9506–9517. 129. Wiedemann SH, Lewis JC, Ellman JA, Bergman RG. Experimental and computational studies on the mechanism of N-heterocycle C–H activation by Rh(I). J Am Chem Soc. 2006;128:2452–2462.

Dearomatization of N-Heterocyclic Ligands

113

130. Lewis JC, Bergman RG, Ellman JA. Rh(I)-catalyzed alkylation of quinolines and pyridines via C–H bond activation. J Am Chem Soc. 2007;129:5332–5333. 131. Berman AM, Lewis JC, Bergman RG, Ellman JA. Rh(I)-catalyzed direct arylation of pyridines and quinolines. J Am Chem Soc. 2008;130:14926–14927. 132. Johnson DG, Lynam JM, Mistry NS, Slattery JM, Thatcher RJ, Whitwood AC. Ruthenium-mediated C–H functionalization of pyridine: the role of vinylidene and pyridylidene ligands. J Am Chem Soc. 2013;135:2222–2234. 133. Conejero S, Lara P, Paneque M, et al. Monodentate, N-heterocyclic carbene-type coordination of 2,2ʹ-bipyridine and 1,10-phenanthroline to iridium. Angew Chem Int Ed. 2008;47:4380–4383. 134. Paneque M, Poveda ML, Vattier F, A´lvarez E, Carmona E. Synthesis and structural characterization of a binuclear iridium complex with bridging, bidentate N-heterocyclic carbene coordination of 2,2ʹ:6ʹ,200 -terpyridine. Chem Commun. 2009;5561–5563. 135. Song G, Li Y, Chen S, Li X. Hydrogen bonding-assisted tautomerization of pyridine moieties in the coordination sphere of an Ir(I) complex. Chem Commun. 2008;3558–3560. 136. Song G, Su Y, Periana RA, et al. Anion-exchange-triggered 1,3-shift of an NH proton to iridium in protic N-heterocyclic carbenes: hydrogen-bonding and ion-pairing effects. Angew Chem Int Ed. 2010;49:912–917. 137. Sini G, Eisenstein O, Crabtree RH. Preferential C-binding versus N-binding in imidazole depends on the metal fragment involved. Inorg Chem. 2002;41:602–604. 138. Sundberg RJ, Bryan RF, Taylor Jr IF, Taube H. Nitrogen-bound and carbon-bound imidazole complexes of ruthenium ammines. J Am Chem Soc. 1974;96:381–392. 139. Tan KL, Bergman RG, Ellman JA. Intermediacy of an N-heterocyclic carbene complex in the catalytic C–H activation of a substituted benzimidazole. J Am Chem Soc. 2002;124:3202–3203. 140. Miranda-Soto V, Grotjahn DB, DiPasquale AG, Rheingold AL. Imidazol-2-yl complexes of Cp*Ir as bifunctional ambident reactants. J Am Chem Soc. 2008;130: 13200–13201. 141. Miranda-Soto V, Grotjahn DB, Cooksy AL, Golen JA, Moore CE, Rheingold AL. A labile and catalytically active imidazol-2-yl fragment system. Angew Chem Int Ed. 2011;50:631–635. 142. Araki K, Kuwata S, Ikariya T. Isolation and intercoversion of protic N-heterocyclic carbene and imidazolyl complexes: application to catalytic dehydrative condensation of N-(2-pyridyl)benzimidazole and allyl alcohol. Organometallics. 2008;27:2176–2178. 143. He F, Braunstein P, Wesolek M, Danopoulos AA. Imine-functionalised protic NHC complexes of Ir: direct formation by C–H activation. Chem Commun. 2015;51:2814–2817. 144. Burling S, Mahon MF, Powell RE, Whittlesey MK, Williams JMJ. Ruthenium induced C–N bond activation of an N-heterocyclic carbene: isolation of C- and N-bound tautomers. J Am Chem Soc. 2006;128:13702–13703. 145. Wang X, Chen H, Li X. Ir(III)-induced C-bound to N-bound tautomerization of a N-heterocyclic carbene. Organometallics. 2007;26:4684–4687. 146. Wacker A, Pritzkow H, Siebert W. Borane-substituted imidazol-2-ylidenes: syntheses, structures and reactivity. Eur J Inorg Chem. 1998;843–849. 147. Vagedes D, Kehr G, Ko¨nig D, et al. Formation of isomeric BAr3 adducts of 2-lithio-Nmethylimidazole. Eur J Inorg Chem. 2002;2015–2021. 148. Ko¨sterke T, Ko¨sters J, Wu¨rthwein E-U, et al. Synthesis of complexes containing an anionic NHC ligand with an unsubstituted ring nitrogen atom. Chem Eur J. 2012;18:14594–14598.

114

R. Arévalo et al.

149. Huertos MA, Pe´rez J, Riera L, Dı´az J, Lo´pez R. From bis(N-alkylimidazole) to bis(NH–NHC) in rhenium carbonyl complexes. Angew Chem Int Ed. 2010;49: 6409–6412. 150. Diaconescu PL. Reactions of aromatic N-heterocycles with d0fn-metal alkyl complexes supported by chelating diamide ligands. Acc Chem Res. 2010;43:1352–1363. 151. Ruiz J, Perandones BF, Van der Maelen JF, Garcı´a-Granda S. On the existence of an N-metalated N-heterocyclic carbene: a theoretical study. Organometallics. 2010;29:4639–4642. 152. Hsieh C-H, Pulukkody R, Darensbourg MY. A dinitrosyl iron complex as a platform for metal-bound imidazole to N-heterocyclic carbene conversion. Chem Commun. 2013;49:9326–9328. 153. Chien SW, Yen SK, Hor TSA. N,S-heterocyclic carbene complexes. Aust J Chem. 2010;63:727–741. 154. Ruiz J, Perandones BF. Metal-induced tautomerization of oxazole and thiazole molecules to heterocyclic carbenes. Chem Commun. 2009;2741–2743. 155. Raubenheimer HG, Stander Y, Marais EK, et al. Group 6 carbene complexes derived from lithiated azoles and the crystal structure of a molybdenum thiazolinylidene complex. J Organomet Chem. 1999;590:158–168. 156. Raubenheimer HG, Kruger GJ, Lombard AA, Linford L, Viljoen JC. Sulfurcontaining metal complexes. 12. Reactions of α-thio carbanions with carbene complexes of the type [M(CO)5{O(alkyl)Ar}] and with the carbyne [(η5-MeC5H4)Mn(CO)2(CPh)][BCl4]. Organometallics. 1985;4:275–284. 157. Ikariya T, Masakatsu S. Topics in Organometallic Chemistry. Bifunctional Catalysis. vol. 37. Berlin: Springer; 2011. 158. Gunanathan C, Milstein D. Bond activation and catalysis by ruthenium pincer complexes. Chem Rev. 2014;114:12024–12087. 159. Zell T, Milstein D. Hydrogenation and dehydrogenation iron pincer catalysts capable of metal–ligand cooperation by aromatization/dearomatization. Acc Chem Res. 2015;48:1979–1994. 160. van der Vlugt JI, Reek JNH. Neutral tridentate PNP ligands and their hybrid analogues: versatile non-innocent scaffolds for homogeneous catalysis. Angew Chem Int Ed. 2009;48:8832–8846. 161. Zhang J, Leitus G, Ben-David Y, Milstein D. Facile conversion of alcohols into esters and dihydrogen catalyzed by new ruthenium complexes. J Am Chem Soc. 2005;127:10840–10841. 162. Kohl SW, Weiner L, Schwartsburd L, et al. Consecutive thermal H2 and light-induced O2 evolution from water promoted by a metal complex. Science. 2009;324:74–77.

CHAPTER THREE

Insight into Metal-Catalyzed Water Oxidation from a DFT Perspective D. Balcells* University of Oslo, Oslo, Norway *Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Manganese Biomimetic Catalysts 2.1 The Crabtree–Brudvig Catalyst 2.2 The Åkermark Catalyst 2.3 The McKenzie Catalyst 2.4 Tetrameric Catalysts 3. The Blue Dimer and Other Dinuclear Ruthenium Catalysts 3.1 The Meyer “Blue Dimer” Catalyst 3.2 The Llobet Catalyst 3.3 The Tanaka Catalyst 3.4 The Sun Catalyst 4. Mononuclear Catalysts 4.1 Iridium 4.2 Iron 5. Conclusions and Perspectives Acknowledgments References

115 120 120 127 128 129 132 132 139 142 145 147 147 156 165 168 168

1. INTRODUCTION One of the main issues already challenging modern societies is the generation, storage, and distribution of energy. The main sources of energy used nowadays include oil, coal, and natural gas (Fig. 1). All of these are carbon-based and often require cumbersome extraction, treatment, and transport processes. Further, they release energy upon combustion generating pollutants like the toxic oxides of nitrogen and sulfur, which are a major Advances in Organometallic Chemistry, Volume 65 ISSN 0065-3055 http://dx.doi.org/10.1016/bs.adomc.2016.01.001

#

2016 Elsevier Inc. All rights reserved.

115

116

D. Balcells

Fig. 1 Worldwide consumption of primary energy sources (IEA report 2014).

concern regarding public health in large urban areas. Another major concern is the massive emission of carbon dioxide to the atmosphere, which is the main greenhouse effect gas stimulating the climate change. From an economic perspective, cost is also an issue in the long term due to the constant grow of overpopulated countries combined with the depletion of the deposits and the political instability on major extraction regions. There is thus an urgent need for a new energy scheme1 which should be cheap, environment friendly, and based on renewable sources. Nowadays, these represent c.1–2% of the total worldwide consumption—a very small share that should grow strongly in the near future. Sunlight is one of the most abundant energy sources available on Earth’s crust. It is indeed the primary source in nature and, indirectly, provides the essential nutritional needs of the humankind. Photosynthesis is the technology evolved by nature over hundreds of millions of years for transforming sunlight into chemical energy.2 The key reaction in this process is the oxidation of water, also referred to as water splitting: 2H2 O ! 4H + + 4e + O2 "

(1)

Despite its simplicity, this reaction is extremely challenging due to its strong endoergic character, involving an oxidation potential of 1.23 V vs the standard hydrogen electrode. Further, it is kinetically encumbered by a high energy barrier, thus requiring significant overpotentials. Plants tackle these two issues efficiently by using the photosystem II (PSII),3 which is a large protein complex embedded into the chloroplasts thylakoid membrane. The key components of PSII are the chlorophylls, which harvest sunlight energy to compensate the unfavorable thermodynamics of Eq. (1), and

117

Metal-Catalyzed Water Oxidation

the oxygen-evolving complex (OEC), which lowers the high energy barriers encumbering the kinetics of Eq. (1).4 The Kok cycle (Fig. 2) gives an overall picture of the reaction mechanism driving the catalytic oxidation of water in the OEC. The latter is based on a manganese tetranuclear core, which has the ability of managing the four moles of H+ + e generated by the reaction through different states, labeled S0–S4, which contain different combinations of manganese (III) and (IV) centers. Plants use the reducing equivalents generated in reaction (1) to synthesize biomolecules from carbon dioxide. Alternatively, these equivalents can be used to generate hydrogen: 4H + + 4e ! 2H2 "

(2)

This simple reaction is a key component in the construction of artificial photosynthesis devices (Fig. 3).5 The reductive formation of hydrogen in the cathode is coupled with the oxidative splitting of water in the photoanode. Once hydrogen has been collected and stored, this solar fuel can be

Fig. 2 The Kok cycle in natural photosynthesis.

Fig. 3 General scheme for an artificial photosynthesis device.

118

D. Balcells

thereafter used to generate electricity in a fuel cell by performing the reverse exoergic reaction: 4H + + 4e + O2 ! 2H2 O

(3)

This energy scheme for power generation is extremely attractive due to several major advantages6; ie, (1) it is fully based on renewable sources, including sunlight, oxygen, and water, which, further, is recycled within the process; (2) it is free of greenhouse effect gases emissions; and (3) it can be implemented at a rather small “home-scale,” thus excluding the need for building and maintaining complex distribution grids. Nonetheless, the practical use of artificial photosynthesis is still hampered by several limitations in the technologies available nowadays. These include the need for more cost-effective fuel cells7 and high-capacity materials for safe hydrogen storage.8 Further, water splitting is not enough efficient yet due to its endoergic and slow character. As for the OEC in PSII, artificial photosynthesis relies on efficient molecular catalysts accelerating the oxidation of water.9 The ideal catalyst must have all these four features: (1) high activity, ie, the reaction must be fast enough as to run at significant rates under mild conditions (eg, room temperature at atmospheric pressure); (2) high robustness, ie, the catalyst must be stable over long time periods of months or years and thereafter be easily recyclable; (3) cheap and environment friendly, ie, the metal center should be based on an earth-abundant metal with low levels of toxicity (eg, late transition metals in the first row); and (4) modular, ie, the catalyst must be easily attached to other molecular components and materials (eg, the photosensitizer for sunlight energy harvesting and the semiconductor for electronic coupling to the cathode). Despite intensive research in the field during the last two decades,10–14 there is not yet a single catalytic system capable of optimizing all these four features. In order to achieve this goal, the mechanistic details of the reaction should be precisely known at the molecular level. Catalytic water oxidation mechanisms are diverse and strongly dependent on the nature of the catalyst used, but their elementary steps can be grouped into three general stages, including catalyst activation, O–O bond formation, and oxygen release. The first stage involves the oxidation of the catalyst to a high energy state, driven by the electron holes opened by the photosensitizer, which is often modeled by a chemical sacrificial oxidant. In this regard, the key data of interest are the redox potentials of the species involved. The second stage

Metal-Catalyzed Water Oxidation

119

is the most important one, since the formation of the O–O bond is the ratedetermining step of the overall process. The intrinsic difficulty of this step originates from the same and high electronegativity of the two oxygen atoms. This issue is eased by stimulating the electrophilic character in the acid–base (AB) and direct-coupling (DC) mechanisms (Fig. 4). In the AB mechanism, O-electrophilicity is promoted by electron-delocalization in a ligand-supported metal-oxo moiety, which undergoes the intermolecular nucleophilic attack of a water molecule. In this scenario, the nature of the proton acceptor plays a key role. In the DC mechanism, two electrophilic oxo moieties form the O–O bond upon intramolecular coupling. In most cases, these formal oxo moieties, M]O, are actually radical oxyl moieties, M–O%, and the DC mechanism can be thus envisaged as the coupling of two oxygen atom radicals. Regardless of the mechanism, this oxidation chemistry requires antagonistic properties in the ancillary ligands, since these should act as both donors to promote catalyst oxidation and acceptors to promote electrophilicity in the oxygen atoms involved. The donor–acceptor ability of these ligands should hence be carefully tuned and balanced. The last stage of the mechanism, oxygen release, can also be rather complex since it requires further redox chemistry and may involve spin crossover toward the triplet ground state of the product. Catalytic water splitting has been studied extensively with several experimental techniques, including X-ray diffraction, cyclic voltammetry, stopped-flow kinetics, and several spectroscopic methods (NMR, IR, UV–VIS, etc.). The insight given by these techniques is often limited by the rather complex nature of the oxidation chemistry involved. Once the

Fig. 4 Acid–base (AB; top) and direct-coupling (DC; bottom) mechanisms in the catalytic formation of the O–O bond.

120

D. Balcells

active species is generated, this can initiate an intricate reaction network involving elementary steps with largely different thermodynamic and kinetic parameters. This network may mask the key intermediates which, in general, are difficult to isolate and characterize. In this regard, computational chemistry is a powerful tool complementing and adding to the experimental work. Thanks to the excellent compromise between accuracy and cost offered by modern density functional theory (DFT),15 this approach can be used to characterize the molecular structure of transient intermediates and the otherwise inaccessible transition states16 and minimum-energy crossing points (MECP) for spin crossover.17 Further, free reaction energies and barriers are also available for each elementary step, which allows for the prediction of redox potentials and rate constants. Spectra can also be simulated and used to verify those recorded in the experiments and to assign the structure.18 The computational data also help in the interpretation of the results; eg, time-dependent DFT (TDDFT) can be used both to simulate the intensity and broadness of the peaks in a UV–VIS spectrum and to identify the electronic transitions originating them. Another powerful feature of the computational approach is the analysis of the electronic structure based on the frontier molecular orbitals,19 natural bond orbitals,20 and electron density.21 These analyses provide concepts relating molecular properties, including structures, local charges, and spectroscopic parameters, to catalytic properties, including energy barriers, oxidation potentials, and resistance toward self-oxidation. These concepts can be exploited in the rational design and development of new catalysts with enhanced activity and robustness.22 This review focuses on both purely theoretical and joint theoretical– experimental studies on molecular systems used in the homogeneous catalytic oxidation of water. Studies on the OEC and the heterogeneous catalysts, which have been already reviewed,23,24 are thus excluded. This review is not comprehensive but rather contains a collection of works which have been selected on the basis of their interest and contribution to the field.

2. MANGANESE BIOMIMETIC CATALYSTS 2.1 The Crabtree–Brudvig Catalyst An obvious approach to metal-catalyzed water oxidation is to mimic the structure and function of the OEC in PSII (Fig. 5). This species is based on a cubane structure with three manganese centers, four bridging oxo ligands, and one calcium cation on its vertices. An extra manganese center is oxo-bridged to one Mn–O edge of the cubane and all cationic metal

Metal-Catalyzed Water Oxidation

121

Fig. 5 Structures of the oxygen-evolving complex in photosystem II (Mn4O6Ca core shown only; top-left), the Crabtree–Brudvig catalyst (top-right), the Åkermark catalyst (bottom-left), and the McKenzie catalyst (bottom-right).

centers are stabilized by carboxylate groups from the amino acids of the protein backbone. The numerous computational studies on the structure and reactivity of this complex system have been already reviewed by Siegbahn23 and Batista.25 The first functional model of the OEC based on oxo-bridged manganese moieties was developed by Crabtree and Brudvig.26 This catalyst contains a mixed-valence (μ-O-Mn(III,IV))2 core, in which the metal centers are supported by strongly N-donating terpyridine (terpy) ligands capable of stabilizing the high oxidation states required. 18O-isotope-labeling experiments proved the evolution of oxygen from water in sodium hypochlorite solutions. EPR measurements supported an initial mechanistic proposal involving the oxidation of one metal center to a Mn(V)]O state, which forms oxygen upon nucleophilic attach of a hydroxide anion. Lundberg, Blomberg, and Siegbahn carried out a systematic DFT study on the Crabtree–Brudvig catalyst (Fig. 5), which focused on the electronic structure of the oxidized species and the mechanism of the critical O–O bond formation process.27 The computational tools used in this work included a modified B3LYP hybrid functional, with the amount of HF exchange reduced to 15% for a better description of the spin states.

122

D. Balcells

Double- and triple-ζ basis sets were used to optimize structures and refine energies, respectively. The latter also included the solvent effects of water by using a continuum model. Antiferromagnetic spin states were computed with the broken-symmetry DFT approach. Due to the limitations of the resources available at that time, two different models were used in order to reduce computational costs. In the small model, which was used in the systematic exploration of all possible reaction pathways, two phenyl rings were removed from the terpy ligand. This requires using imine C]N–H moieties to avoid dangling bonds and these introduce artificial H-bonds with the reacting molecules of water. The key stationary points were thus refined in a large full model, by performing restricted optimizations with the geometrical parameters of the small model. The authors carried out a thorough exploration of all possible electronic states of the postulated active species, which, formally, is a mixed-valence MnVMnIV-oxo complex. The calculations showed that this complex is only stable in open-shell configurations with radical character on the oxygen atom. The formal MnV-oxo moiety is thus better described as MnIV-oxyl. The unpaired electron on oxygen can couple either ferro- or antiferromagnetically with the unpaired electrons on the MnIV center, which, in addition, can have either all spin-up, """, or spin-up/spin-down, ""#, configurations. Further, these spin configurations are also coupled through the bridging (μ-O)2 moiety to the other MnIV atom. This complex scenario yielded a total of nine electronic states optimized with different spin configurations. The configuration of the ground state was described as MnIV(###)O(")MnIV("""). This doublet 3+ charged species has oxyl character, which elongates and activates the Mn–O bond. Remarkably, the oxyl character was preserved on a series of computational experiments, including (1) altering the amount of HF exchange of the functional, (2) introducing Cl counteranions to avoid possible artifacts from the gas-phase optimization of a 3 + charged system, (3) reoptimization in a water-solvating continuum model, and (4) extending the basis set with an additional set of polarization functions. The reactivity of the MnIV-oxyl moiety was explored by considering the nucleophilic attack of one water molecule in the AB mechanism (Fig. 4). The reaction takes place in a single step, in which water transfers a proton to the closest μ-O ligand and the incipient hydroxide forms the O–O bond with the oxyl. The transition state involves a five-membered MnO3H ring (Fig. 6). On the reactants side, water is H-bound to the μ-O ligand, whereas, on the products side, water appears split across the new μ-OH and κ-OOH

Metal-Catalyzed Water Oxidation

123

Fig. 6 Transition state for the formation of the O–O bond in the nucleophilic attack of water to the ground state of [(terpy)(OH2)Mn(μ-O)2Mn(O)(terpy)]3+.

ligands. Upon the attack of water, the unpaired oxyl electron reduces the metal center from MnIV to MnIII. This mechanism was observed in two different spin states—the ground state, MnIV(###)O(")MnIV("""), and the closest in energy, MnIV(""")O(")MnIV(###), which stands 8.8 kcal mol1 above. The ground state involves the lowest energy barrier, ΔG{ ¼ 23.4 kcal mol1 (vs 27.6 kcal mol1) but is more endoergic, ΔG ¼ 19.1 kcal mol1, than the other, ΔG ¼ 8.9 kcal mol1. The lowest barrier is in reasonable agreement with the kinetics observed experimentally, which are consistent with an effective barrier of 19–21 kcal mol1 in the rate-determining step. Interestingly, calculations on manganese complexes with no activity showed that these species have closed-shell MnV-oxo moieties in their ground states, thus suggesting that oxyl character is an essential requirement for catalytic water oxidation. The practical application of the Crabtree–Brudvig catalyst on artificial photosynthesis devices (Fig. 3) requires its implementation on a photoanode capturing the energy of sunlight. With this aim, the same group performed the direct adsorption of the catalyst onto titanium oxide nanoparticles. The surface structure of this system was characterized by means of several techniques, including computational modeling by Batista and coworkers.28 This required the combination of several different methods due to the high complexity of this system. Nanoparticle models were constructed from bulk TiO2 anatase previously optimized by means of periodic DFT calculations at the PW91/GGA level with plane-wave basis and ultra-soft Vanderbilt pseudopotentials. Each nanoparticle consists on a fragment including 32 [TiO2] units with dangling bonds capped with hydrogens, yielding Ti–OH moieties. Adsorption of the catalyst is modeled with the reaction in which one Mn–OH bond of the catalyst reacts with a Ti–OH bond on the (101) surface leading to the formation of water and the [Ti–ONP]–MnCAT adsorbate (NP ¼ nanoparticle; CAT ¼ catalyst); ie, only one of the two manganese centers is bound to the NP through a Ti–O–Mn

124

D. Balcells

bond. This system is optimized in the antiferromagnetic state by means of QM/MM calculations. The QM part includes the catalyst at the DFT (B3LYP) level with double-ζ quality basis sets including pseudopotentials on manganese. The MM part includes the TiO2 NP described with the universal force field (UFF). The QM/MM approach used included electron embedding with both partitions polarized through bidirectional electrostatic interactions. Electronic structures were also explored by means of Extended-Hu¨ckel simulations of the UV–VIS spectra. The QM/MM calculations showed that the adsorption process is strongly exothermic with a binding enthalpy of 54 kcal mol1. This suggests that catalyst adsorption takes place in near-amorphous TiO2 NP, which has a higher content of Ti–OH moieties compared to crystalline NP. This is also consistent with the experimental data collected in the same study, including EPR, UV–VIS and electrochemical data. The spin density of the catalyst is not affected to a significant extent upon adsorption. Nonetheless, the geometry does undergo significant distortions compared to the free catalyst; eg, the Mn–N bonds between the metal centers and the terpy ligands are weakened as proven by their distance elongation. The UV– VIS spectra are also affected, with the bands observed at 450 and 650 nm blue-shifted to 420 and 650 nm upon adsorption. Other transitions are observed between these two bands, albeit with lower intensities, involving electron transitions between the catalyst and the TiO2 surface. A key feature in catalytic oxidations is the generation of the active species from the precatalyst, which in most cases is a highly oxidized transition metal complex. The understanding of this process requires a detailed understanding of its redox and AB chemistry, which can be predicted by means of quantum chemical calculations. Nonetheless, both the theory level and model should be carefully defined in order to obtain reliable results. Batista and coworkers benchmarked a protocol for the prediction of redox potentials, E0, and pKas based on DFT calculations on the [(bpy)2Mn2(μ-O)2]3+ complex, considered as a model of the OEC structure (Fig. 5).29 This work proved that DFT (B3LYP) calculations can yield accurate E0 and pKa values, if energies are refined with large basis sets (correlation consistent of triple-ζ quality with a double set of polarization functions) on a continuum solvation model (Poisson–Boltzmann self-consistent reaction field [SCRF]). The computational results were validated by comparison to experimental IR and cyclic voltammetry data. The calculations showed that the pKa of the bridging oxo ligands has a strong dependence on the oxidation state of the metal; eg, oxidation from the Mn(III)Mn(III) to the Mn(IV)Mn(III)

Metal-Catalyzed Water Oxidation

125

state causes a variation of 10 pH units in the pKa. This redox process is thus better described as the proton-coupled electron transfer (PCET) reaction shown in Eq. (3):  3 +  3 + ðbpyÞ2 Mn2 ðμ  OHÞðμ  OÞ ! ðbpyÞ2 Mn2 ðμ  OÞ2 + H + + e (4)

This benchmarked protocol was thereafter used by the same group to explore the redox chemistry of the original Crabtree–Brudvig catalyst (Fig. 5).30 Similar to the bridging oxo ligands in the benchmark bpy system, the pKa of the terminal aqua ligands was found to be extremely sensitive to the oxidation state of the metal core, with variations as large as 13 pH units. However, in contrast with the bpy system, the terpy catalyst is characterized by redox potentials that are highly dependent on the pH of the reaction media, as shown by simulated Pourbaix diagrams. Interestingly, acetate anions from the pH buffer used experimentally can undergo ligand exchange with the aqua ligands, which also has a strong impact on the redox potentials. This study concluded that the oxidation of the Mn(III)(IV) state to the Mn(IV)(IV) involves the PCET shown in Eq. (4): 

2 + ðH2 OÞðterpyÞMnðμ  OÞ2 MnðterpyÞðAcOÞ  2 + ! ðHOÞðterpyÞMnðμ  OÞ2 MnðterpyÞðAcOÞ + H + + e

(5)

Reference calculations on the decoupled reactions with the aqua-bound systems show that both PCET and H2O $ AcO exchange contribute to lower oxidation potentials by preventing an excessive accumulation of positive charge upon the active species. The role played by the acetate anions once the active species is generated was also explored by Batista and coworkers by using their benchmarked DFT approach.31 This study focused on the three key steps shown in Fig. 7 involving the full oxidation of water into triplet oxygen, ie, concerted atom–proton transfer (APT), followed by intramolecular proton-coupled electron transfer (IPCET), followed by proton transfer (PT). The initial oxyl species is generated by PCET from the product Mn(IV)Mn(IV) species in Eq. (5). In the initial APT step, water interacts with both the catalyst, through an H-bond connected to the bridging oxo ligand, and one acetate anion, through an H-bond connected to one of the carboxylic oxygen atoms. In the lowest energy transition state, the O–O bond forms upon PT to the acetate. The barrier associated with this step, 2.8 kcal mol1, is dramatically lower than that associated to the alternative pathway in which

126

D. Balcells

Fig. 7 Evolution of the metal core of the Crabtree–Brudvig catalyst along the reaction pathway for the acetate-assisted oxidation of water.

the proton transfers to the bridging oxo (Fig. 6), 22.3 kcal mol1. This reaction takes place from the antiparallel antiferromagnetic state, MnIV(""") MnIV(###)O("); the alternative parallel configuration, MnIV(###) MnIV(""")O("), was also explored, leading to higher energy pathways. The resulting hydroperoxo intermediate undergoes a low-barrier IPCET (ΔG{ ¼ 2.7 kcal mol1) yielding a Mn(III) end-on η1-superoxo species, which readily releases triplet oxygen upon protonation of the other bridging oxo ligand. Overall, this study reveals the cocatalytic role played by the acetate buffer. In addition to reducing the high oxidation potentials required for the generation of the active species, the acetate anions act as a Brønsted AB catalyst by accepting the first proton released by water. The cocatalytic effects of the acetate buffer are only possible after ligand exchange with one of the aqua ligands of the catalyst. This process was studied computationally by Zheng and coworkers at the DFT level,32 by using

Metal-Catalyzed Water Oxidation

127

the hybrid B3LYP functional with an ECP on manganese combined with double- (optimizations) and triple-ζ (single points for energy refinement) quality basis sets. The solvation effects of water were included in the model by means of the continuum polarized continuum model (PCM) method. Further, additional calculations on the key energy barriers with the B97D functional showed the minor influence of dispersion forces in this catalytic system. The Crabtree–Brudvig catalyst is mixed-valence with two nonequivalent metal centers: one Mn(III) and one Mn(IV). The calculations revealed that ligand exchange follows an associative mechanism with concomitant association of the acetate ligand and dissociation of the aqua. The concerted transition state involves the four-member MnIII/IV⋯O(H)⋯ H⋯O(Ac)⋯MnIII/IV. The reaction is almost thermoneutral for both metal centers, with a slight preference for the Mn(IV), ΔG ¼ 1.5 vs 2.0 kcal mol1. Nonetheless, from the kinetics point of view, ligand exchange is much faster at the Mn(III) center, ΔG{ ¼ 22.0 vs 28.8 kcal mol1, which is consistent with the model used by Batista in the study of the acetate buffer effects.

2.2 The Åkermark Catalyst Following the work of Crabtree and Brudvig, other research groups developed similar manganese biomimetics based on the dinuclear Mn2(μ-O)2 ˚ kermark and coworkers (Fig. 5),33 core. In the catalyst developed by A the tridentate-N,N,N-terpy ligand was replaced by a penta-dentate-O,N, O,N,O ligand containing a fused mix of imidazole, carboxylate, and phenoxide donating moieties. The manganese centers are linked by the phenoxide moiety and two bridging ligands, including one methoxide and one acetate. This system offers two key advantages: (1) the possibility of driving the reaction with light, by using either [Ru(bpy)3]2+ or [Ru(bpy)2(deeb)]2+ as photosensitizer, and (2) the higher robustness conferred by the imidazole moieties, which are more resistant to oxidation than the pyridine. In collaboration with the theoretical group of Siegbahn, A˚kermark further developed this system with the aim of increasing its catalytic activity.34 Eight different catalysts were synthesized with different groups installed at the para-phenoxide position. The most active catalyst was functionalized with an amide bound to a distal carboxylate group through an aliphatic (CH2)4 linker. This catalyst was studied computationally at the DFT (B3LYP) level with Pople’s double- and triple-ζ quality basis sets, including Stuttgart/Dresden (SDD) pseudopotentials accounting for the relativistic

128

D. Balcells

Fig. 8 Stabilization of the Åkermark catalyst active species before (left) and after (right) intramolecular proton transfer.

effects introduced by manganese. Solvation effects were modeled with the continuum universal solvation model (SMD) method, and dispersion forces were described by means of the Grimme’s approach. The postulated active species is on a mixed-valence Mn(IV)Mn(V) state, with one hydroxo ligand bound to each metal center trans to the bridging acetate. A full model was used in the calculations, including the distal carboxylate group. The geometry optimization of this system revealed that this group can reach the Mn2(μ-O)2 core by adopting an eclipsed conformation, which is stabilized by means of two hydrogen bonds between the carboxylate and the Mn–OH moieties (Fig. 8). This suggests that the distal group can promote both the PT and PCET reactions required by the formation of the O–O bond. The calculations also showed that these reactions may involve the N–H moiety of the imidazole rings; eg, PT from imidazole to the distal carboxylate raises the energy of the system by only 2.0 kcal mol1. Further geometry optimizations excluded the substitution of the bridging acetate by the distal carboxylate group.

2.3 The McKenzie Catalyst McKenzie and coworkers also contributed to the development of dinuclear biomimetic catalysts for water oxidation.35 The McKenzie catalyst (Fig. 5) is rather unique when compared to the Crabtree–Brudvig and A˚kermark systems; ie, it is based on a lower oxidation state Mn(II)Mn(II) core, in which the bridging oxygen atoms are monoanionic and part of the mcbpen ligand. This ligand is a tetra-dentate N-donor with two pyridine and two amine moieties. One of the latter is functionalized by one anionic carboxylate group, which is believed to play a key role by bridging the two metal centers, by introducing H-bonding with the apical aqua ligands and by

Metal-Catalyzed Water Oxidation

129

stabilizing higher oxidation states with different coordination numbers (7 for Mn(II) and 6 for Mn(IV)). In the postulated mechanism, oxygen is formed by intramolecular O⋯O coupling (Fig. 4) in a highly oxidized Mn2IV(μ-O)2 species. This mechanism was thereafter excluded in a computational study with Sameera and McGrady (vide infra), which revealed a low energy pathway for the AB mechanism. One distinct feature of this catalytic system, proven by isotope-labeling experiments, is the fact that one oxygen atom in the O2 evolved comes from water, whereas the other comes from the oxidant, which is tert-butyl-peroxide. The reaction mechanism in the McKenzie system was studied computationally in collaboration with Sameera and McGrady.36 The model included the full catalyst treated at the broken-symmetry DFT level with the hybrid B3LYP functional and double-ζ quality basis set with an ECP for manganese. This method is able to reproduce the crystal structure of the dimer with fairly high accuracy. Geometries were fully optimized in gas phase and energies thereafter refined considering water as solvent by means of the SCRF approach. The calculations show that the catalyst can be hydrolyzed yielding the corresponding Mn(II)-aqua monomer with a free energy cost of 11.8 kcal mol1. The oxidation of this species to a Mn(IV)-oxo complex gives access to an equilibrium between three species, which is the mechanistic core of the McKenzie catalyst (Fig. 9). Dimerization leads to the formation of a Mn2IV(μ-O)2 intermediate in equilibrium with a mixed-valence MnIII(μ-O)MnIV(O%) oxyl species. Thanks to the unusual tetra-dentate N-donor nature of the ancillary mcbpen ligand, the generation of the oxyl species has a moderate energy cost of 8.7 kcal mol1. The oxyl intermediate is the true active species giving access to the catalytic cycle. This may involve singlet-to-triplet spin crossover for a lower barrier, 1.8 V vs normal hydrogen electrode (NHE). The ground state of the dioxo species, [(RuIV(O))2]3+, is an antiferromagnetic "#"","#"## configuration with radical oxyl character. Both the DC and AB mechanisms were explored (Fig. 4). Interestingly, the DC mechanism involves a rather low barrier for the formation of the O–O bond, 13.9 kcal mol1 but was not considered operative due to the high barrier thereafter required to displace the peroxo ligand by an incoming molecule of water, 32.5 kcal mol1. In contrast, the AB mechanism involves a higher barrier, 47.7 kcal mol1, which is dramatically reduced to 24.5 kcal mol1 upon protonation of the oxo ligand accepting the proton released by the water nucleophile at the transition state. This Brønsted acid catalysis effect was also tested in the DC mechanism but failed by yielding even higher energy barriers. The main conclusion of this study, eg, the preference for the AB mechanism over the DC, was thereafter refuted by the group of Llobet, which reached the opposite conclusion.60 They designed a sophisticated isotopelabeling experiment aimed at identifying the most favorable reaction mechanism. Both solvent (water) and catalyst were labeled by 18O at three different degrees: c.0/12%, 16/12%, and 23/19%. The relative isotopic ratios of the oxygen evolved (18O2, 18O–16O, and 16O2) were then calculated for three different mechanistic scenarios (prevailing AB, DC, and catalyst-solvent 18O–16O exchange) and compared to those observed experimentally. For all three different degrees of 18O-labeling, the product distribution calculated for the DC mechanism yields the best fit with the experiments. Nonetheless, the values calculated for the AB and DC mechanisms were in several instances rather similar,