Low-Valent Transition Metalate Anions in Synthesis, Small Molecule Activation, and Catalysis

Abstract

This review surveys the synthesis and reactivity of low-oxidation state metalate anions of the d-block elements, with an emphasis on contributions reported between 2006 and 2022. Although the field has a long and rich history, the chemistry of transition metalate anions has been greatly enhanced in the last 15 years by the application of advanced concepts in complex synthesis and ligand design. In recent years, the potential of highly reactive metalate complexes in the fields of small molecule activation and homogeneous catalysis has become increasingly evident. Consequently, exciting applications in small molecule activation have been developed, including in catalytic transformations. This article intends to guide the reader through the fascinating world of low-valent transition metalates. The first part of the review describes the synthesis and reactivity of d-block metalates stabilized by an assortment of ligand frameworks, including carbonyls, isocyanides, alkenes and polyarenes, phosphines and phosphorus heterocycles, amides, and redox-active nitrogen-based ligands. Thereby, the reader will be familiarized with the impact of different ligand types on the physical and chemical properties of metalates. In addition, ion-pairing interactions and metal–metal bonding may have a dramatic influence on metalate structures and reactivities. The complex ramifications of these effects are examined in a separate section. The second part of the review is devoted to the reactivity of the metalates toward small inorganic molecules such as H₂, N₂, CO, CO₂, P₄ and related species. It is shown that the use of highly electron-rich and reactive metalates in small molecule activation translates into impressive catalytic properties in the hydrogenation of organic molecules and the reduction of N₂, CO, and CO₂. The results discussed in this review illustrate that the potential of transition metalate anions is increasingly being tapped for challenging catalytic processes with relevance to organic synthesis and energy conversion. Therefore, it is hoped that this review will serve as a useful resource to inspire further developments in this dynamic research field.

1. Introduction

This review article covers the chemistry of ionic complexes composed of s-block or organic cations and d-block metalate anions where the d-block metal atom is in an “uncommonly” low (sometimes even negative) oxidation state. Such metalates are highly electron-rich species, which may serve as powerful reagents in a variety of chemical transformations. Prime examples are the transition metal carbonylate anions, the first examples of which were described already a century ago.¹ In the second half of the last century, various other types of transition metalates stabilized by π-acids such as alkenes and arenes became available.²⁻¹⁰ Early reports on the behavior in fundamental reactions such as transmetalation, ligand substitution, and oxidative addition strongly suggested that metalates often are highly reactive compounds, paving the way for applications in small molecule activation and homogeneous catalysis. As early as the 1970s, the oxidative addition of alkyl halides to anionic complexes such as [Fe(CO)₄]^2– (1) or [(η⁵-Cp)Fe(CO)₂]⁻ (2, Cp = C₅H₅) had been intensively investigated.¹¹⁻¹³ However, the structural diversity and synthetic versatility of transition metalate anions have begun to be fully appreciated only recently. Due to their highly reduced nature, low-valent metalates have the potential to activate inert element–element bonds (including those of fundamentally important inorganic molecules such as N₂, CO₂, and H₂) and to mediate catalytic reactions and multielectron transformations. Indeed, the interest in applications of metalate anions in bond activation, organic synthesis, stabilization of transient species, and homogeneous catalysis is growing rapidly. The purpose of this review is to summarize the most recent developments in this vibrant field in a comprehensive manner and to familiarize interested readers with current applications of low-oxidation state metalate anions in small molecule activation and catalysis.

Since we aim to focus on the chemistry of anionic transition metal complexes in “uncommonly” low (sometimes even negative) oxidation states, the concept of metal oxidation state deserves brief comment.¹⁴ For the purpose of this review, we consider the oxidation state of any particular metal as “uncommonly low” when it is below the range of oxidation states in which most of its compounds are found.^15,16 In this case, the compounds discussed will be denoted as low-valent transition metalate anions. Naturally, this definition will depend on each element. For example, chromium complexes are frequently found in metal oxidation states of +III to +VI,¹⁷ while metal +II and +III states are typical for iron and cobalt complexes.^18,19 Furthermore, while assigning oxidation numbers to metal atoms is straightforward for most transition metal compounds, the correct assignment may not be immediately obvious in cases where redox-active ligands are present. However, even if the formal oxidation state of the metal is debatable due to charge transfer to the ligands, the overall complex will still be in a highly reduced electronic state. For this reason, we will not apply the criterion of metal oxidation state in a very strict and exclusive manner, but it will be a useful guideline for selecting the appropriate material presented for this review. Complexes with redox-active ligands will be included when a low metal oxidation state is likely to be present. However, a treatment of transition metal complexes with redox-active ligands (e.g., bipyridines, diminopyridines, and diimines) and metal atoms in the more common oxidation states is beyond the scope of this review.

The chemistry of low-valent anionic transition metal compounds has been previously reviewed by other authors.^1−3,9,20 In 2006, a seminal review by J. E. Ellis, who has been a pioneer in the field for many decades, discussed the chemistry of s-, p-, and d-block compounds in negative oxidation states.² Furthermore, Ellis summarized the state of the art of the “Chatt reaction” as a method to synthesize arene metalates, but metalates with other ligand frameworks were not discussed.³ C. G. Werncke recently dedicated a book chapter to the organometallic chemistry of main group as well as transition metal compounds with metal atoms in very low oxidation states, which gives a succinct overview of the field from its inception up to 2019.⁹ Only a few reviews describe applications of metalates in small molecule activation and homogeneous catalysis. In 2018, Gómez-Suárez and Nolan reviewed the chemistry of anionic complexes of the late transition metals, including low-valent species as well as those in higher oxidation states, with particular emphasis on their catalytic performance.²⁰ Two recent accounts summarize the chemistry of very specific transition metalate complexes, i.e., dicarbonyl cyclopentadienyl ferrates and nacnac-stabilized rhenium(I) cyclopentadienyl anion.^10,13 The chemistry of linear 3d-metal(I) compounds has been summarized by Werncke and Gómez in a recent book chapter.⁹⁰¹ Metallocene anions are the subject of another recent review.²¹ The chemistry of cobalt complexes and their application in C–H functionalization reactions, presumably involving low-valent intermediates, was reviewed in 2014.²² A recent account describes the uses of cobalt-N-heterocyclic carbene (NHC) compounds, including low-valent Co(0) species, in homogeneously catalyzed reactions,²³ and another one accounts for the efforts to achieve hydrofunctionalization and cross-coupling reactions at low-valent chromium catalysts, mostly generated in situ from air-stable Cr(II) or Cr(III) salts.¹⁷ A review article by Xu and co-workers discusses heterometallic complexes with Mg- or Zn-metalloligands.²⁴ Several reviews discuss dinitrogen fixation and functionalization, but the use of metalates is not a focal point of these articles.²⁵⁻³⁰ Considering the varying scope of the previous reviews and the impressive growth of the field in the most recent years, a comprehensive treatment of the latest developments is highly desirable. This review therefore summarizes the chemistry of low oxidation state transition metalates in the period from 2006 to 2022. Recent selected literature from 2023 is also included. For work prior to 2006, readers are referred to the previously published reviews, though earlier reports will be included for context when appropriate.^1−3,9

Upon inspection of the relevant literature, it becomes clear that most of the results in transition metalate chemistry have been achieved with first-row transition elements. While metalates are known for most of the 3d series, significant focus has been directed toward the study of the late first row metals iron, cobalt, and nickel. Since iron is the most earth-abundant transition metal, the use of iron complexes has gained considerable attention in catalysis.^18,19 Although the chemistry of the heavier second and third row metals is often distinct from their lighter first-row transition metal counterparts due to the primogenic effect,^31,32 many parallels are observed between group congeners in the chemistry of low-valent transition metalate anions. Examples for 4d- and 5d-transition metalates are therefore included throughout this review. A number of metalates are known for the lanthanides and actinides.³³⁻⁴⁷ However, their chemistry is often distinct due to the more electropositive nature of these elements, and the more core-like character of the f orbitals, which gives rise to less directional, and more electrostatic, bonding. Due to the redox-active nature of the ligands, the oxidation state of the metal cations can be debatable in some cases. Therefore, we believe that the chemistry of the f-block metalates warrants a separate treatment, which is beyond the scope of this review.

The strong influence of ion-pairing effects on the stability and reactivity of metalates has become increasingly recognized in the last 15 years. Ion-pairing is observed in a wide range of metalates (especially highly charged ones) and can have dramatic effects on their structure and reactivity. To remain within the boundaries of this review, our treatment of ion-pairing effects will be limited to metalates that likely contain “low-oxidation state” metal atoms. Although interesting and important, a systematic treatment of ion-pairing in higher valent transition-metal species, e.g., imido⁴⁸ and nitrido complexes⁴⁹ and complexes with redox-active formazanate ligands,⁵⁰ is beyond the scope of this work.

Covalent bonds between p-block and d-block metal cations or two d-block metal cations can be used to stabilize highly reactive, nucleophilic species. Therefore, the utility of metal–metal interactions in stabilizing low-valent metalate anions will be discussed in this review as well. However, it is important to note that the chemistry of metal-based Lewis pairs in general is beyond the scope of this review.⁹⁰² This type of compound will only be considered here if they are formally anionic species, i.e., when s-block or organic counteranions are present. Specifically, the chemistry of nucleophilic gold(I) complexes having “auride character” and heterobimetallic transition-metal–zinc compounds has recently been covered by other reviews and thus will only be mentioned where appropriate for context.^24,51,52 Furthermore, heterometallic complexes featuring elements other than alkali or alkaline earth metals, such as tetrel atoms or heavier transition metals (4d, 5d), are not covered in this review even though such complexes may display transition metal atoms in formally negative oxidation states, e.g., Fe(−I), Co(−I), or Rh(−I).⁵³⁻⁵⁸

The review article is divided into four parts. Following the present section, section 2 summarizes the different methods for the preparation of metalates to familiarize readers with the synthetic protocols, the commonly used ligand types, and basic reactivity patterns. These highly reactive, electron-rich complexes require a suitable, stabilizing ligand framework, normally consisting of strong π-backbonding units. One of the most common ligand types is carbonyl. Numerous studies have proven that inter alia isocyanide, arene and alkene, phosphane, phosphite, carbene, alkyl and amido complex, and imine ligand frameworks also provide sufficient stabilization, allowing for comprehensive investigations of the reactivity and applications of low-valent transition metalate complexes. Accordingly, section 2 covers the chemistry of mononuclear transition metalate complexes according to ligand type. In addition, the importance of ion-pairing interactions and metal–metal bonding in the synthesis and reactivity of metalates is highlighted in a separate subsection. Section 3 summarizes the state-of-the-art in small molecule activation by metalate complexes, including catalytic transformations, focusing on the activation of dihydrogen, dinitrogen, carbon monoxide, carbon dioxide, white phosphorus, and other related molecules and model compounds. Section 4 presents a summary of the reviewed results and some perspectives for the future development of the field.

2. Synthesis and Basic Reactivity Patterns of d-Block Metalates

Documented examples of carbonyl metalates date back to the early 1930s, when initial reports described the synthesis of [Fe(CO)₄]^2– (1, Figure 1),⁵⁹ although its structure was not confirmed until the 1950s.¹ Since then, reported carbonyl metalates include almost every metal from groups 4–10 with varying charge and nuclearity. Additional examples, showcased in Figure 1, are [V(CO)₆]⁻ (3) and [Cr(CO)₅]^2– (4).^1,2 Isocyanide ligands can be used as tunable CO surrogates thanks to their R substituent. The first isocyano metalate, [Co(CNXyl)₄]⁻ (5, Xyl = 2,6-Me₂C₆H₃; see Figure 1), was reported in 1989 by the Cooper group.^60,61 Furthermore, alkene and arene metalates constitute one of the most versatile groups of low-valent metalates.^2,3 Within this type of compound, selected examples are the arene titanate K[Ti(benzene)₂] (6), the manganate [Li(dme)]₂[Mn(cod)₂] (7) (see Figure 1), and the nickelate [Ni(C₂H₄)₃]^2– (8), which are homoleptic compounds with the metal in formally negative oxidation states.^7,62−65 By contrast, ferrate 2 (Figure 1) is heteroleptic and formally zerovalent.¹¹⁻¹³

Figure 1.

Ligands commonly used for stabilizing low-valent (uncommon) oxidation states in transition metal complexes.

2.1. Synthetic Routes

The methods for the synthesis of highly reduced d-block metalates can be summarized into the main types of reactions shown in Scheme 1 for selected early examples of such compounds.^{7,8,62−64,66−68} Most metalate syntheses reported since 2007 generally follow one of these methods.³ Jonas and co-workers pioneered the reductive cleavage of metallocenes with lithium in the presence of an olefin as a strategy to access highly reactive alkene metalates (Scheme 1, reaction type 1a).^7,8,62,68,69 Examples of complexes obtained using such strategy are [Li₂(thf)₄{Ni(η²-cod)(η⁴-cod)}] (9) or the alkenemanganate 7, shown in Scheme 1.^7,65 The use of other alkali metals (sodium, potassium, or their Hg amalgams) to reduce a transition metal salt or precursor is also a common synthetic route (see Scheme 1, reaction type 1b),⁷⁰ with examples including the reduction of a Co(II) precursor with sodium to afford [(N₂)Co(PEt₂Ph)₃]⁻ (10), or the formation of bis(arene)titanates(−I), K[Ti(arene)₂] [arene = benzene (6, Figure 1, vide supra), toluene (11)] from their Ti(0) parent compounds.^63,64,66 Reducing agents such as KC₈⁷¹⁻⁷⁹ and cobaltocene (Cp₂Co)⁸⁰ can also be employed. The strategy used by Jonas is certainly a turning point in the synthesis of arene and alkene metalates.^900,65,82

Scheme 1. General Methods for the Synthesis of (Highly Reduced) Anionic Complexes.

The reaction of alkali metal arene radical anions with transition metal (TM) salts is another powerful route to access highly reduced metalate complexes (Scheme 1c). This method was discovered in the early 1960s by J. Chatt as a route to hitherto unknown transition metal compounds and later expanded to enable the synthesis of a wide range of arene metalates.^83,84 To honor Chatt’s groundbreaking work on the interaction of transition metal compounds with alkali metal arenides, Ellis recently proposed naming such methodology “the Chatt reaction”.³ Examples of metalates obtained by this pathway are two titanates analogous to the ones reported by Green and co-workers, 12 (arene = 1,1′-biphenyl) and 13 (arene = 4,4′-ditert-butyl-1,1′-biphenyl; Scheme 1, reaction type 1c),⁶³ and the first homoleptic polyarene cobaltate, bis(1,2,3,4-η⁴-anthracene)cobaltate(−I) (14), usually regarded as a source of the “naked” atomic cobalt anion, Co^–.^64,67 Finally, new metalate compounds have been synthesized by displacement/exchange of labile alkene/arene ligands, in complexes such as [Li₂(TMEDA)₂{Fe(η²-C₂H₄)₄}] (15, TMEDA = 1,2-bis(dimethylamino)ethane Scheme 1, reaction type 2). This example by the Jonas group shows the synthesis of [Fe(cod)₂]^2– (16) as two different lithium salts.⁸ Thus, metalates can be accessed either by reduction of a high-valent compound or using precursors already featuring a low-valent metal center.^2,3

The selected early examples of metalates shown in Scheme 1 share similar features. In most cases, they are synthesized via reduction of transition metal salts/precursors or metallocene compounds with alkali metals (e.g., lithium, sodium, potassium), and the metal centers are stabilized by π-acid ligands, capable of accepting the excess electron density on the metal atom through synergistic π-backbonding. Therefore, the presence of π-acceptor ligands, such as carbonyls, isocyanides (isolobal with the carbonyl ligand) and unsaturated hydrocarbons (arenes, alkenes), is common (see Figure 1).

Due to the lability of the hydrocarbon ligands, alkene and arene metalates are often referred to as “naked-metal atom” reagents, or storable sources of transition metal anions. Nonetheless, Ellis and co-workers have emphasized that the use of poly(arenemetalates) as sources of “naked” atomic metal anions, due to their highly reduced nature, might be limited to the compatibility of the desired products with redox processes.⁸⁵ With a few exceptions, carbonyl-, isocyanide-, alkene-, and arene-metalates obey the 18-electron rule. Other metalates have been stabilized by carbenes, redox-active ligands, or phosphorus ligands (Figure 1), and examples of these will be discussed throughout the text.

In many cases, the successful isolation of a complex with the metal in a formally negative oxidation state has been enabled by a strong interaction with the countercation, generally obtained from the reducing agent. Consequently, a breadth of examples of carbonyl, alkene/arene or isocyanide metalates have been synthesized and crystallized as alkali metal salts, stabilized with coordinating solvents (e.g., THF or DME; see Scheme 1 for examples by Jonas and co-workers^7,8,62), encapsulating agents ([18]crown-6, [2.2.2]cryptand), or by further exchange of the metal cations with organic cations [e.g., PPN⁺ (PPN⁺ = [Ph₃PNPPh₃]⁺), NBu₄⁺ or NMe₄⁺].

2.2. Survey of Metalates Reported since 2006: The Crucial Influence of Ligands on Complex Structure and Reactivity

The following subsections describe the efforts made in the period from 2006–2022 to synthesize and understand the reactivity of low-valent anionic d-block metal complexes. The material is organized according to ligand type. The types of ligands capable of stabilizing d-block metalates have already been summarized in Figure 1. The chemistry of neutral or cationic species or complexes with metalloradical character goes beyond the scope of the present work and will not be included unless examples of these are related to low-valent anionic complexes.

2.2.1. Complexes Containing Carbonyl Ligands

The carbonyl ligand (CO) is among the most studied types of ligands in organometallic chemistry.⁸⁶ Upon coordination to a metal center, carbon monoxide behaves as a weak σ-donor and as a strong π-acceptor (strong π-acid). In fact, carbonyl ligands are very effective in stabilizing highly reduced metal centers and metals in uncommon oxidation states since they can accept the excess electron density through M–CO π-backbonding (Figure 2).^1,2,9

Figure 2.

Bonding interactions in carbonyl or isocyanide transition metal complexes.

Consequently, homoleptic metal–carbonyl compounds featuring metal centers in formally negative oxidation states have been known for many decades.¹ To the best of our knowledge, the lowest formal oxidation state reported so far has been M(−IV), achieved in the series of homoleptic carbonyl metalates of the group 6 metals (Cr, Mo, W).⁸⁷

Isocyanide ligands (C≡NR) are isolobal with CO but are weaker π-acids and stronger σ-donors than carbon monoxide.^88,89 While it has long been recognized that isocyanide ligands can be used as surrogates of carbonyl ligands, the former can undergo electronic and steric modulation in a way unavailable to CO.⁸⁸

In many cases, the oxidation state of the metal atom in low-valent compounds featuring carbonyl or carbonyl-like ligands is ambiguous. For example, the description of the electronic structure of the NO ligand-ferrate compound [Fe(CO)₃(NO)]⁻ (17) has been subject to considerable debate. Several groups have studied its spectroscopic characteristics, comparing its behavior to that of known [Fe(CO)₄]^2– (1).^90,91 Despite being isoelectronic with complex 1, which is widely accepted to be a Fe(−II) complex, there is no clear agreement on either the oxidation state of the Fe center or, consequently, on the charge/electronic description of the NO ligand in 17. In their independent studies, the group of Plietker⁹¹ and Gruden and Zlatar⁹⁰ concur that 17 cannot be described as a Fe(−II)/NO⁺ species, and that, accordingly, it can be more accurately regarded as an Fe(0) species featuring a covalently bonded anionic (NO^–) ligand. The spectroscopic evidence (⁵⁷Fe Mössbauer, EXAFS and XES spectroscopy) and the theoretical calculations (DFT, TDDFT, and CASSCF) are not only in agreement⁹¹ but consistent with the structural characterization (X-ray and FT-IR).^92,93 Based on their calculations, Plietker and co-workers suggested that in the ground-state the complex features an Fe(0) atom, and that the Fe–NO bond consists of two covalent π-bonds to an NO^– moiety.⁹¹ Gruden and Zlatar performed a NOCV (natural orbitals for chemical valence) analysis on [Fe(CO)₃]^2– and NO⁺ fragments, finding large charge flow from the Fe center to the NO⁺ ligand, which was held as additional evidence that the iron center cannot be in an Fe(−II) formal oxidation state.⁹⁰ Furthermore, a comparison of this analysis with that of the analogue 1 seems to confirm the disparate electronic configurations of the iron centers in each species.

The results by Plietker⁹¹ and Gruden and Zlatar⁹⁰ contrast with a study by Bauer and co-workers.⁹⁴ The latter work investigated the bonding in 17 via a combined experimental and theoretical approach using valence-to-core X-ray emission spectroscopy (VtCXES), X-ray absorption near-edge structure spectroscopy (XANES), and high-energy-resolution fluorescence-detected XANES (HERFD-XANES) in combination with high-level quantum chemical calculations. The Fe(0) complex Fe(CO)₅ (18) was used as a model of a 3d⁸ metal, which would represent the proposal by Plietker.⁹¹ According to the spectroscopic and theoretical results, the electronic structures of [Fe(CO)₄]^2– (1) and [Fe(CO)₃(NO)]⁻ (17) appear to be very similar, which would imply a formal Fe(−II) oxidation state in the nitrosyl compound 17.⁹⁴ It should be noted though that complex 18 might not be an ideal reference point due to the simultaneous changes in oxidation state, coordination geometry, and coordination number in comparison with 1 and 17. Thus, the electronic situation of [Fe(CO)₃(NO)]⁻ (17) is still debatable. Nonetheless, the contrasting behavior in catalytic reactions exhibited by compounds 1 and 17 could well be the result of changes in the electronic characteristics of the metal center and the ligand. While 1 was inactive in most of the studied transformations, 17 showed activity in allylic substitutions, hydrosilylations, transesterifications, and carbene-transfer reactions.⁹⁵⁻⁹⁹

A series of AFe(CO)₃^– (A = B, Ge, Sn, Pb, Sc, Y, La) anions and the nature of the formed A–Fe bonds were investigated by means of experimental and theoretical studies by Zhou, Li, and co-workers.^100−102 In all cases, the anionic compounds were generated experimentally in the gas phase and characterized by mass-selected infrared photodissociation spectroscopy. Quantum chemical calculations assisted in the identification and analysis of the electronic structure of the complexes. For instance, the BFe(CO)₃^– (19) anion was found to have a cylindrical C_3v symmetric structure.¹⁰⁰ Geometry optimizations (DFT) indicated a very short equilibrium B–Fe bond distance of 1.63 Å, which is considerably shorter than the sum of the triple-bond covalent radii of iron and boron atoms (1.75 Å).¹⁰³ The latter observation suggests that the B–Fe bond order should be higher than three. Electronic structure and (quantum) chemical bonding analyses support the existence of a quadruple B≣Fe bonding interaction, composed of one electron-sharing σ bond, two Fe → B dative π bonds, and a weak B → Fe dative σ bonding interaction. DFT calculations predict a Fe–B stretching frequency of very low IR intensity for 19, absorbing at 863 cm^–1, that could not be directly observed.¹⁰¹ Similar studies on AFe(CO)₃^– (A = Ge, Sn, Pb) indicated that the complexes exist in a ²A₁ doublet electronic ground state and present an A≡Fe triply bonded C_3v structure. The carbonyl ligands are all coordinated to the iron center. Bonding analyses indicated that the valence np atomic orbitals of the group 14 atoms and the hybridized 3d and 4p atomic orbital of the iron center are the main contributors to the triple bonding interactions.¹⁰¹ Analogous to compound 19, and contrasting with the heavier group 14 congeners, compounds AFe(CO)₃^– (A = Sc, Y, La) were found to feature quadruple A–Fe metal–metal bonding interactions based on combined experimental and quantum chemical investigations.¹⁰²

Similarly, the heteronuclear magnesium–iron carbonyl anionic complexes MgFe(CO)₄^– (20) and Mg₂Fe(CO)₄^– (21) were generated in the gas phase and their structures studied using a combination of mass-selected infrared photodissociation spectroscopy and quantum chemical calculations.¹⁰⁴ According to this study, the most stable structures of compounds 20 and 21 have a ²A₁ ground state and C_3v symmetry, presenting heteronuclear Mg–Fe or Mg–Mg–Fe bonds, respectively. Theoretical calculations indicate, however, that despite having some covalent character, the interactions between magnesium and iron have a significant electrostatic component. For both complexes, 20 and 21, population analyses revealed that the negative charge is mainly located at the Fe(CO)₄ unit. In 20, analysis of the bonding situation suggested the existence of a Mg–Fe σ-bond. In turn, in complex 21 the central Mg atom has a partial positive charge, and the terminal Mg atom is neutral, and the interaction was described as a σ-type electron-sharing bonding. The valence of the metal atoms was also analyzed, finding that in the σ-bonded complexes the atoms are formally in the Mg(I)–Fe(−II) oxidation states for 20, whereas in 21 the covalent σ-bond is relatively weak, and involves Mg(0)–Mg(I)–Fe(−II) formal oxidation states.¹⁰⁴ This group extended their investigations to heteronuclear magnesium–iron carbonyl cationic complexes, MgFe(CO)_n⁺ (n = 4–9) finding that for n > 5 the valence of the metal centers involved can formally be described as featuring Mg(II)–Fe(−I).¹⁰⁵

By reaction of the radical anion salt of fullerene [K([2.2.2]cryptand)][C₆₀^•–] with [Fe₃(CO)₁₂] (22), the Konarev group obtained the anionic iron-bridged fullerene C₆₀ dimer [K([2.2.2]cryptand)]₂[Fe(CO)₂-μ,η²:η²-C₆₀]₂·2.5C₆H₄Cl₂ (23, see Figure 3).¹⁰⁶ Each iron atom coordinates to two C–C bonds of C₆₀ hexagons in the two fullerene units, and to two carbonyl ligands. The coordination environment of the metal centers is completed by an Fe–Fe interaction (2.978(4) Å). In the solid state, the dianion presents two [K([2.2.2]cryptand)]⁺ cations interacting with carbonyl ligands. The combined experimental and DFT data indicated that the diamagnetic dimer 23 is in a singlet ground state and that the negative charges are localized both on the iron atoms and the fullerene cages. The reduced state of the metal centers is reflected by the shift of the IR stretching bands to lower frequencies (1880, 1902, 1920, and 1958 cm^–1) with respect to the parent cluster 22 (1909, 1991, and 2051 cm^–1). A similar anionic cobalt dimer [nBu₄N][{Co(Ph₃P)}₂(μ-Cl)(μ,η²:η²-C₆₀)₂] (24) was obtained by the same group.¹⁰⁷ The latter species, however, features phosphine ligands coordinated to the cobalt atoms instead of carbonyls and has a bridging chloride atom linking both metal centers.

Figure 3.

Representation of the molecular structure of 23, according to X-ray crystallographic analysis.

Zhou and Frenking studied the species ANi(CO)₃^– (25, A = Li, Na, K, Rb, Cs) using mass-selected infrared photodissociation spectroscopy in the gas phase, like with the previously discussed iron compounds AFe(CO)₃^– (19).^{100−102,108} In 25, the alkali metal is covalently bound to the anionic nickel–carbonyl fragment, interacting with the metals and with the carbon atoms of the carbonyl ligands. The nickel center is in a nearly planar C_3v geometry, with the alkali metal located above this plane.¹⁰⁸

In turn, the tricarbonyl nickelate [Ni(CO)₃]^2– (26) was identified, and structurally characterized, as the anionic part of the carbonyl nickel salt [A([18]crown-6)]₂[Ni(CO)₃]·8NH₃ (A = K, Rb).¹⁰⁹ [Ni(CO)₃]^2– completes the series of known first-row carbonyl metalates. A previous attempt to obtain a similar compound via reduction of Ni(CO)₄ (27) with alkali metals had led only to the synthesis of two different nickelates: [Ni₂(CO)₆]^2– (28) in pyridine or H₂Ni₂(CO)₆ [H₂(28)] in liquid ammonia.¹¹⁰ The serendipitous discovery of anion 26 came after an attempt to functionalize germanium clusters with the Ni(0) compound (PPh₃)₂Ni(CO)₂ (29) in the presence of a crown ether and liquid ammonia. Single crystal X-ray structural analysis of the thermally unstable, air- and moisture sensitive crystalline material revealed that the nickel center is surrounded by the carbonyl ligands in an almost perfect trigonal planar geometry, with a D_3h point group symmetry (see Figure 4). The anion interacts with a crown ether-stabilized potassium cation and weakly with three ammonia molecules in the crystal lattice. Theoretical calculations supported X-ray diffraction results, thus ruling out the C_3v nonplanar structure previously predicted for 26 almost 40 years earlier.^109,111

Figure 4.

Representation of the structure of [K([18]crown-6)]₂[26]·8NH₃, according to X-ray crystallographic analysis.¹⁰⁹

2.2.2. Cyanido Complexes

Contrasting with the wide variety of highly reduced carbonyl metalates, cyanido compounds with metal centers in formally negative oxidation states have rarely been reported.¹¹² Exceptional examples of highly reduced complexes at group 8 and 9 transition metals are [Fe(CN)₃]^7– (30), [Fe(CN)₄]^6– (31), [Ru(CN)₃]^7– (32), and [M(CN)₃]^6– [M = Co (33), Rh (34), Ir (35)], which were reported by Jach and co-workers.^112−117 Unlike the cases summarized in Scheme 1, these cyanido complexes are synthesized by solid-state routes in which pelletized mixtures of the corresponding transition metal, an alkaline earth metal subnitride, graphite, and either nitride or cyanide salts (as a nitrogen source) react under argon at temperatures higher than 1020 K. The tricyanidometalates 30 and 32–35 were isolated as alkaline earth metal salts, AE_3.5[M(CN)₃] (for M = Fe, Ru, AE = Sr, Ba) and AE₃[M(CN)₃] (for M = Co, AE = Sr, Ba; for M = Rh, Ir, AE = Ba), respectively. Complex 30 also formed as the mixed salt LiSr₃[Fe(CN)₃], whereas ferrate 31 was isolated as (Sr₃N)₂[Fe(CN)₄]. In all cases, the ligands exhibited elongated C–N bonds (1.20 to 1.28 Å) with respect to the classical cyanido ligand (1.15 Å, on average), and significantly lower ν_CN stretching frequencies in IR, in the range 1490–1688 cm^–1 (vs 2000–2100 cm^–1 in classical CN^– ligands).^112,118 These values indicate strong π-backbonding interactions with the metal atoms. Physical characterization and quantum chemical calculations indicate that the reduced cyanidometalates are best described as [M^2–(CN^1.67–)₃]^7– (M = Fe, Ru) and [M^–(CN^1.67–)₃]^6– (M = Co, Rh, Ir) instead of the formal [M^4–(CN^–)₃]^7– or [M^3–(CN^–)₃]^6–, respectively, with the transition metal centers in d¹⁰s⁰ configurations (18 valence electrons, VE). In turn, complex [Fe(CN)₄]^6– (31) was also regarded as featuring an Fe(−II) center in a d¹⁰s⁰ configuration (18 VE). Given the similarities between the isoelectronic cyanido and carbonyl complexes, e.g., in the pair [Fe(CN)₄]^6– (31) and [Fe(CO)₄]^2– (1), or [Co(CN)₃]^6– (33) and [Co(CO)₃]^3– (36),^119,120 the authors anticipated that the isolation of further examples of highly reduced cyanido species analogous to known carbonyl metalates might be feasible.^112−117

2.2.3. Isocyanide Complexes

2.2.3.1. Manganese Isocyanide Complexes

Figueroa and co-workers have contributed to the chemistry of 3d manganese,^121,122 iron,¹²³ and cobalt^124,125 metalates. In 2011, mixed carbonyl/isocyanide manganese monoanions were reported by this group.¹²¹ Since isocyanides are both weaker π-acceptors and stronger σ-donors than CO, the synthesis of mixed carbonyl/isocyano complexes by substitution with isocyanides has been used as a strategy to increase the electron density, and therefore nucleophilicity, of carbonyl-containing complexes.¹²⁶ In an attempt to combine the steric and electronic tunability of isocyanide ligands with the reactivity pattern offered by the known [Mn(CO)₅]⁻ (37), the group of Figueroa synthesized two examples of mixed complexes, namely [Mn(CO)₂(CNAr^Mes2)₃]⁻ (38) and [Mn(CO)₃(CNAr^Dipp2)₂]⁻ [39, synthesized from BrMn(CO)₃(CNAr^Dipp2)₂ (40) see Scheme 2; Mes = 2,4,6-Me₃C₆H₂, Dipp = 2,6-(iPr)₂C₆H₃]. The synthesis of trigonal bipyramidal complex 38 proved challenging. Only after sequential treatment of the isomeric mixture of BrMn(CO)₂(CNAr^Mes2)₃ (41) with potassium anthracenide (KC₁₄H₁₀) and [18]crown-6 was it possible to isolate and characterize complex 38 in a 20% yield. This compound is stable at −35 °C in the solid state but decomposes rapidly at room temperature in solution (over the course of 4 h in C₆D₆). Consequently, its reactivity could not be conveniently assessed. By contrast, with the more sterically demanding ligand CNAr^Dipp2, complex 39 is stable in C₆D₆ solution for days. IR stretching frequencies ν_CO and ν_CN for both complexes are listed on Table 1. The low-frequency ν_CO [38: 1841 cm^–1; 39: 1896 and 1773 cm^–1] and ν_CN [38: 1891 cm^–1 vs free CNAr^Mes2 = 2118 cm^–1;¹²⁷39: 1910 cm^–1 vs free CNAr^Dipp2, C₆D₆ solution = 2118 cm^–1]¹²⁸ stretches observed by FT-IR reflect the significant π-backdonation from the reduced manganese centers.

Scheme 2. Synthesis of Mixed Carbonyl/Isocyanide Mn-Metalates by the Group of Figueroa¹²¹.

Table 1. ν_CO and ν_CN Stretching Frequencies (cm^–1) for Selected Metalates Bearing Carbonyl or Isocyanide Ligands.

Complex	νCO (cm–1)	νCN (cm–1)	Ref
[K([18]crown-6)(dme)][Mn(CO)2(CNArMes2)3]	1841 (s, vb)a	1891	(121)
Na[Mn(CO)3(CNArDipp2)2]	1896, 1773	1910	(121)
K2[Fe(CNXyl)4]	-	1670	(131)
Na2[Fe(CNArMes)4]	-	1682	(123)
Na[HFe(CNArMes2)4]	-	1994–1828	(123)
K2[Fe(CO)2(CNArTripp2)2]	1878 (w), 1793 (m)	1599 (vw), 1562 (s)	(71)
Fe(CO)3(CNArTripp2)2	1940	2146 (vw), 2092 (vs)	(71)
Fe(BF)(CO)2(CNArTripp2)2	1980 (s), 1942 (vs)	2126 (vw sh), 2056 (s)	(71)
Fe(N2)(CO)2(CNArTripp2)2	1953 (m), 1918 (s)	2122 (vw), 2075 (vs)	(71)
[K(Et2O)]2[Fe(CNAr3NC)2]2	-	1982 (w) and 1961 (w)	(72)
[K([18]crown-6)(dme)][Co{1,2-(PMes)2C2B10H10}(CNCy)2]	-	2029 and 1938	(132)
[K([18]crown-6)(THF)2][Co(CNtBu)4]	-	1778 (vs br)	(85)
Na[Co(CNArMes2)4]	-	1903, 1821, and 1761	(124)
[PPN][Co(CNArMes2)4]	-	1821	(124)
(η2-PPN)Co(CNArMes2)3	-	1952, 1860, and 1824	(125)
Na[Co(CO)(CNArMes2)3]	1830	1904, 1864	(133)
Na[Co(CO)2(CNArMes2)2]	1852	1891	(133)
Na[Co(CO)3(CNArMes2)]	1870 (vs), 1846 (s)	1932	(133)
[(μ-CNArMes2)2{CpCo}2]	-	1834	(134)
K[(μ-CNArMes2)2{CpCo}2]	-	1666	(134)
K2[(μ-CNArMes2)2{CpCo}2]	-	1511	(134)
[{K(Et2O)}2{Cp*Co≡CNArTripp2}]	-	1509	(135)
[K(Et2O)][Cp*Co(H)(CNArTripp2)]	-	1710	(135)
[K(Et2O)][Cp*Co(SiMe3)(CNArTripp2)]	-	1707	(135)
Ph2B(tBuIm)2Fe(CO)3	1987, 1900	-	(73)
[K([2.2.2]cryptand)][Ph2B(tBuIm)2Fe(CO)3]	1926, 1836, 1800	-	(73)
K2[Ph2B(tBuIm)2Fe(CO)2]2	1838, 1742	-	(73)
[PhB(MesIm)3Fe(CO)3][B(C6F5)4]	2100, 2040	-	(74)
PhB(MesIm)3Fe(CO)2	1956, 1886	-	(74)
K[PhB(MesIm)3Fe(CO)2]	1812, 1728	-	(74)
K[Fe(PCcarbeneP)(CO)2]	1826, 1767	-	(77)
[Na([15]crown-5)][(DippPDI)Fe(CO)2]	1935, 1863	-	(136)
[(Dippnacnac)Ca(μ-OC)2Fe(η5-Cp)(thf)2]2	1823, 1780	-	(137)
[(Dippnacnac)MgFe(η5-Cp)(CO)2(thf)]	1926, 1857	-	(138)
[(Dippnacnac)Mg{(NTol)2CFe(η5-Cp)(CO)2}]	2010, 1956	-	(138)
[(P3B)Fe(CO)]	1857	-	(139)
[(PPh3Si)Fe(CO)]	1881	-	(140)
[(PiPr3Si)Fe(CO)]	1850	-	(141)
[(PiPr3Si)Fe{CONa(thf)3}]	1717	-	(141)
[Na([12]crown-4)2][(PPh3Si)Fe(CO)]	1757	-	(141)

^aν_CO + ν_CN.

Furthermore, the reactivity of 39 differs notably from that of the related complex [Mn(CNXyl)₅]⁻ (Xyl = 2,6-Me₂C₆H₃; 42).¹²⁹ Cooper and co-workers found that 42 is thermally unstable and undergoes multiple insertion processes. The reaction of 42 with alkyl electrophiles led to an alkylation-induced coupling of two isocyanide ligands to form a structurally characterized 1,4-diazabutadien-2-yl complex of manganese.¹³⁰ Contrastingly, compound 39 reacted with HCl, MeI, or heavier main-group electrophiles such as MeSiCl₃ or SnCl₂ to afford the series of products mer,trans-RMn(CO)₃(CNAr^Dipp2)₂ [R = H- (43), Me- (44), Cl₂MeSi- (45), ClSn- (46), Scheme 3], where the electrophile binds to the metal center. Remarkably, for the hydride and methyl derivatives 43 and 44, the complexes behaved similarly to the carbonyl analogue RMn(CO)₅ (47), mimicking the chemistry of 47 while benefiting from the steric protection provided by the isocyanide ligand.¹²¹

Scheme 3. Reactivity of Na[Mn(CO)₃(CNAr^Dipp2)₂] (Na[39]) by Figueroa and Co-workers^121,122.

Later, Figueroa and co-workers reported that the reaction of complex 39 with the triflate Mn(I) complex [Mn(OTf)(CO)₃(CNAr^Dipp2)₂] (48) affords the isolable Mn(0) monoradical [Mn(CO)₃(CNAr^Dipp2)₂] (49) in high yield (92%) via comproportionation. The same species is also accessible, albeit in a less convenient synthetic pathway, through oxidation of 39 with thallium triflate (TlOTf; 1.0 equiv), generating metallic thallium and NaOTf as byproducts.¹²²

2.2.3.2. Isocyanide Ferrates

The groups of Ellis¹³¹ and Figueroa¹²³ independently reported on the synthesis and reactivity of iron-isocyano metalates (Scheme 4). The dianionic complexes [Fe(CNR)₄]^2– [CNR = CNXyl (50), CNAr^Mes2 (51)] are the isocyanide analogues of the first known carbonylmetalate, [Fe(CO)₄]^2– (1).^1,2,11

Scheme 4. Synthesis of the Iron-Isocyano Metalates [Fe(CNR)₄]^2– (50, 51) by the Groups of (a) Ellis¹³¹ and (b) Figueroa¹²³.

Despite being a successful strategy for the isolation of metalates, the direct treatment of the labile anthracene-Fe(−I) complex [Fe(η⁴-C₁₄H₁₀)₂]⁻ (52, vide infra) with CNXyl did not afford homoleptic isocyanide-iron complexes.¹⁴² Instead, compound 50 was obtained after reduction of iron(II) bromide by potassium naphthalenide at low temperatures, in the presence of CNXyl (4 equiv) (Scheme 4a).¹³¹ The solid-state molecular structures of the Fe(II)-isocyanide complex FeBr₂(CNXyl)₄ (53), the zerovalent Fe(CNXyl)₅ (54), and a cocrystallized form containing 54 and the dinuclear [Fe(CNXyl)₃]₂(μ₂-CNXyl)₃ (55) were later reported.¹⁴² For metalate 50, both the FT-IR stretching bands (ν_CN = 1670 cm^–1; see Table 1) and the C–N bond length (1.237(7) Å) indicate a strong metal-to-ligand backbonding, as a consequence of the electron-rich character of the Fe(−II) center.¹³¹

The reaction of 50 with Ph₃SnCl (2.0 equiv) furnishes the air-stable complex trans-[Fe(CNXyl)₄(SnPh₃)₂] (56) (see Scheme 5). The CNtBu analogue trans-[Fe(CNtBu)₄(SnPh₃)₂] (57) was also isolated, via the proposed intermediate [Fe(CNtBu)₄]^2– (58), which was not isolated due to thermal instability. The formation of complexes 56 and 57 contrasts with the behavior previously observed for carbonylmetalate 1, for which similar reactions afford a series of compounds, among which is cis-[Fe(CO)₄(SnPh₃)₂].¹⁴³ The trans arrangement instead observed in 56 and 57 was attributed to the poorer acceptor properties and higher steric hindrance of isocyanides compared to CO ligands.^89,131

Scheme 5. Reactivity of [Fe(CNR)₄]^2– [CNR = CNXyl (50), CNAr^Mes2 (51), CNtBu (58)]^123,131.

The bulkier iron-isocyano metalate 51, reported by Figueroa and co-workers,¹²³ was obtained via reduction of FeCl₂ with sodium amalgam in the presence of the isocyano ligand CNAr^Mes2 (Scheme 4b). The complex formed as an unsolvated contact ion triple with the sodium cations, as confirmed via crystallographic characterization. Again, the FT-IR and ¹³C NMR data obtained are consistent with significant metal-to-ligand backbonding, and the minimal structural differences observed are attributed to the steric constraint exerted by the bulkier isocyanide ligands. Attempts to disrupt the ion-pairing with [18]crown-6 resulted in decomposition of the ferrate Na₂[51] to the formally Fe(I)-1-azabenz[b]azulene product [Na([18]crown-6)][(η⁵-Me₆-1-azabenz[b]azulene)Fe(CNAr^Mes2)₂] (59), seemingly obtained after an aza-Büchner ring expansion of an isocyanide CNAr^Mes2 ligand (Scheme 5).

Metalate 51 reacts with trimethylsilanol (Me₃SiOH) to yield the protonated product Na[HFe(CNAr^Mes2)₄] (Na[60], Scheme 5), in behavior similar to that of the tetracarbonyl analogue 1.¹²³ Different reactivity was observed upon treating 51 with stronger Brønsted acids (3,5-dimethylbenzoic acid, pivalic acid, pyridinium chloride, or ethylammonium chloride) at low temperatures. These reactions yielded only complex mixtures of free ligand, unreacted material, and unidentified species. The authors attributed such behavior to the presumable formation of the doubly protonated product, H₂Fe(CNAr^Mes2)₄, which would rapidly decompose under the studied conditions. This species could not be accessed from 60 and additional equivalents of Me₃SiOH, which would indicate that 60 displays only moderate Brønsted basicity. However, the characterization of complex Na[60], in particular its higher energy stretching bands in the FT-IR spectrum (ν_CN = 1994–1828 cm^–1; see Table 1), evidenced that the iron center in the complex remained highly reduced and would be expected to react with strong electrophiles.

Therefore, 60 was treated with MeOTf at low temperatures (Scheme 5), affording methane and the zerovalent product Fe(N₂)(CNAr^Mes2)₄ (61), which could be regarded as a trapped form of the tetraisocyano complex Fe(CNAr^Mes2)₄. The latter is analogous to the elusive Fe(CO)₄, which has been photochemically generated in matrix isolation experiments and characterized by infrared spectroscopy.^144−146 Complex 61, however, decomposed completely in solution at room temperature in a relatively short time frame (4 h), through an intramolecular ligand C–H activation mechanism.¹²³ Such decomposition complicated the study of additional reactivity patterns for 61. Nevertheless, this work highlighted the parallel in reactivity between carbonyl and isocyano metalate complexes.

The mixed isocyano/carbonyl metalates K₂[Fe(CO)₂(CNAr^Tripp2)₂] [62, Ar^Tripp2 = 2,6-(2,4,6-(iPr)₃C₆H₂)₂C₆H₃] and K₂[Fe(CO)₂(CNAr^Dipp2)₂] [63, Ar^Dipp2 = 2,6-(2,6-(iPr)₂C₆H₃)], analogous to 50,¹²³ were also reported by the Figueroa group.^71,147 In a multistep synthetic protocol, pentacarbonyliron(0) is photolytically transformed, in the presence of the more sterically encumbered ligand CNAr^Tripp2 (2.2 equiv), to the iron(0) complex Fe(CO)₃(CNAr^Tripp2)₂ (64), which is subsequently oxidized by molecular iodine (1.0 equiv) to afford [Fe(I)₂(CO)₂(CNAr^Tripp2)₂] (65). The latter complex reacted with KC₈ (4.5 equiv) to yield the formally Fe(−II) mixed isocyano/carbonyl complex 62 (Scheme 6).⁷¹ Likewise, [Fe(I)₂(CO)₂(CNAr^Dipp2)₂] (66) was reduced with KC₈ (4.5 equiv) to obtain the analogue K₂[Fe(CO)₂(CNAr^Dipp2)₂] (63).¹⁴⁸

Scheme 6. Synthesis of Mixed Carbonyl/Isocyano Fe(−II) Metalates by Figueroa and Co-workers^71,147,148.

Treatment of complex 62 with 2.8 equiv of BF₃·Et₂O (Scheme 7) afforded the unprecedented terminal fluoroborylene-iron complex, Fe(BF)(CO)₂(CNAr^Tripp2)₂ (67). The same product was obtained at lower ratios of 62:BF₃·Et₂O (1:1 and 1:2), although in lower yields/conversions and/or with the formation of the tetrafluoroborate ([BF₄]⁻) ion as a byproduct. Based on such observations, the authors suggested that the transformation of complex 62 into 67 proceeds via an unobserved anionic irondifluoroboryl intermediate, presumably [Fe(BF₂)(CO)₂(CNAr^Tripp2)₂]⁻ (68), and that the loss of the second fluoride is promoted by the excess of BF₃. Compound 67 was the first crystallographically characterized terminal fluoroborylene complex. Its kinetic stability is sufficient to allow isolation at room temperature.⁷¹ The structural and spectroscopic characterization of 67 indicates that the terminal fluoroborylene ligand displays characteristics of both a strong σ-donor and π-acceptor.

Scheme 7. Reactivity of the Mixed Complexes K₂[Fe(CO)₂(CNAr^Tripp2)₂] (62) and K₂[Fe(CO)₂(CNAr^Dipp2)₂] (63)^71,147,148.

As with [Fe(CNXyl)₄]^2– (50), ferrate 62 can be oxidized (with iodine) in the presence of N₂ to yield an analogue of 61, namely Fe(N₂)(CO)₂(CNAr^Tripp2)₂ (69, Scheme 7).^71,147 Compared with 61, which exhibited poor thermal stability, 69 is more stable at room temperature in benzene, n-pentane, and Et₂O solution under an N₂ atmosphere. Such enhanced stability is a consequence of the greater steric protection of the bulkier isocyano ligands CNAr^Tripp2 in combination with the additional electronic stabilization offered by the two CO ligands.¹⁴⁷ The authors found that the dinitrogen ligand has a low degree of activation, as reflected by the short N–N distance determined via X-ray diffraction (d_NN = 1.1059(41) Å) and by the high-energy FT-IR stretching band (ν_N≡N = 2194 cm^–1; see Table 2 below). This lack of activation was attributed to the presence of the strongly π-acidic CO and isocyanide ligands. Furthermore, it was found that the N₂ ligand was sufficiently labile to be displaced, opening the coordination sphere of 69 to engage in small molecule activation. The reactivity of 69 toward H–H or Si–H moieties contrasts with that of its analogue 61. Oxidative addition of any of the studied substrates afforded the respective Fe(II) complexes HFeI(CO)₂(CNAr^Tripp2)₂ (70, R = H, SiEt₃, or SiH₂Ph), with loss of N₂. In addition, the reaction of 69 with white phosphorus led to the cleavage of one edge of the P₄ tetrahedron, forming the butterfly-P₄ complex Fe(κ²-P₄)(CO)₂(CNAr^Tripp2)₂ (71).¹⁴⁹

Table 2. Selected ν_N≡N Stretching Frequencies (cm^–1) for Dinitrogen-Bound Anionic Complexes.

Complex	νN≡N (cm–1)	Ref
free N2 (g)	2331	(547)
Fe(N2)(CO)2(CNArTripp2)2	2194	(147)
[K([18]crown-6][(N2)Fe(CAAC)2]	1850	(76)
[(ICy)3Co(N2)]	1917	(548)
[K(ICy)2Co(N2)2]	1807	(548)
K2[(tBunacnac)Fe(NN)Fe(nacnactBu)]	1589	(549)
K2[(nacnac)Co(NN)Co(nacnac)]	1599	(550)
[(AdPPymDI)Fe(N2)]	2050	(391)
[Na([18]crown-6)(thf)2][(AdPPymDI)Fe(N2)]	1935	(391)
(N2)Co(SiMe3)(CNArMes2)3	2224	(125)
[(P3B)Fe(N2)]	2011	(139)
[Na([12]crown-4)2][(P3B)Fe(N2)]	1905a	(139)
[Na([12]crown-4)2][K(DME)x][(P3B)Fe(N2)]	1836	(484)
[(PPh3Si)Fe(N2)]	2041	(140)
[Na([12]crown-4)2][(PPh3Si)Fe(N2)]	1967	(140)
[(PiPr3Si)Fe(N2)]	2008	(140)
[Na([12]crown-4)2][(PiPr3Si)Fe(N2)]	1920	(430)
[(P3B)Co(N2)]	2089	(526)
[Na([12]crown-4)2][(P3B)Co(N2)]	1978	(551)
[(N2)FeAl{N[o-(NCH2PiPr2)C6H4]3}]	2010	(432)
[(N2)CoAl{N[o-(NCH2PiPr2)C6H4]3}]	2081	(432)
[K([18]crown-6)]+ [(N2)FeAl{N[o-(NCH2PiPr2)C6H4]3}]−	1925	(432)
[K([2.2.2]cryptand)]+ [(N2)CoAl{N[o-(NCH2PiPr2)C6H4]3}]−	1995	(432)
[N(PPh3)2]+ [(N2)CoGa{N[o-(NCH2PiPr2)C6H4]3}]−	1999	(525)
[N(PPh3)2]+ [(N2)CoIn{N[o-(NCH2PiPr2)C6H4]3}]−	2021	(525)
[K([2.2.2]cryptand)]+ [(N2)CoV{N[o-(NCH2PiPr2)C6H4]3}]−	1971	(494)
[K([2.2.2]cryptand)]+ [(N2)CoCr{N[o-(NCH2PiPr2)C6H4]3}]−	1990	(494)
[K([2.2.2]cryptand)]+ [(N2)Co2{N[o-(NCH2PiPr2)C6H4]3}]−	1994	(552)

^aAverage value.

Reaction between the nucleophilic complex K₂[Fe(CO)₂(CNAr^Dipp2)₂] (63) and anthracendiyl-substituted chlorophosphine, ClP(anthr) (1 equiv; anthr = 9,10-anthracendiyl) afforded the anionic complex [Fe(P(anthr))(CO)₂(CNAr^Dipp2)₂]⁻ (72, Scheme 7, bottom).¹⁴⁸ Single crystal X-ray diffraction experiments revealed that the molecular structure features an intact anthracendiyl-substituted phosphanyl ligand. Although 72 is stable in solution for several days at room temperature, it reacts quickly with additional ClP(anthr) (1 equiv) to yield complex [(η²-P₂)Fe(CO)₂(CNAr^Dipp2)₂] (73), in which a side-on coordinated diphosphorus ligand has been formed. It was demonstrated that the susceptibility of ClP(anthr) to suffer C–P bond homolysis and its steric hindrance are essential in the formation of the diphosphorus ligand at the metal center: similar reactions between 72 and other electrophilic chlorophosphines, i.e., ClPPh₂ and ClP(iPr)₂, afforded only the phosphanylphosphine complexes [Fe(κ¹-P(anthr)PR₂)(CO)₂(CNAr^Dipp2)₂] (74, R = Ph, iPr). It was proposed that the formation of 73 could go through an intermediate analogous to 74, in which sequential anthracene extrusion leads to the formation of the diphosphorus scaffold. The average P–P bond distance for the diphosphorus ligand in 73 [d_PP = 1.988(1) Å] is longer than the experimentally determined distance for free P₂ [d_PP = 1.8934 Å], but still evidence of a strong interaction, with significant P–P multiple bond character. This is further corroborated by ³¹P NMR analysis, DFT, and natural bond orbital (NBO) calculations. A comparison between the structural features of 73 and those of the η²-alkyne complex [(η²-BTMSA)Fe(CO)₂(CNAr^Dipp2)₂] (75, BTMSA = bis-trimethylsilylacetylene) confirmed that both complexes are isostructural, suggesting a close electronic analogy. Nonetheless, these complexes exhibit disparate reactivity: while 73 reacts with diene nucleophiles (1,3-butadiene or 1,3-cyclohexadiene) to afford diphospha-Diels–Alder adducts, complex 75 failed to react with these substrates.¹⁴⁸

Bidentate isocyanide ligands form chelates larger than those of typical N,N- or P,P-ligands due to the linear arrangement of the C≡N unit, which imposes rigidity to the structure of the complexes. Considering this, and the known ability of isocyanide ligands to stabilize low-valent iron centers,^{71,123,131,147,148} Wolf, Reiser, and co-workers reported a series of mono- and polynuclear iron complexes featuring the bidentate ligands BINC (bis(2-isocyanophenyl)phenylphosphonate) and CNAr₃NC (2,2′′-diisocyano-3,5,3′′,5′′-tetramethyl-1,1′:3′,1′′-terphenyl).⁷² Although the reductions of isocyanide halide complexes led mainly to neutral complexes, we are including these results for context. Treating anhydrous FeBr₂ with either of the selected bidentate ligands (2.1 equiv; see Scheme 8a and b) afforded mononuclear compounds of general formula [FeBr₂(diisocyanide)₂] [diisocyanide = BINC (76), CNAr₃NC (77)]. Reduction of 76 with KC₈ (2 equiv) resulted in the formation of [Fe₃(BINC)₆] (78, see Scheme 8a), a structural analogue of the carbonyl-iron cluster [Fe₃(CO)₁₂] (22). 78 possesses a triangular iron arrangement in which the shorter Fe–Fe bond (2.495(6) Å vs 2.682(6) and 2.685(6) Å) features two bridging isocyanide units. In solution, 78 has a fluxional behavior between the coordinated bridging and terminal isocyanide moieties. The difference in the stretching bands of the free ligand (ν_C≡N = 2126 cm^–1) with respect to the coordinated isocyanide in complex 76 (ν_C≡N = 2122 cm^–1) or 78 (ν_C≡N = 2035 cm^–1) evidenced the low-valent character of the metal centers in the latter. While the band in the Fe(II) complex 76 is essentially at the same frequency as in the free ligand, in the formally zerovalent 78 this is shifted to lower frequencies, thus indicating substantial metal-to-ligand backbonding. This is additionally supported by the shortened Fe–C bond lengths (1.83(1) Å in 78 vs 1.87(4) Å in 76, on average).

Scheme 8. Synthesis of Bidentate Isocyanide Iron Complexes by the Group of Wolf and Reiser⁷².

Analogous reduction of complex 77, featuring the bulkier isocyanide CNAr₃NC, afforded either the dimeric complex [Fe(CNAr₃NC)₂]₂ (79) or the anionic dimer [K(Et₂O)]₂[Fe(CNAr₃NC)₂]₂ (80), depending on the stoichiometry of KC₈ used (2 or 3.2 equiv; Scheme 8b).

Both dimers possess a butterfly arrangement caused by the restrictions imposed by the bridging isocyanide ligands. Unlike its carbonyl analogue [Fe₂(CO)₈] (81), 79 is stable and isolable at room temperature. Such stability was attributed to the steric effect of the bulky isocyanide ligands. A contrast between the structural data of 80 and that of its carbonyl analogue [Fe₂(CO)₈]^2– (82)^150−152 revealed that, in the latter, the Fe–Fe bond is significantly longer (2.804(1) Å) than in 80 (d_Fe–Fe = 2.552(8) Å) due to the absence of bridging carbonyl ligands. Treatment of the Fe(II) precursor [Cp*FeCl(tmeda)] (83, Cp* = C₅Me₅, tmeda = tetramethylethylene-1,2-diamine) with the diisocyanide ligand BINC (Scheme 8c) led to the formation of the mononuclear complex [Cp*FeCl(BINC)] (84), which undergoes reduction by reaction with KC₈ (1 equiv) to form the dinuclear heteroleptic compound [Cp*Fe(BINC)]₂ (85). This crystallizes as the cis-isomer, with both Cp* groups coordinated above the Fe–Fe bond. However, in solution (THF-d₈), the NMR spectroscopic measurements revealed an equilibrium between the cis and trans isomers.⁷²

2.2.3.3. Cobalt Isocyanides

The first examples of homoleptic isocyanide cobaltates were [K(dme)][Co(CNXyl)₄] (5), reported by Cooper and co-workers (vide supra, Figure 1),^60,61 and a closely related alkylisocyanocobaltate complex [K([18]crown-6)(thf)₂][Co(CNtBu)₄] (86), reported by Ellis and Brennessel.⁸⁵ In 2010, Figueroa and co-workers described the synthesis and reactivity of bulky arylisocyanocobaltates, which have an extensive reaction chemistry.¹²⁴ The ion contact complex Na[Co(CNAr^Mes2)₄] (Na[87]), featuring a Co(−I) center, was synthesized by reduction of CoCl₂ with sodium amalgam, in the presence of the encumbering ligand CNAr^Mes2 (Scheme 9). The observed ion-pairing is similar to that previously discussed (vide supra) for the iron metalate 51, reported by the same group.¹²³ Crystallographic characterization confirmed the ion contact, which was also observed in 5. Addition of [Ph₃PNPPh₃]Cl {[PPN]Cl; PPN = [Ph₃PNPPh₃]⁺) was found to disrupt the contact ion pair, and the ion-separated metalate [PPN][Co(CNAr^Mes2)₄] ([PPN][87]) was obtained. Stretching frequencies (FT-IR) for both the ion-paired and the ion-separated complexes are given in Table 1, with the number of bands reflecting the changes in pairing.

Scheme 9. Synthesis of Sterically Encumbering Isocyanide Cobaltates¹²⁴.

Reaction of Na[87] with FcOTf (1 equiv; Fc = ferrocenium) yields the neutral paramagnetic complex Co(CNAr^Mes2)₄ (88; Scheme 10a), contrasting with the behavior of the previously reported [K(dme)][Co(CNXyl)₄] (5), which dimerizes to Co₂(CNXyl)₈ after oxidation. Hence, the greater steric encumbrance of the isocyanide ligands CNAr^Mes2 stabilizes the monomeric neutral Co(0) species. The zerovalent tetraisocyanide complex 88 is a sterically encumbered analogue of the binary cobalt carbonyl complex Co(CO)₄. Addition of two equivalents of FcOTf to tetrahedral Na[87] in the presence of NaBAr^F₄ affords a diamagnetic square planar tetraisocyanocobalt(I) complex, [Co(CNAr^Mes2)₄]BAr^F₄ (89), while analogous reaction in THF as a donor solvent results in pyramidalization of the cobalt(I) species, yielding [Co(CNAr^Mes2)₄(thf)]OTf (90). Furthermore, it was observed that the Co(I) species 89 and 90 comproportionate with the cobaltate complex 87 to generate the neutral compound 88. Therefore, the authors were able to identify a fully reversible one-electron modulation of the +I, 0, and −I formal oxidation states for the [Co(CNR)₄] fragment.¹²⁴

Scheme 10. (a) Oxidation of the Co(−I) Complex to Yield Co(0)/Co(I)-Isocyanide Compounds, and Synthesis of a Well-Defined Isocyano-Hydride-Cobalt Complex; (b) Formation of a Tris(isocyano)cobaltate and Its Subsequent Reactivity^124,125,153.

Moreover, Figueroa and co-workers isolated an unusual isocyano hydrido cobalt complex HCo(CNAr^Mes2)₄ (91) from Na[Co(CNAr^Mes2)₄] (Na[87]) and 3,5-dimethylbenzoic acid (1.05 equiv, Scheme 10a, right).^124,153 The hydride ligand gives rise to a strongly high-field shifted ¹H NMR resonance (δ = −13.1 ppm).¹⁵³ Compound 91 is the isocyano analogue of the carbonyl complex HCo(CO)₄ (92), known for its relevance in catalytic hydroformylation processes.^154,155 Hydride 91 reacts with H₂ (1 atm, r.t.) to afford the methylenimine derivative H₂C=NAr^Mes2 as the main product, along with small quantities of an η⁶-arene iminoformyl species, [Co(κ¹-CC(H)NAr^Mes2)(CNAr^Mes2)₃] (93). The catalytic 1,1-hydrogenation of the isocyanide CNAr^Mes2 was achieved in good yields (76 ± 13%) using 91 as a catalyst, with a loading of 5 mol % (16 h, r.t.). It is worth noting that the authors carried out a full evaluation of the relative acidity of 91, and its reactivity toward olefins and Brønsted acids, demonstrating its ability to undergo insertion chemistry and H atom transfer reactions.^133,153,156

Prolonged stirring of [PPN][87] in n-hexane resulted in the dissociation of one isocyanide CNAr^Mes2 ligand, followed by precipitation of the zwitterionic tris(isocyano) complex (η²-PPN)Co(CNAr^Mes2)₃ (94, Scheme 10b), which has been proposed as a highly reactive synthon for the coordinatively unsaturated monoanionic cobaltate [Co(CNAr^Mes2)₃]⁻.¹²⁵ The authors suggested that the dearomatization of the bound phenyl ring in 94, as observed via X-ray diffraction analysis, evidenced a strong π-basic character on the [Co(CNAr^Mes2)₃]⁻ fragment. However, regardless of the strong π-back-donation toward the phenyl ring, in the solid state, the cobalt center in 94 remains a highly reduced one, as indicated by the shift on the FT-IR stretching bands of the isocyano moieties with respect to the free ligand [ν_CN(94) = 1952, 1860, and 1824 cm^–1 (see Table 1) vs ν_CN (free CNAr^Mes2) = 2118 cm^–1].¹²⁷ It was also suggested that the ligation of the phenyl unit from the [PPN]⁺ cation, usually noncoordinating, is essential for the (kinetic) stabilization of the coordinatively unsaturated, electron-rich fragment [Co(CNAr^Mes2)₃]⁻. Consequently, the isolation of analogues of the latter with different cations (alkali metal, alkaline-earth metal, tetra-alkylammonium or tetra-alkylphosphonium) was unsuccessful. The authors underlined the contrasting behavior between [PPN][87] and the tetracarbonyl compound [PPN][Co(CO)₄] ([PPN][95]),¹⁵⁷ for which no CO-ligand displacement by the [PPN]⁺ cation was reported. This disparate behavior highlights the utility of the isocyanide ligand analogues in the synthesis and stabilization of low-valent, low-coordinate transition metal complexes. The metalate 94 could then be proposed as an isocyano analogue of the tricarbonyl monoanionic compound [Co(CO)₃]⁻, previously observed exclusively in the gas phase.¹⁴⁶

Despite the kinetic stability observed for 94, the compound slowly decomposed, after 30 h in a solution of C₆H₆, to the η⁶-arene, phenyl-substituted iminoacyl complex {[η⁶-MesAr^MesN=C(Ph)]Co(CNAr^Mes2)} (96). Experiments with the deuterium-labeled analogue (η²-[D₃₀]-PPN)Co(CNAr^Mes2)₃ ([D₃₀]-94) afforded exclusively {[η⁶-MesAr^MesN=C(C₆D₅)]Co(CNAr^Mes2)} ([D₅]-96), thereby indicating that the iminoacyl Ph group came from the [PPN]⁺ counterion, through P–C bond cleavage and formal transfer of a [C₆H₅]⁺ moiety. The tris(isocyano)cobaltate 94 represented a highly reactive species toward electrophilic reagents (Scheme 10b). Treatment of 94 with bis(diethylamino)chlorophosphine (ClP(NEt₂)₂), or with trimethylsilyl chloride (Me₃SiCl) in the presence of N₂, yielded the complexes (Et₂N)₂PCo(CNAr^Mes2)₃ (97) and (N₂)Co(SiMe₃)(CNAr^Mes2)₃ (98), respectively. These results evidenced the nucleophilic character of metalate 94, while, in the formation of 98, remaining active for small molecule binding after the electrophilic functionalization. Additionally, metalate 94 acted as a nucleophile in multistep transformations of electrophilic organic carbonyls (Scheme 10b). The reaction of 94 with maleic anhydride forms the isomaleimide-containing complex [PPN][(η²-C,C-(IMAr^Mes2)Co(CO)(CNAr^Mes2)₂] [99; IMAr^Mes2 = 5-(Ar^Mes2-imino)furanone], whereas analogous treatment with pivaloyl chloride [tBuC(O)Cl] afforded a 4:1 mixture of HCo(CO)(CNAr^Mes2)₃ (100) and the iminoacyl complex Co(CO)(η²-C,N-(tBuC=NAr^Mes2)(CNAr^Mes2)₂ (101). Mechanistic studies on the formation of 99 suggest a decarbonylation step of a bound acyl ligand generated after anhydride ring opening, and these observations would also explain the results obtained in the analogous reaction with tBuC(O)Cl. The inherent reactivity of the C–N bond in these isocyano ligands could be considered an impediment to further explore their interaction with other reagents.¹²⁵

Interesting reactivity was also observed for the zerovalent tetraisocyanide complex Co(CNAr^Mes2)₄ (88), which undergoes L-type ligand substitution or exhibits (metallo)radical behavior. Both types of reactivity were explored by Figueroa and co-workers.¹⁵⁸ Although further details on such reactivity go beyond the scope of this work, in general, ligand-exchange reactions displacing one CNAr^Mes2 unit were observed upon interaction of 88 with two electron donors (triphenylphosphine, tert-butyl isocyanide, tert-butylethylene, acetylenes, maleic anhydride, and benzaldehyde), yielding a series of substituted products, (L)Co(CNAr^Mes2)₃ (102). Despite the large steric profile of 88, the complex undergoes associative ligand substitution reactions, as demonstrated through kinetic experiments. HSnBu₃ and HCCl₃ failed to react with 88 via a hydrogen- or chlorine-atom (X·) abstraction pathway, while other substrates (e.g., diphenyl disulfide, elemental sulfur, and P₄) led to inner-sphere multielectron transformations, instead of one electron reactions, in every case.¹⁵⁸

A series of mixed ligand (carbonyl/isocyano) analogues of [Co(CNAr^Mes2)₄]⁻ (87) and their derived hydrides were similarly described.¹³³ Initial attempts to synthesize the mixed ligand cobaltates from either of the tetraisocyano complexes, Na[87] or [PPN][87], or from the masked trisisocyano compound (η²-PPN)Co(CNAr^Mes2)₃ (94) by stoichiometric addition of CO (1, 2, or 3 equiv) resulted in a mixture of all monoanionic cobaltate complexes [Co(CO)_4–n(CNAr^Mes2)_n]⁻ [n = 3 (103), 2 (104), 1 (105)] and free isocyanide ligand (CNAr^Mes2),^124,125 which was attributed to a rapid redistribution process.

Independent synthetic routes were subsequently developed for compounds 103–105 and their corresponding hydrides (Scheme 11). HCo(CNAr^Mes2)₄ (91) reacted with one equivalent of carbon monoxide to yield the hydride complex HCo(CO)(CNAr^Mes2)₃ (100, Scheme 10b, vide supra) by ligand displacement. Compound 100 was subsequently deprotonated by Na[N(SiMe₃)₂] to afford Na[Co(CO)(CNAr^Mes2)₃] (Na[103], Scheme 11a). The dicarbonyl compound 104 could not be similarly obtained from hydride 100. An alternative synthesis, based on reductive cleavage of the reported dimer Co₂(CO)₄(CNAr^Mes2)₄ (106) by sodium amalgam, afforded complex Na[Co(CO)₂(CNAr^Mes2)₂] (Na[104], see Scheme 11b).^133,159 Use of the dimer Co₂(CO)₆(CNAr^Mes2)₂ (107) led to Na[Co(CO)₃(CNAr^Mes2)] (Na[105]). However, since dimer 107 exists in solution as a ∼1:1 mixture of the 1,1-Co₂(CO)₆(CNAr^Mes2)₂ and 1,2-Co₂(CO)₆(CNAr^Mes2)₂ isomers at room temperature, the reduction by sodium amalgam afforded a mixture of Na[Co(CO)₄] (Na[95]), Na[104], and Na[105], in ca. 1:1:2 ratio according to NMR and IR spectroscopy (Scheme 11c).^133,159 After fractional crystallization, Na[105] was obtained (∼40% yield). X-ray crystallographic analysis of the sodium salts of 103–105 revealed that substantial ion-pairing interactions exist in the solid state between anions and cations. Similar to the case of Na[87], the ion-pairing interactions could be disrupted by addition of [PPN]Cl, generating the respective PPN salts [PPN][103], [PPN][104], and [PPN][105]. These three compounds give rise to low-energy ν_CO and ν_CN stretching bands (see Table 1) which reflect the highly reduced, π-basic character of the cobalt centers. The difference in the number of bands in each case evidences the changes in coordination environment and symmetry of the three mixed ligand cobaltates. The hydrides of complexes Na[104] and Na[105] were obtained by reaction with 3,5-dimethylbenzoic acid, as in the case of HCo(CNAr^Mes2)₄ (91, Scheme 10a).^133,153 An assessment of the relative Brønsted acidity revealed a progressive increase in pK_α^THF values [in the range 40.7–28.6 for the different hydride complexes HCo(CO)_4–n(CNAr^Mes2)_n, n = 1–4] as more isocyanide ligands were added to the Co center. This increase results from the inherently higher σ-donor/π-acid character of the isocyanide ligands. Therefore, this approach is a way to modulate the acidity of metal hydride complexes.

Scheme 11. Synthesis of Mixed Ligand (Carbonyl/Isocyano) Cobaltates¹³³.

The Figueroa group also developed the chemistry of heteroleptic “Cp^R′Co(isocyanide)” complexes (R′ = H, Me) of different nuclearities, including anionic complexes.^134,135,160 Compounds [Cp^R′Co(I)₂(CNAr^R)] [Cp^R′ = Cp, CNAr^R = CNAr^Mes2 (108); Cp^R′ = Cp, CNAr^R = CNAr^Dipp2 (109); Cp^R′ = Cp*, CNAr^R = CNAr^Dipp2 (110); Cp^R′ = Cp, CNAr^R = CNAr^Tripp2 (111)] served as starting materials for the synthesis of dinuclear or mononuclear complexes featuring either bridging isocyanide ligands or a terminal carbyne, respectively.^134,135 The dinuclear complexes [(μ-CNAr^Mes2)₂{CpCo}₂] (112), K[(μ-CNAr^Mes2)₂{CpCo}₂] (K[113]), and K₂[(μ-CNAr^Mes2)₂{CpCo}₂] (K₂[114]) were obtained by reduction of 108 with increasing equivalents of KC₈ (see Scheme 12).¹³⁴ In the solid state, both anionic species, K[113] and K₂[114], form an ion pair with the potassium counterions and the arenes in the isocyanide ligand. The increasing electron density in the complexes as a function of the degree of reduction is reflected in the red shift for the ν_CN band of the bridging isocyanide ligands (ν_CN(112) = 1834 cm^–1, ν_CN(K[113]) = 1666 cm^–1, ν_CN(K₂[114]) = 1511 cm^–1; see Table 1), observed by FT-IR spectroscopy, demonstrating the successive increase in Co → (CN)π* back-donation. The three dimers constitute the formal triad of d⁸–d⁸, d⁸–d⁹, d⁹–d⁹ complexes, analogous to the carbonyl species [(μ²-CO)₂[CpCo]₂]ⁿ, for which the dianion has remained elusive to isolation.^161,162 In contrast, the dianionic isocyanide complex K₂[114] was isolated and characterized, despite isocyanide ligands being poorer π-acids than CO, thereby decreasing the reduction potentials of their complexes. This divergent behavior was ascribed to the ability of the CNAr^Mes2 ligands to form CN π-bond and π-arene/alkali-metal-cation interactions, as observed in the anionic compounds K[113], K₂[114], and other isocyanocobaltates such as the homoleptic Na[Co(CNAr^Mes2)₄] (Na[87]).¹²⁴ Unlike with Na[87], however, the π-arene/K⁺ interactions in K₂[114] could not be conveniently disrupted with either sequestering agents ([18]crown-6 or dibenzo-[18]crown-6) or noncoordinating cations such as [PPN]⁺, with reactions leading to intractable mixtures. Therefore, it was concluded that the structural stabilization of dimer K₂[114] critically depends on the tight ion pair formed with the K⁺ counterion. Computational analysis of the bonding in the M₂(μ-CNAr^R)₂ core indicated that the predominant interaction is the π-backbonding with the bridging ligands instead of the formation of direct metal–metal bonds and that the former determines the geometric and electronic structures observed. Furthermore, the authors also synthesized the analogous d⁸–d⁹ and d⁹–d⁹ pairs [Co₂{(μ-CNAr^Mes)₂(η⁶-Mes)Co}₂]ⁿ (115, n = 0, 1+), in which one mesityl substituent on the isocyanide ligand coordinates to the cobalt center in an η⁶-fashion, and the d⁹–d⁹ nickel dimer [(μ-CNAr^Mes2)₂{CpNi}₂] (116), isoelectronic with K₂[114]. For these complexes, the observed electronic and structural trends indicate that, as in the cobalt dimers, the π-backbonding interactions influence the electronic structure significantly more than metal–metal interactions.¹³⁴ Contrasting behavior was observed for the reduction of 109 or 110, which afforded only neutral complexes: a dinuclear compound analogous to 112–114 or a mononuclear dinitrogen(isocyanide)cobalt adduct, respectively.¹⁶⁰

Scheme 12. Synthesis of Heteroleptic Cp^R′Co(isocyanide) Complexes by the Group of Figueroa^134,135,160.

Reduction of [Cp*Co(I)₂(CNAr^Tripp2)] (111) by KC₈ yielded a low valent dianionic cobalt terminal carbyne (see Scheme 12, top left).¹³⁵ The product, [{K(Et₂O)}₂{Cp*Co≡CNAr^Tripp2}] (117), shows a “pogo stick” structure additionally stabilized by contact ion pairing of the potassium counterions with the aryl (Tripp) substituents, the carbon atoms, and the nitrogen atoms of the ligand. FT-IR analysis revealed the presence of a significantly reduced C–N bond order through a substantial red-shift observed for the ν_CN band with respect to the free isocyano ligand [ν_CN(117) = 1509 cm^–1 vs ν_CN(CNAr^Tripp2) = 2122 cm^–1] or reported Co(isocyano) complexes (see Table 1). This, along with the short Co–C bond length (d_CoC(117) = 1.670(3) Å), supports the formulation of the complex as a metal carbyne. However, the molecular structure of 117 also features a relatively long C_carbyne–N bond distance (d_CN (117) = 1.307(3) Å), a short distance between the nitrogen atom and the m-terphenyl-ipso carbon (d_Ncipso = 1.372(3) Å), and altered metric parameters in the central aromatic ring indicating dearomatization. These findings point toward a significant N → C π-donation opposite to the carbyne interaction, indicating the presence of a highly reduced (electron-rich) cobalt center. Therefore, the authors proposed a possible azabenzallyl resonance form as one of the contributors to the electronic structure of the dianionic complex 117 (Scheme 12, bottom left). DFT calculations on the simplified model complex [Cp*Co≡CNXyl]^2– (117′, Xyl = 2,6-Me₂C₆H₃) indicate that the compound has a substantial atomic d-orbital character at the cobalt center, which suggests the presence of a metal–carbon multiple bond, like those in Fischer-type carbynes. The calculations support a bonding situation in which a low-valent cobalt center with highly filled d-orbitals binds to the ligand at the carbon atom, establishing a significant metal-to-carbon π-backdonation interaction.¹³⁵

Insights into the electronic structure of 117 indicate nucleophilic character at cobalt. This was confirmed by treatment of 117 with HCCSiMe₃ and Me₃SiCl (Scheme 13), which afforded the two-legged piano stool complexes [K(Et₂O)][Cp*Co(H)(CNAr^Tripp2)] (118) and [K(Et₂O)][Cp*Co(SiMe₃)(CNAr^Tripp2)] (119). Both reactions illustrate cobalt-centered reactivity, with an electronic reorganization yielding the coordinated isocyano CNAr^Tripp2 ligand and formal concomitant oxidation of the cobalt center to Co(I). The significant π-backbonding in these species was evidenced by their ν_CN stretching bands [ν_CN(118) = 1710 cm^–1; ν_CN(119) = 1707 cm^–1]. The observation that the metal center is the preferred site of electrophilic attack additionally supports the notion that the carbyne ligand is stabilized via an azabenzallyl resonance form. By contrast, the reaction of 117 with diphenylacetylene led to the formation of a new C–C bond after [2 + 2] cycloaddition to the Co≡C motif, with protonation of the α-carbon of the intermediate metallacycle, to finally form the vinyliminacyl complex [K(Et₂O)][120] (see Scheme 13, bottom). The authors proposed a possible mechanism to explain the formation of the vinyliminacyl product.¹³⁵

Scheme 13. Reactivity of [{K(Et₂O)}₂{Cp*Co≡CNAr^Tripp2}] (117) toward Protic Substrates and Silyl Electrophiles¹³⁵.

2.2.3.4. Isocyanide Complexes of the 4d and 5d Metals

Isocyanide metalates based on 4d or 5d metals have also been reported in the period from 2006–2022.^163−167 Ellis described the isolation of niobium and tantalum homoleptic isocyanidemetalate complexes, namely Cs[M(CNXyl)₆] [M = Nb (121), Ta (122), Xyl = 2,6-Me₂C₆H₃; Scheme 14].^163,168 The compounds Cs[M(CNXyl)₆] were obtained in two reaction steps which include, first, ligand exchange from the corresponding carbonyl metalate [Et₄N][M(CO)₆] [M = Nb (123), Ta (124)]¹⁶⁹ while oxidizing with iodine, followed by subsequent reduction of the resulting [MI(CNXyl)₆] [M = Nb (125), Ta (126)] with CsC₈. These octahedral complexes, bearing formally d⁶ M(−I) centers, were characterized by X-ray diffraction analysis, and ¹³C{¹H} NMR spectroscopy for the tantalum species. In 2017, the Ellis group reported the tantalate complex K[Ta(CNDipp)₆] (127, Dipp = 2,6-iPr₂C₆H₃) in two steps by a similar procedure starting from 124 (see Scheme 14).^165,170 Displacement of CO ligands by CNDipp under oxidative conditions (I₂) formed the precursor [TaI(CNDipp)₆] (128). 128 was then reduced by an excess of KC₈ to afford 127 in 70–80% yield. In this work, Ellis disclosed the discovery of appropriate oxidizing agents to access the until then elusive, isolable complex [Ta(CNDipp)₆] (129) from 127. Oligomeric MoO₃, [Ta(CNDipp)₇][BF₄] (130) or [NEt₃H][BPh₄] cleanly oxidized metalate 127 to the desired homoleptic zerovalent tantalum species 129. Later, Ellis and co-workers reported the crystal structure of the related complex [Ta(CNDipp)₇][Ta(CNDipp)₆] (131), obtained by reducing 128 with an excess CsC₈ (5.8 equiv).¹⁶⁴ In compound 131, both the complex cation and anion are homoleptic, featuring the same transition metal and π-acceptor ligand. 131 undergoes comproportionation in solution to form the 17-electron complex 129.

Scheme 14. Isocyanide Tantalate and Niobiate Complexes by the Ellis Group^163,165.

Figueroa reported that upon reduction of [ReBr(CO)₃(CNAr^Dipp2)₂] (132), the isolation of the monomeric rhenium(−I) anion Na[Re(CO)₃(CNAr^Dipp2)₂] (133, Ar^Dipp2 = 2,6-(2,6-(iPr)₂C₆H₃)₂C₆H₃) was achieved (Scheme 15a).¹⁶⁶133 contains an encumbering isocyanide scaffold well-suited to support the highly reduced rhenium atom, which is in a formally negative oxidation state. Later, Abraham and Figueroa described the convenient preparation of starting materials to access unprecedented low-valent rhenium and technetium anionic complexes (Scheme 15b).¹⁶⁷

Scheme 15. Synthesis of Mixed Ligand Technetium and Rhenium Metalates Containing Formally M(−I) Metal Centers^166,167.

The compounds [MX(CO)(CNAr^DArF2)₄] [M = Tc, X = Cl (134); M = Re, X = Br (135); Ar^DArF2 = 2,6-(3,5-(CF₃)₂C₆H₃)₂-4-F-C₆H₂] can be reduced with Na/Hg to yield the respective Na[M(CO)(CNAr^DArF2)₄] species [M = Tc (136), M = Re (137)], featuring formally M(−I) centers. The encumbering fluorinated isocyanide ligand CNAr^DArF2 has been key to stabilize the uncommonly low oxidation state by increasing the π-accepting properties, as well as to provide steric protection, thereby preventing dimerization. These anionic metalates are surprisingly robust and, for instance, the Re(−I) complex Na[Re(CO)(CNAr^DArF2)₄] (137) readily reacted with electrophiles such as MeI, HCl, or F₆C₅C(O)Cl. The latter behavior mimicked that of the homoleptic carbonyl [Re(CO)₅]⁻ (138). These mixed ligand Tc and Re [M(CO)(CNAr^DArF2)₄]⁻ complexes are heavier analogues of the manganese compounds [Mn(CO)₂(CNAr^Mes2)₃]⁻ (38) and [Mn(CO)₃(CNAr^Dipp2)₂]⁻ (39), described by the same group and depicted in Scheme 2 (vide supra).¹²¹ This family of group 7 metal complexes [M(CO)_x(CNAr)_y]⁻ has the coordination number x + y = 5 in common, but different carbonyl/isocyanide ratios, x/y = 1/4, 2/3, or 3/2 due to steric demand of the respective CNAr ligand and the relative size of the corresponding metal center (M = Mn, Tc, Re), resulting in disparate geometries.

Besides the series of [Re(CO)_x(CNAr)_y]⁻ complexes by Figueroa and co-workers,^166,167 other rhenium metalates bearing carbonyl or isocyanide ligands with the metal atom in a formally negative oxidation state are rare. For instance, Kubiak and co-workers reported the [Re(CO)₃(bpy)]⁻ anion (139), which is likely involved in the electrocatalytic reduction of CO₂ by [Re(CO)₃(Cl)(bpy)] (140).^171,172 Nevertheless, a follow-up study revealed that the [Re(CO)₃(bpy)]⁻ species is better described as [Re⁰(CO)₃(bpy^–)], rather than [Re^–(CO)₃(bpy)], due to the presence of the redox-active α-diimine chelate.¹⁷³

Figueroa and co-workers isolated mono- and dianionic triplatinum-isocyanide clusters related to the monomeric Chini cluster, [Pt₃(CO)₆]^2– (141).¹⁷⁴ The encumbering m-terphenyl isocyanide ligands CNAr^Dipp2 (Ar^Dipp2 = 2,6-(2,6-(iPr)₂C₆H₃)₂C₆H₃) provide kinetic stability to the series [Pt₃(μ-CO)₃(CNAr^Dipp2)₃]ⁿ⁻ (n = 0, 1, 2), allowing for its isolation and full characterization. The preparation of the heteroleptic trinuclear complexes is straightforward, starting with ligand exchange of [Pt(CNAr^Dipp2)₂] (142) under a CO atmosphere affording the neutral compound [Pt₃(μ-CO)₃(CNAr^Dipp2)₃] (143, Scheme 16). 143 was reduced with KC₈ to yield the dianion [Pt₃(μ-CO)₃(CNAr^Dipp2)₃]^2– (144), and comproportionation of 143 and 144 produces the radical monoanion [Pt₃(μ-CO)₃(CNAr^Dipp2)₃]^·– (145), as depicted in Scheme 16. The highly reduced cluster 145 reacted cleanly with various electrophiles, such as R₃ECl (R = Et and E = Si or R = Ph and E = Sn) or [AuCl(PPh₃)] (146), giving rise to molecularly defined and isolable compounds. Interestingly, spectroscopic analysis (FT-IR, multinuclear NMR) and computational studies (DFT) suggest that the highest-occupied molecular orbitals (HOMO) of the charged clusters consist of a combined π*-framework of the CO and CN^ArDipp2 ligands, which would imply that this set of isocyanide and carbonyl ligands exhibit redox noninnocent character.

Scheme 16. Synthesis of Dianionic and Radical Anionic Chini-type Triplatinum Clusters with Encumbering Isocyanide Ligands¹⁷⁴.

Using the same bulky isocyanide ligands, Figueroa and co-workers very recently reported the isolation and reactivity studies of 16 valence-elctron rhodate and iridate anions, [M(CNAr^Dipp2)₃]⁻.¹⁷⁵

Ruiz and co-workers have extensively studied the chemistry of mixed-ligand complexes of the group 6 metals, featuring multiple metal–metal bonds.¹⁷⁶ These complexes usually bear carbonyl, cyclopentadienyl, and phosphinidene ligands and, therefore, their chemistry is summarized in section 2.2.6.4 (vide infra).

2.2.4. Arene and Alkene (Hydrocarbon) Metalates

Within the area of transition metal complexes in uncommonly low oxidation states, the chemistry of homoleptic alkene- and arene metalates has attracted considerable attention. Ellis and co-workers have made significant contributions to the chemistry of this type of metalate.³ This group has synthesized numerous examples of low-valent arene and alkene complexes with different transition metals, including those of the 3d series and heavier congeners.^2,67,177,178 Many of these are nowadays considered key synthons in low-valent transition metal chemistry, and deemed storable sources of transition metal anions,^67,179 also known as “naked-metal atom” reagents.²

Due to the large number of metalates containing arene or alkene ligands, this subchapter is compartmentalized into sections according to the metal ion involved. Group 4 metalates (Ti, Zr, Hf) are discussed together due to their similar properties. Metalates of the 3d transition metals are then discussed sequentially, followed by an account of the smaller number of reported 4d and 5d metalates, which sometimes contrast with the 3d metalates.

2.2.4.1. Group 4 Metalates

A striking example of the chemistry of alkene/arene transition metal anions is the synthesis of the triads of tris(anthracene)metalate(−II), [M(C₁₄H₁₀)₃]^2– [M = Ti (147), Zr (148), Hf (149)] and tris(naphthalene)metalate(−II), [M(C₁₀H₈)₃]^2– [M = Ti (150), Zr (151), Hf (152)] complexes.^177,180 The synthetic protocol for the corresponding tris(polyarene)titanates(−II) [polyarene = anthracene (147), naphthalene (150)] is illustrated in Scheme 17.¹⁸⁰

Scheme 17. Synthesis of Tris(polyarene)titanates(−II) [Polyarene = Anthracene (147), Naphthalene (150)] by Ellis and Co-workers¹⁸⁰.

Although the characterization of these titanates in solution by NMR spectroscopy indicated that the three polyarene ligands are magnetically equivalent, the X-ray diffraction analysis showed that the structures observed in the solid state are more accurately formulated as [Ti(η⁴-polyarene)₂(η²-polyarene)]^2–. Thus, 147 and 150 are 16-electron homoleptic polyarene complexes with polyarene ligands of different hapticities. This reduced hapticity was attributed to the existence of strong intramolecular polyarene repulsions due to the short distance between the ligand and the Ti center, caused by a high degree of backbonding from the electron-rich metal to the ligand.

Brennessel and Ellis have reported the structural characterization of titanium¹⁸¹ and vanadium^181,182 alkene or arene metalates. The Ti(−I) compound {[K([18]crown-6)][Ti(C₁₆H₁₀)₂]}_n (153, see Figure 5, left) was obtained by ligand exchange on the highly reactive titanate 150 with pyrene. The M(0) complexes [M(C₁₆H₁₀)₂] [M = V (154), Nb (155); see Figure 5 for the structure of the vanadium(0) complex] were obtained by reduction of the corresponding precursors MCl_4–n(thf)_2+n (M = V: n = 1; M = Nb: n = 0) by alkali metal pyrene radical anion salts. These sandwich compounds bear pyrene ligands in an eclipsed conformation.¹⁸¹

Figure 5.

Homoleptic pyrene complexes of titanium and vanadium.¹⁸¹

Various other highly reduced naphthalene and anthracene complexes of Sc, Ti, Zr, and Hf supported by amide, anilide, triamidoamine, and triaryloxidemethyl ligands have been reported.^183−188 In these, the arene ligands are reduced to their dianionic form. Although these complexes are beyond the scope of the present review, because they feature metal atoms in the common oxidation states +III (for Sc) and +IV (for Ti, Zr, Hf), it is noteworthy that their reactivity with unsaturated substrates (e.g., with N₂, P₄, alkynes, and fluoroarenes) often resembles the behavior of low-valent metalate anions. We would like to refer to the original literature for further discussion.^187−191

Arene complexes of titanium and vanadium supported by σ-/π-bonded pyrrolide-based ligands were developed by Gambarotta and co-workers.^192,193 The chemistry of these complexes is discussed in section 2.2.8.2 (vide infra).

2.2.4.2. Vanadium, Chromium, and Manganese

The synthesis and structural characterization of a family of naphthalene- and anthracene-vanadates were reported.¹⁸² Brennessel and Ellis described the reduction of the known compound bis(naphthalene)vanadium(0) (156)¹⁹⁴ by potassium naphthalenide (Scheme 18a), which affords brown solutions of a paramagnetic product too unstable for isolation or characterization, but able to act as an intermediate in the synthesis of well-defined vanadates. Given the subsequent reactivity observed, the identity of this intermediate was speculated to be [V(η⁴-C₁₀H₈)₂(thf)]⁻ (157), although the presence of the possible 18-electron vanadate [V(η⁶-C₁₀H₈)₂]⁻ (158) could not be ruled out.

Scheme 18. Synthesis of Naphthalene and Anthracene Vanadates by the Group of Brennessel and Ellis¹⁸².

Regardless, the in situ generated species reacted with a series of reagents: further reduction with potassium naphthalenide (1 equiv) in the presence of PMe₃ yielded the V(−II) complex [K(thf)₂]₂[V(η⁴-C₁₀H₈)(η⁶-C₁₀H₈)] (159), whereas reaction with 1,2-bis(dimethylphosphino)ethane (dmpe, 1 equiv) returned [V(η⁴-C₁₀H₈)₂(dmpe)]⁻ (160, L = [18]crown-6 or [2.2.2]cryptand; Scheme 18a, top). Additionally, the intermediate species reacts with anthracene (2 equiv) at low temperatures to give complex [V(η⁴-C₁₄H₁₀)₂(thf)]⁻ (161) (Scheme 18a, down). More efficient synthetic protocols for 159 and 161 consist of the direct reduction of VCl₃(thf)₃ with potassium naphthalenide (5 equiv, in the presence of PMe₃) or potassium anthracenide (5 equiv), respectively, in THF.

In addition, the anthracene vanadate 161 served as starting material for the synthesis of the phosphine adducts [V(η⁴-C₁₄H₁₀)₂(PMe₃)]⁻ (162) and [V(η⁴-C₁₄H₁₀)₂(dmpe)]⁻ (163) (Scheme 18a, bottom). Furthermore, in an attempt to synthesize a mixed arene-alkene anthracene-COD vanadate, VCl₃(thf)₃ was treated with sodium anthracenide (3 equiv) in the presence of 1,5-cyclooctadiene to afford crystals of the dimetallabis(anthracene) sandwich complex {[Na(thf)₃][V₂(C₁₄H₁₀)₂]} (164, Scheme 18b).¹⁸²

In 2018, the Caulton group reported the dichromium sandwich complex [Cr₂(naphthalene)₂]⁻ (165, as [K([18]crown-6)]⁺ salt; see Scheme 19), obtained as a radical anion.¹⁹⁵ The sandwich compound is formed after reaction between the previously reported Cr[N(SiMe₃)₂]₂(thf)₂ (166)^196,197 and an equimolar amount of [K([18]crown-6)][C₁₀H₈]. The structure of the obtained species significantly differs from that of the long-known bis(naphthalene)chromium(0) complex [Cr(η⁶-C₁₀H₈)₂] (167), generated via reduction of CrCl₃(thf)₃ by lithium or sodium naphthalenide.^194,198,199

Scheme 19. Proposed Chemical Reaction for the Synthesis of the Bis(naphthalene)dichromium Sandwich Complex [Cr₂(naphthalene)₂]⁻ (165)¹⁹⁵.

Crystallographic characterization of 165 revealed the formation of a 1D coordination polymer with alternating units of the anion [Cr₂(naphthalene)₂]⁻ and the cation [K([18]crown-6)]⁺, forming an ion pair with two carbon atoms in the naphthalene rings of different anionic scaffolds. Furthermore, each chromium center in the dinuclear sandwich is coordinated by both η⁴ and η⁶ naphthalene ligands. Based on the analysis of EPR spectra recorded at 298 and 77 K, it was proposed that the spin is strongly localized in the π-aromatic system, but there is also a significant contribution of the metal orbitals in the SOMO. Altogether, the analytical data indicate a certain degree of axial symmetry. Therefore, it was concluded that the two chromium centers in the sandwich compound do not form strong Cr–Cr interactions. The mass balance of the obtained material does not entirely account for the amounts used in the synthetic protocol and, consequently, two possible chemical reactions were proposed. The more feasible path is shown in Scheme 19, in which [K([18]crown-6)][C₁₀H₈] acts as the reducing agent and displaces the amide ligands on chromium, which are sequestered by excess of the Lewis acidic chromium reagent. Consequently, the authors discussed the displacement of amide leaving groups for delivery of a metal center as a potentially generalizable method for the synthesis of low-valent compounds.¹⁹⁵

In 2005, (electro)chemical studies showed that the manganese complexes [(η⁶-polyarene)Mn(CO)₃]⁺ [polyarene = naphthalene (168), 1,4-Me₂naphthalene (169)] undergo full and reversible two-electron reduction to afford [(η⁴-polyarene)Mn(CO)₃]⁻ (170–172).⁸⁰ The series of complexes 170–172 was obtained for different polyarenes [naphthalene (170), 1,4-dimethylnaphthalene (171) and phenanthrene (172), see Scheme 20a for the synthesis of these examples) by reaction of the corresponding complexes 168 and 169 with an excess of cobaltocene (Cp₂Co). In contrast, the reaction between [(η⁶-naphthalene)Mn(CO)₃]⁺ (168) or [(η⁶-1,4-Me₂naphthalene)Mn(CO)₃]⁺ (169) with only one equivalent of Cp₂Co afforded the bimetallic compounds [(η⁴,η⁶-1,4-R₂naphthalene)Mn₂(CO)₅] [R = H (173) or Me (174)], which contain Mn–Mn bonds. For the acenaphthene and 1,2,3,6,7,8-hexahydropyrene analogues of 168 and 169, different reaction temperatures afforded distinct products in the presence of an excess Cp₂Co: whereas at low temperatures the η⁴-polyarene complex is obtained [polyarene = acenaphthene (175), hexahydropyrene (176)], the reduction at room temperature yields the corresponding bimetallic compounds with the polyarenes coordinated in the η⁴,η⁶-fashion (177 and 178). While complex [(η⁴,η⁶-1,4-Me₂naphthalene)Mn₂(CO)₅] (174) is relatively stable toward carbon monoxide (CO) at room temperature, in the presence of ferrocenium ions, a rapid oxidative activation occurs with rupture of the Mn–Mn bond, generating the zwitterionic syn-facial homonuclear product 179 (Scheme 20b). Furthermore, reduction of 168 in the presence of [(η⁶-naphthalene)FeCp]⁺ (180) leads to the formation of the zwitterionic heteronuclear anti-facial η⁴,η⁶-naphthalene bimetallic compound 181 (Scheme 20b). It was proposed that the formation of both the homo- and the heteronuclear bimetallic compounds 179 and 181 is the result of facilitated displacement of the polyarene ligand in the cationic precursor by the nucleophilic fragments in the in situ generated reduced species, 170–172.⁸⁰

Scheme 20. Synthesis of (Polyarene)(carbonyl) Complexes of Manganese, Affording Mono- and Bimetallic Compounds⁸⁰.

2.2.4.3. Iron

The synthesis of bis(1,2,3,4-η⁴-anthracene)cobaltate(−I) (14) (see Scheme 1, above) and its iron analogue bis(1,2,3,4-η⁴-anthracene)ferrate(−I), [Fe(η⁴-C₁₄H₁₀)₂]⁻ (52), and the follow-up chemistry done with these as precursors, are two remarkable examples of the versatility of metalates.^67,179 The Fe(−I) compound 52 is a paramagnetic 17-valence-electron (VE) compound, unlike its Co(−I) analogue 14 which is diamagnetic, having 18 VE. The synthetic procedure used to obtain 52 is analogous to that of cobaltate 14 described in section 2.1, and consists of the reduction of iron(II) bromide by potassium anthracenide, as depicted in Scheme 21a. Single-crystal X-ray diffraction of 52 showed that its structure is also very similar to that of 14.

Scheme 21. (a) Synthesis and (b) Initial Reactivity Screening of Bis(1,2,3,4-η⁴-anthracene)ferrate(−I) (52) by the Group of Ellis¹⁷⁹.

Initial investigations of the reactivity of 52 (Scheme 21b) demonstrated the labile character of the coordinated polyarene ligands in ligand-substitution reactions. Complete ligand displacement was observed upon treatment of 52 with carbon monoxide or 1,3-butadiene, yielding the previously reported Fe(−I) complex, [Fe₂(CO)₈]^2– (82),^150−152 and the first 17-electron homoleptic 1,3-butadiene complex, [Fe(η⁴-C₄H₆)₂]⁻ (182), respectively. Slightly different behavior was observed upon reaction with 1,5-cyclooctadiene, where only one anthracene ligand was substituted even when an excess of 1,5-cyclooctadiene was used, affording the heteroleptic complex [Fe(η⁴-C₁₄H₁₀)(η⁴-cod)]⁻ (183). The naphthalene analogue of 183, [Fe(η⁴-C₁₀H₈)(η⁴-cod)]⁻ (184) was later reported by the same group, and obtained by direct reduction of FeBr₂ with KC₁₀H₈ (3 equiv) in the presence of excess 1,5-cyclooctadiene in THF, at −78 °C.²⁰⁰ Furthermore, the interaction of 52 with donor ligands stronger than cod resulted in oxidation to give Fe(0) complexes. For example, the reaction with P(OMe)₃ afforded the complex [Fe(η⁴-C₁₄H₁₀){P(OMe)₃}₃] (185) and the salt of anthracene radical anion, [K([18]crown-6)(thf)₂][C₁₄H₁₀].¹⁷⁹

The influence of the electron density of the Fe(−I) center was evidenced by the reaction of 52 with 2,2′-bipyridine (bpy), affording the homoleptic complex [Fe(bpy)₃]⁻ (186),^201−204 which bears reduced bpy ligands.²⁰⁵ The reduced nature of the bpy ligands in 186, and similar species with the related chelating pyridines tpy (2,2′:6′,2′′-terpyridine) and tbpy (4,4′-di-tert-butyl-2,2′-bipyridines), originally formulated as [Fe^–I(bpy⁰)₃]^1– or [Fe⁰(bpy⁰)₃]⁰,^201−204 was confirmed through a combination of spectroscopic, crystallographic, electro-, and magnetochemical studies, EPR, and theoretical calculations.²⁰⁶ As mentioned in section 2.2.3.2, an evaluation of the interaction between the anthraceneferrate 52 and CNXyl revealed that this compound was not a suitable precursor for the synthesis of isocyanoferrates due to oxidation of the Fe(−I) center to Fe(0) to form Fe(CNXyl)₅ (54), possibly via an intermediate Fe(0) polyarene-isocyano complex, [Fe(η⁴-C₁₄H₁₀)(CNXyl)₃] (187), which was structurally characterized.²⁰⁷

Wolf and co-workers also used the bis(anthracene)metalates(−I) [M = Fe (52); Co (14)]^67,179 as a platform to synthesize a variety of low-valent complexes. For example, a metal-mediated dimerization of phosphaalkynes was achieved at such complexes (Scheme 22, top).^208−210 The reaction of [M(η⁴-C₁₄H₁₀)₂]⁻ with four equivalents of the corresponding phosphaalkyne results in the cyclodimerization of the phosphaalkyne and the formation of either the open-shell sandwich complex [Fe(1,3-P₂C₂tBu₂)₂]⁻ (188)²⁰⁹ or the 18-electron compound [Co(1,3-P₂C₂R₂)₂]⁻ [R = tBu (189), tPent (190), Ad (191)].²⁰⁸ The molecular structures of these compounds feature coplanar 1,3-diphosphacyclobutadiene rings in a staggered orientation (i.e., the two P₂C₂R₂ rings are rotated, with respect to one another, by 90°).^208−210 Reactivity of anions 188 and 189–191 is described in section 3.5 (vide infra).

Scheme 22. Divergent M(−I)-Mediated Phosphaalkyne (M = Fe, Co)^208−210 and Alkyne (M = Fe)²¹¹ Cyclo-oligomerizations.

By contrast, reaction of the metalate 52 with diphenylacetylene afforded the unusual hexaphenylbenzene complex [K([18]crown-6)(thf)₂][Fe(η⁶-C₆Ph₆)(η²-C₂Ph₂)] (192), formed via alkyne cyclotrimerization (Scheme 22, down).²¹¹ While phosphaalkynes are often invoked as analogues of alkynes, the reactivity of 52 toward both substrates showed marked differences in their cyclo-oligomerization reactions. DFT calculations performed on the simplified model complexes of the isomers [Fe(η⁴-C₄Me₂H₂)₂]⁻ (188C′), [Fe(η⁶-C₆Me₃H₃)(η²-C₂MeH)]⁻ (192′), [Fe(η⁴-P₂C₂Me₂)₂]⁻ (188′), and [Fe(η⁶-P₃C₃Me₃)(η²-PCMe)]⁻ (192P′), products of alkyne or phosphaalkyne cyclo-oligomerization, revealed that their relative stabilities are in accordance with the experimental results: the Fe^–I-mediated cyclotrimerization is the favored reaction pathway for alkynes, whereas for phosphaalkynes the cyclodimerization is preferred. The divergent behavior of alkyne and phosphaalkyne cyclo-oligomerizations was correlated with the relative thermodynamic stabilities of the corresponding arene vs cyclobutadiene complexes and shown to be intimately dependent on the incorporation of phosphorus atoms into the ligand structures. Furthermore, the theoretical calculations and the characterization by physical methods (EPR, magnetic moment, X-ray diffraction) demonstrated that complex 192 is paramagnetic with one unpaired electron. The positive spin density is mainly localized on the Fe atom, indicative of a d⁹ configuration. However, the electronic structure might differ from a formal [(L⁰)Fe^–I(L⁰)] since both of its ligands are potentially redox active.²¹¹

Inspired by the contributions of the Jonas group,^8,900,68,69 Fürstner and co-workers reported on the synthesis of the heteroleptic Cp^RFe-alkene metalates [Li(tmeda){CpFe(C₂H₄)₂}] (193), [Li(chelate){CpFe(cod)}] [chelate = tmeda (194), dme (195)], and [Li(tmeda){Cp*Fe(C₂H₄)₂}] (196), featuring zerovalent iron centers.^212−214 As shown in Scheme 1 (reaction type 1a), under reducing conditions, the Cp rings in metallocenes can be successively displaced. The group of Fürstner reported that performing a reductive cleavage on ferrocene (197) under an atmosphere of ethylene yields the heteroleptic Fe(0) half-sandwich complex 193 (Scheme 23a), which after prolonged stirring reacts further to form the homoleptic ferrate [Li₂(TMEDA)₂{Fe(η²-C₂H₄)₄}] (15). Performing the reaction in the presence of 1,5-cyclooctadiene (cod) instead of ethylene leads to the formation of the complexes [Li(chelate){CpFe(cod)}] (194, 195). Furthermore, [Li(dme){CpFe(cod)}] (195) reacts with Ph₃CCl to afford the one-electron oxidation product [CpFe(cod)] (198). A similar synthetic approach, starting from precursor [Cp*FeCl(tmeda)] (83), yields complex [Li(tmeda){Cp*Fe(C₂H₄)₂}] (196) (Scheme 23b). Using these protocols (Scheme 23), various cyclopentadienyl-alkene metalates can be obtained in a multigram scale (up to 85 g, for 195).^212−214 The authors also demonstrated the ability of these metalates to serve as intermediates in the formation of challenging scaffolds at transition metals. For instance, the Fe(I) complex [Cp*Fe(C₂H₄)₂] (199), obtained via one-electron oxidation as for 198, promotes the cyclodimerization of diphenylacetylene to yield the neutral cyclobutadiene iron(I) complex [Cp*Fe(η⁴-C₄Ph₄)] (200, Scheme 23b).²¹³

Scheme 23. Synthesis of Low-Valent Heteroleptic Cyclopentadienyl-alkene Metalates by the Group of Fürstner^212−214.

In the search for further examples of heteroleptic ferrates, the Wolf group reported the complex [K([18]crown-6){Cp*Fe(η⁴-C₁₀H₈)}] (201, Cp* = C₅Me₅, via reduction of “Cp*FeCl”, prepared in situ from Cp*Li and [FeCl₂(thf)_1.5], with 2 equiv of potassium naphthalene, in the presence of 1 equiv of crown ether (Scheme 24a).^215−217 Unlike the naphthaleneferrate 184, 201 is a diamagnetic 18-electron compound with a formally iron(0) center, obtained as a contact ion pair with the countercation [K([18]crown-6)]⁺. Furthermore, it is analogous to the species previously described by the Jonas group, [Li(thf)₂][CpFe(η⁴-naphthalene)] (202, Cp = C₅H₅).⁹⁰⁰ The naphthalene precursor 201 behaves as an efficient Cp*Fe^– synthon in its reaction with diphenylacetylene, mediating its cyclodimerization to form the anionic 18-electron cyclobutadiene iron(0) complex (203, Scheme 24b), analogous to the 17-electron Fe(I) compound 200 (Scheme 23b).²¹³ Also, when 201 is treated with tBuC≡P, a cyclodimerization of the phosphaalkyne occurs, affording the 1,3-diphosphabutadiene iron(0) sandwich compound 204 (Scheme 24b, top).²¹⁵ Follow-up studies demonstrated that the cyclodimerization of bis(pyridyl)acetylene (pyridyl = 2-pyridyl, 3-pyridyl, 4-pyridyl) at 201 was also possible, generating potentially redox-active multitopic substituted cyclobutadiene ligands with potential to form supramolecular assemblies.^218,219

Scheme 24. (a) Synthesis and (b) Reactivity toward tert-Butylphosphaalkyne and Substituted Acetylenes of Complex [K([18]crown-6)][Cp*Fe(η⁴-C₁₀H₈)] (201)^215,218,219.

The complexes [K([18]crown-6){Cp*Fe(η⁴-C₄py₄)}] [py = 2-pyridyl (205), 3-pyridyl (206), 4-pyridyl (207)] present structural differences, as determined by X-ray diffraction; while the 2-pyridyl isomer 205 is monomeric with a contact ion pair,²¹⁸ the 3-pyridyl (206) and 4-pyridyl (207) analogues form coordination polymers with alternating cation–anion sequences.²¹⁹ The neutral complex {Cp*Fe[η⁴-C₄(2-py)₄]} (208) was also obtained after oxidizing 205 with [CuBr(tht)] (1 equiv; tht = tetrahydrothiophene). Complexation of ZnCl₂ or FeCl₂ by 205 afforded 1:1 or 2:1 adducts [K([18]crown-6){Cp*Fe[C₄(2-py)₄]}(ZnCl₂)] (209), [K([18]crown-6){Cp*Fe[C₄(2-py)₄]}(FeCl₂)] (210), and [K([18]crown-6)(thf)][Cp*Fe[C₄(2-py)₄](ZnCl₂)₂] (211) in which the anionic complex acts as a bidentate chelate (see Figure 6).²¹⁹

Figure 6.

Iron(II) and zinc(II) adducts of complex [K([18]crown-6){Cp*Fe[η⁴-C₄(2-py)₄]}] (205).

[K([18]crown-6){Cp*Fe(η⁴-C₁₀H₈)}] (201) also acts as a two-electron reducing agent toward the imidazolium salt [L^DippPCl₂]⁺ (212, L^Dipp = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene); see Scheme 25) to afford the diphosphorus cation [L^Dipp₂P₂Cl₃]⁺ (213) and the related [L^Dipp₂P₂Cl]⁺ (214).²²⁰ Along with the mixture of the phosphorus cations, the oxidized cationic complex [Cp*Fe(η⁶-naphthalene)]Cl (215) was identified. The relative ratio in which the phosphorus cations were obtained depended strongly on the reaction temperature and order of addition of the reagents. While 213 is the major product (>90% of the total P content) after 8 h when 201 is slowly added to a refluxing solution of [212]OTf in THF, with decreasing temperature (65 to −30 °C) the ratio shifts progressively in favor of 214, which becomes the major product at −90 °C, in a ratio 213:214 = 1:8. Compound 213 is also efficiently synthesized by reduction of [212]OTf with sodium.²²⁰

Scheme 25. Reduction of the Imidazolium Salt [L^DippPCl₂]OTf ([212]OTf) by K([18]crown-6)][Cp*Fe(η⁴-C₁₀H₈)] (201)²²⁰.

In many cases, it is difficult to provide a straightforward description of the electronic structure of transition metalates due to the redox noninnocence of their stabilizing ligands. To gain insight into the electronic structure of heteroleptic ferrates, Wolf and co-workers applied a combination of experimental analyses (NMR, UV–vis, ⁵⁷Fe Mössbauer spectroscopy, electron paramagnetic resonance (EPR) spectroscopy, magnetic susceptibility measurements and single-crystal X-ray crystallography) and theoretical calculations (DFT)²¹⁷ to investigate the behavior of the anionic complexes [K([18]crown-6)][Cp*Fe(η⁴-polyarene)] [polyarene = naphthalene (201),^215−217 anthracene (216); Cp* = C₅Me₅],²¹⁷ and the neutral analogues [Cp*Fe(η⁴-polyarene)] [polyarene = naphthalene (217) anthracene (218)]. Although compound 216 can be prepared via a procedure similar to that used for 201 (Scheme 24a) by using anthracene instead of naphthalene, this synthetic pathway is hampered by difficulties in the purification of the product. Consequently, an alternative protocol was devised, consisting of the exchange of the naphthalene ligand on 201 with anthracene (Scheme 26a). It was also observed that the reduction of “Cp*FeCl” with 2 equiv of potassium anthracenide in the absence of crown ether affords the neutral compound 218 (Scheme 26b). The naphthalene complex 217 cannot be similarly obtained and was synthesized via oxidation of the parent species 201 with AgOTf (Scheme 26a).

Scheme 26. Preparation of Low-Valent Naphthalene and Anthracene Iron Complexes by Wolf and Co-workers²¹⁷.

The redox properties of these complexes were studied through cyclic voltammetry and spectroelectrochemical analysis, revealing that the neutral complexes 217 and 218 can be both reversibly reduced to the monoanions [Cp*Fe(η⁴-polyarene)]⁻ (201, 216) and reversibly oxidized to the cationic species [Cp*Fe(η⁶-polyarene)]⁺ (217⁺, 218⁺). Density functional theory (DFT) calculations indicated reduced orbital charges and spin densities in the series of complexes [Cp*Fe(polyarene)]^−/0/+ (polyarene = naphthalene, anthracene). Altogether, the structural, spectroscopic and DFT studies showed that the formal oxidation state of the iron center in such species varies from Fe(0) to Fe(II). However, the spectroscopic oxidation state in all cases is close to Fe(II), with the sole exception of compound 217, which possesses significant Fe(I) character. In any case, clearly, the naphthalene and anthracene motifs in these complexes behave predominantly as redox noninnocent ligands.²¹⁷

In 2012, independent reports by Wolf and co-workers,²²¹ and by the group of Yoshizawa and Tatsumi,²²² described the synthesis of the homodinuclear compounds [Cp*Fe(μ–η⁴:η⁴-C₁₀H₈)FeCp*] (219, Cp* = C₅Me₅) and [Cp*Fe(μ–η⁴:η⁴-C₁₄H₁₀)FeCp*] (220). An analogue of complex 219 featuring the unsubstituted cyclopentadienyl ligand, [CpFe(μ–η⁴:η⁴-C₁₀H₈)FeCp] (219′), was previously described, but its characterization was not fully reported.⁹⁰⁰ The procedure for the synthesis of [Cp*Fe(μ–η⁴:η⁴-polyarene)FeCp*] consists of a salt metathesis between [Cp*FeCl(tmeda)] (83) and the anionic Fe(0) complex 201 (Scheme 27a), in the case of 219,²²¹ or of 1:1 reduction of [FeCl₂(thf)_1.5] by potassium anthracenide in the presence of stoichiometric equivalents of LiCp*, for 220 (Scheme 27b). Yoshizawa and Tatsumi obtained compound 219 via direct reduction of 83 with potassium naphthalenide (2 equiv; see Scheme 27a).²²² These complexes feature 17-electron, formally Fe(I) centers, and the structural characterization of 219 confirmed that two Cp*Fe moieties bind to the opposite faces of the naphthalene bridge. A similar transoid arrangement was found for the anthracene-bridged complex 220 (Scheme 27b). Despite being 17e^– species and having one unpaired electron per metal center, the complexes are diamagnetic in the ground state, due to the strong magnetic coupling of both Fe(I) centers efficiently mediated by the polyarene bridge, as demonstrated via electrochemical, spectroscopic, and DFT investigations.^221,222 Other early reports described, contrastingly, paramagnetic [Cp*Fe(μ-polyarene)FeCp*] compounds (polyarene = dihydrophenanthrene, triphenylene, phenanthrene, pyrene) with weakly coupled iron centers.^223,224 For 219 and 220, two well-separated one-electron oxidation steps were identified after studying their redox behavior by cyclic voltammetry and UV–vis spectroscopy. Reversible oxidations to the monocations [Cp*Fe(μ-polyarene)FeCp*]⁺ [polyarene = naphthalene (221), anthracene (222)] and the dications [Cp*Fe(μ-polyarene)FeCp*]²⁺ [polyarene = naphthalene (223), anthracene (224)] as well as reduction to the monoanions [Cp*Fe(μ-polyarene)FeCp*]⁻ [polyarene = naphthalene (225), anthracene (226)] were observed.^221,222 Chemical oxidation with [Cp₂Fe]PF₆ allowed for the isolation of the mixed-valent Fe^I–Fe^II cationic compounds 221 and 222 as the BAr^F₄^– salts. The molecular structure of 221 revealed that, upon oxidation, the naphthalene ligand changes its hapticity from η⁴ in complex 219 to η⁶ in the cationic compound. This hapticity change was assumed to account for the high thermodynamic stability of the mixed-valence species with respect to disproportionation.²²²

Scheme 27. Synthesis of Low-Valent Redox-Active Polyarene-Bridged Homo- And Heterodinuclear Complexes^221,222,225.

A slightly modified procedure was employed for the synthesis of dissymmetrical naphthalene-bridged complexes [Cp‴Fe(μ-C₁₀H₈)FeCp*] (227, Cp‴ = η⁵-C₅H₂-1,2,4-tBu₃, Cp* = η⁵-C₅Me₅), [Cp‴Fe(μ-C₁₀H₈)RuCp*] (228) and the homoleptic heterodinuclear compound [Cp*Fe(μ-C₁₀H₈)RuCp*] (229), as shown in Scheme 27c.²²⁵ The main modification of this protocol is that “Cp^RFeCl” is initially generated in situ from [FeCl₂(thf)_1.5] and KCp^R (Cp^R = Cp‴, Cp*), then treated with potassium naphthalenide (KC₁₀H₈) to form the iron(0) species K[Cp^RFe(C₁₀H₈)], onto which either [Cp*FeCl(tmeda)] (83, tmeda = tetramethylethylene-1,2-diamine) or [Cp*RuCl]₄ (230) are added to obtain the final dissymmetrical FeFe or FeRu complexes. 227–229 are structurally analogous to 219 and 220: diamagnetic compounds with coordinating Cp^RM fragments on opposite faces of the bridging naphthalene ligand.

Cyclic voltammetry and UV–vis spectroelectrochemistry studies showed the feasibility of oxidizing complexes 227 and 228 to the monocations [Cp‴Fe(μ-C₁₀H₈)FeCp*]⁺ (231) and [Cp‴Fe(μ-C₁₀H₈)RuCp*]⁺ (232). These paramagnetic complexes are accessible as hexafluorophosphate salts after chemical oxidation with [Cp₂Fe]PF₆. Further oxidation to the corresponding dicationic species was observed via cyclic voltammetry, and proved reversible only for 232 and for the homoleptic [Cp*Fe(μ-C₁₀H₈)RuCp*]⁺ (233). The structural and spectroscopic data are supported by the DFT calculations. Substituting the Cp* ligand by Cp‴ produces only modest perturbations of the electronic structures of the complexes, which do not seem to significantly change the compositions and shapes of their frontier orbitals. According to the DFT calculations, the redox behavior of complexes 227 and 228 seems to be significantly affected by the contributions from the Cp‴Fe motif and the naphthalene bridging ligand. By contrast, substitution of a Cp*Fe motif in complex 227 by a Cp*Ru scaffold in 228 seems to have little influence. Thus, the redox potentials of the complexes appear to respond more strongly to the changes on the Cp^R ligand than to the substitution of the Fe center by Ru.²²⁵

In the 1980s, an in situ generated catalytic system, composed by phenyl lithium and the iron–sulfur cluster [Bu₄N]₂[Fe₄S₄Cl₄] (234) was developed for the hydrogenation of stilbenes and carbonyl compounds.^226,227 In an attempt to elucidate the nature of the catalytically active species, Hu and co-workers synthesized the unsupported Fe(I) organoferrate [Bu₄N][(η⁶-biphenyl)Fe(Ph)₂] (235) by treating the cluster 234 with PhLi (8 equiv; Scheme 28).²²⁸ Although the structure of 235 is very unusual, this complex was not a competent catalyst in the hydrogenation of olefins, nor did it react with H₂. Nonetheless, 235 reacts with 2-bromopyridine, p-bromoanisole, or 1-bromobutane to afford the corresponding C–C coupling products and free biphenyl in low yield.

Scheme 28. Synthesis of an Fe(I) Organoferrate by Hu and Co-workers²²⁸.

2.2.4.4. Cobalt

The versatility of metalate complexes as sources of “naked-metal atom” reagents was further demonstrated through the synthesis of the ternary intermetalloid clusters [Co@Sn₆Sb₆]^3– (236), [Co₂@Sn₅Sb₇]^3– (237), and [Ni₂@Sn₇Sb₅]^3– (238) (see Figure 7 for a representation of the structures of the cobalt clusters) by Dehnen and co-workers.²²⁹ To obtain clusters 236–238, the classic cobaltate [Co(η⁴-cod)₂]⁻ (239, as [K(thf)_x]⁺ salt)⁷ or the Ni(0) precursor [Ni(cod)₂] (240, cod = 1,5-cyclooctadiene),^4,5 in conjunction with [K([2.2.2]cryptand)]₂[Sn₂Sb₂] (241), served as starting materials. The cobalt-containing clusters cocrystallized in the salt [K([2.2.2]cryptand)]₃[(236)_0.83(237)_0.17]·2dmf·2toluene. Crystals allowed for an adequate characterization of the major component 236, while for the minor component 237 the structure was not satisfactory. However, anion 237 was characterized in the salt [K([2.2.2]cryptand)]₃[237], obtained after performing a similar reaction in a mixture of ethylenediamine (en) and DMF at 60 °C, and growing crystals from a filtered solution (in ethylenediamine) of the precipitated powder. The 12-vertex cluster 236 has a fused square-antiprism structure, and is asymmetrically occupied by a Co^– center, constituting a rare example of a compound in which the inner transition metal atom is not located at the center of the cluster.

Figure 7.

Representation of the structures of the anionic clusters [Co@Sn₆Sb₆]^3– (236) in [K([2.2.2]cryptand)]₃[(236)_0.83(237)_0.17]·2dmf·2toluene and [Co₂@Sn₅Sb₇]^3– (237) in [K([2.2.2]cryptand)]₃[237], as determined by X-ray diffraction.²²⁹

Although certainly unusual, DFT calculations not only supported the molecular structure determined by X-ray diffraction but also revealed that other structures with the same stoichiometry are not feasible. The cobalt and nickel clusters 237 and 238 are isoelectronic and isostructural (a representation of the structure of 237 is shown in Figure 7), despite their different compositions. Calculated natural atomic charges for the three clusters support the presence of formal Co^– centers for 236 and 237, and of the isoelectronic Ni(0) centers for 238. Therefore, it was concluded that the three Sn/Sb cages obtained are 56-valence-electron species (C_4v-type topology). The total valence electron count might then be the crucial factor controlling the types of structures obtained. Preliminary ¹¹⁹Sn NMR studies in solution were also reported.²²⁹

Compound [Co(η⁴-cod)₂]⁻ (239) was also used for the synthesis of the unusual Co₂Sn₂ heterobimetallic p-block/d-block element cluster [Ar^Dipp2SnCo]₂ [242, Ar^Dipp2 = C₆H₃-2,6(C₆H₃-2,6-iPr₂)₂; Scheme 29].²³⁰ The cyclic rhomboidal Co₂Sn₂ core in 242 is obtained after reaction between cobaltate 239 and the tin compound [Ar^Dipp2Sn(μ-Cl)]₂ (243). 242 has strong metal–metal bonds between the tin and cobalt atoms, but also features a weaker tin–tin interaction, as evidenced in the molecular structure of the cluster. Furthermore, the cobalt atoms are η⁶-coordinated by one of the flanking 2,6-diisopropylphenyl units in each terphenyl ligand, and a strong cobalt-arene interaction is indicated by the short cobalt–centroid distance (1.560(1) Å). This suggests strong d-π* backbonding between the cobalt center and the coordinated arenes. DFT calculations support the existence of strong intermetallic bonds in the Co₂Sn₂ core and a weaker Sn–Sn interaction.

Scheme 29. Synthesis of the Cyclic Co₂Sn₂ (242) Cluster Using the Co^– Synthon [Co(η⁴-cod)₂]⁻ (239)²³⁰.

Whereas treatment of 242 with O₂ and CO yielded intractable mixtures, reaction with white phosphorus proved highly selective. Insertion of the P₄ tetrahedron into the Co₂Sn₂ core led to the formation of a catena-P₄ complex, [Ar^Dipp2₂Sn₂Co₂P₄] (244, Ar^Dipp2 = C₆H₃-2,6-{C₆H₃-2,6-iPr₂}₂), which constitutes the first molecular cluster composed of phosphorus, cobalt and tin atoms. The molecular structure confirmed a functionalized P₄-chain, resulting from migration of a tin-bound terphenyl substituent to one P-atom.²³⁰

Among the polyarene complexes reported by Ellis,^67,179 naphthalene metalates of the late transition metals proved elusive until 2006.²³¹ The first naphthalenecobaltate(−I), [Co(η⁴-C₁₀H₈)(η⁴-cod)]⁻ (245, see Scheme 30), analogous to [Fe(η⁴-C₁₄H₁₀)(η⁴-cod)]⁻ (183), was obtained by reduction of CoBr₂ by potassium naphthalenide^85,231 in the presence of a stoichiometric amount of 1,5-cod, in a procedure similar to that used for complexes [M(η⁴-C₁₄H₁₀)₂]⁻ [M = Fe (52); Co (14)].^67,179

Scheme 30. Synthesis of Naphthalenecobaltate(−I), [Co(η⁴-C₁₀H₈)(η⁴-cod)]⁻ (245).

Complex 245 is also accessible from cobaltocene, and the authors proposed that, according to their observations, both synthetic routes seem to have a common intermediate, presumably a homoleptic naphthalenecobaltate analogous to the anthracenecobaltate 14. Such a species has proven difficult to isolate in pure form.⁸⁵ However, a mixture of yellow and red-black crystals suitable for X-ray diffraction was obtained after reaction of the known [Co(C₂H₄)₄]⁻ complex (246, as the [K([18]crown-6)]⁺ salt)^2,232,233 with naphthalene (3 equiv; THF, 20 °C). The yellow crystals were identified as the starting material 246, while the red-black ones were characterized as the unusual “triple-salt” [K([18]crown-6)]₃[Co(η⁴-C₁₀H₈)(C₂H₄)₂]₂[Co(η⁴-C₁₀H₈)₂] (247), which provided structural evidence of the new metalate [Co(η⁴-C₁₀H₈)(C₂H₄)₂]⁻ and of the long sought-after homoleptic naphthalene metalate [Co(η⁴-C₁₀H₈)₂]⁻.²³¹

The reactivity of compounds 14 and 245 was examined by the same group (Scheme 31).⁸⁵ An early investigation of the reactivity of 14 was reported in 2002⁶⁷ and was later broadened to include a wider range of substrates.⁸⁵ Alternative and improved methods for the synthesis of classic cobaltates,^2,234−239 such as [Co(η⁴-C₄H₆)₂]⁻ (248), [Co(PF₃)₄]⁻ (249), [Co(PR₃)₄]⁻ [R = OiPr (250) OMe (251)], or [Co(bpy)₂]⁻ (252),^205,240 were developed, involving reactions of precursor 14 with 1,3-butadiene, PF₃, P(OiPr)₃, or 2,2′-bipyridine, respectively. Crystallographic analyses were also performed for complexes 248–250, and for the classic carbonyl compound [Co(CO)₄]⁻ (95, characterized as the [K([2.2.2]cryptand)]⁺ salt).²⁴¹ Additionally, 14 reacted with PMe₃, albeit with different results to those previously obtained with 1,2-bis(diphenylphosphino)ethane (dppe).⁶⁷ Instead of an analogue of the homoleptic compound [Co(dppe)₂]⁻ (253),⁶⁷ the interaction of cobaltate 14 with PMe₃ yielded the known Co(0) species, [Co(η⁴-C₁₄H₁₀)(PMe₃)₃] (254)²⁴² and the anthracene radical anion salt, [K([18]crown-6)(thf)₂][C₁₄H₁₀]. The authors proposed that complex 254 might be formed through an unidentified Co(−I) intermediate, which, due to its strongly reducing character, is oxidized to the Co(0) complex while reducing coordinated or free anthracene to C₁₄H₁₀^–. This behavior might parallel that observed by Klein and co-workers, who reported that another (trimethylphosphine)cobaltate, [Co(PMe₃)₄]⁻ (255), prepared by reduction of [Co(PMe₃)₄] (256) with alkali metals (Li, Na, K), acted as an extraordinarily strong reducing agent.^242,243

Scheme 31. Reactivity of [Co(η⁴-C₁₄H₁₀)₂]⁻ (14) and [Co(η⁴-C₁₀H₈)(η⁴-cod)]⁻ (245) in Contrast: Synthesis of a Family of Metalates from “Naked-Metal Atom” Cobaltates^85,245.

Conditions: (i) Et₂O, 20 °C; (ii) THF, 20 °C; (iii) −78 to 20 °C.

The synthesis of a family of mixed arene/alkene cobaltates was also proposed, and the viability of the reactions using complexes 14 or 245 was compared (Scheme 31).⁸⁵ The heteroleptic complexes [Co(η⁴-C₁₄H₁₀)(η⁴-cod)]⁻ (257), analogous to 245, and [Co(η⁴-C₁₄H₁₀)(C₂H₄)₂]⁻ (258) were obtained from 14 after treatment with an excess 1,5-cod or ethylene, respectively.⁸⁵ When the reduction of CoBr₂ to form 245 is attempted in the presence of an excess 1,5-cod, instead of the heteroleptic complex, the reaction affords the classic cobaltate [Co(η⁴-cod)₂]⁻ (239),⁷ which can also be obtained by direct reaction of 245 with 1,5-cod. Similarly, while 245 remains unchanged when treated with anthracene at room temperature, upon heating the reaction mixture to 60 °C, anthracene displaces the naphthalene ligand in 245 to yield 257. However, the synthesis of 257 using 14 as the starting material remains more efficient than the ligand exchange reaction at higher temperature. Displacement of the anthracene ligand from precursor 14 appears to be facile, explaining the formation of homoleptic complexes by ligand substitution. By contrast, the coordinated cod ligand on 245 appeared to be substantially less labile. As a result, a variety of mixed-ligand cyclooctadiene complexes are accessible by substitution of the naphthalene ligand of 245, e.g., [Co(η⁴-C₄H₄R₂)(η⁴-cod)]⁻ [R = H (259), R = Me (260)] [Co(bpy)(η⁴-cod)]⁻ (261),²⁴⁴ and [Co(η⁴-cod)(PMe₃)₂]⁻ (262). Detailed examination of the bond distances on the bpy scaffold on the solid-state structure of 261 revealed the presence of a coordinated bpy radical anion, bpy^•–. Thus, 261 could be described as a Co(0) species, [Co(bpy^•–)(η⁴-cod)]⁻.⁸⁵ Precursor 245 has also served as starting material for the synthesis of the bis(pyrene)metal complex [Co(η²-pyrene)₂(η⁴-cod)]⁻ (263, as [K([2.2.2]cryptand)]⁺ salt), which was structurally characterized.²⁴⁵ Unlike the Ti(−I) or M(0) (M = V, Nb) sandwich complexes 153 or 154,¹⁸¹263 showcases pyrene ligands coordinated in η²-fashion.²⁴⁵

Ellis and co-workers studied the formation of alkylisocyanometalates using precursors 14 and 245.⁸⁵ While FT-IR spectroscopy of the interaction of 14 with CNtBu suggested the possible formation of a homoleptic alkylisocyanometalate, the product could not be isolated (Scheme 31). However, its formation was corroborated by the isolation of the triphenyltin derivative, (Ph₃Sn)Co(CNtBu)₄ (264), similar to trans-[Fe(CNtBu)₄(SnPh₃)₂] (57). In a similar vein, 245 reacted with CNtBu (4 equiv) at −78 °C to afford the thermally unstable [K([18]crown-6)(thf)₂][Co(CNtBu)₄] (86), the formulation of which was confirmed through structural characterization.

The cobaltates 14, 239, and 245 also served as starting materials for the synthesis of additional examples of homoleptic and heteroleptic alkene cobaltates.²⁴⁶ The heteroleptic complexes [K([18]crown-6)][Co(η⁴-cod)(η²-styrene)₂] (265) and [K([18]crown-6)][Co(η⁴-dct)(η⁴-cod)] (266; dct = dibenzo[a,e]cyclooctatetraene, cod = 1,5-cyclooctadiene), and the homoleptic [K(thf)₂][Co(η⁴-dct)₂] (267), were obtained by ligand exchange with the corresponding alkenes, using either complex 239 or 245 as a starting material (see Scheme 32a). The highly air-sensitive heteroleptic species 265 could be obtained from either 245 in the presence of a slight excess of styrene, or from 239 after reaction with a large excess of the olefin and in the presence of [18]crown-6. The latter procedure afforded 265 in a higher yield than the former. Similarly, addition of dct to 245 led to the formation of 266, albeit not in pure form, being invariably contaminated with the bis(dibenzo[a,e]cyclooctatetraene) complex 267. A more complex mixture of products was obtained by reaction of the bis(1,5-cyclooctadiene) cobaltate 239 with dct, furnishing again both 266 and 267, as well as unreacted 239. Such reactivity parallels behavior observed for 239 by Ellis and co-workers, in which attempted ligand exchange on Jonas’ classic cobaltate led almost invariably to heteroleptic species or unreacted starting material, even in the presence of a large excess of ligand.⁸⁵

Scheme 32. Synthesis of Heteroleptic and Homoleptic Alkene Cobaltates by Jacobi von Wangelin, Wolf, and Co-workers²⁴⁶.

The homoleptic complex 267 is obtained in pure form by reaction of 14 with dct (see Scheme 32b), albeit in low yield (19%). A significant improvement in the reaction yield (62%) was observed when styrene was added to 14 followed by the addition of dct. This observation might be indicative of a two-step substitution, in which an initially formed styrene complex analogous to 265 facilitates the formation of the homoleptic complex 267. X-ray diffraction analysis of 267 revealed that the complex is obtained as a contact ion pair with the [K(thf)₂]⁺ counterion.²⁴⁶

2.2.4.5. 4d and 5d Metalates

As mentioned above, Ellis and co-workers described the triad of group 4 M(−II) complexes [M(C₁₄H₁₀)₃]^2– and [M(C₁₀H₈)₃]^2–, including the heavier tris(polyarene)zirconates(−II) [polyarene = anthracene (148), naphthalene (151)] and tris(polyarene)hafnates(−II) [polyarene = anthracene (149), naphthalene (152)], as illustrated in Scheme 17 for the lighter titanate analogues (vide supra).^177,180 In a similar vein, the first isolable homoleptic butadienemetalates of 4d and 5d metals, [M(η⁴-C₄H₆)₃]⁻ [M = Nb (268), Ta (269)], were also reported by Ellis.²⁴⁷ In previous work, the same group had succeeded in accessing the homoleptic arene tantalates [Ta(η⁴-C₁₄H₁₀)₃]⁻ (270) and [Ta(η⁴-C₁₀H₈)₃]⁻ (271), with 271 serving as the starting material for the corresponding butadienemetalate [Ta(η⁴-C₄H₆)₃]⁻ (269, Scheme 33a).¹⁷⁸ Attempts to isolate and characterize the presumed homoleptic naphthaleneniobate(−I) [Nb(η⁴-C₁₀H₈)₃]⁻ failed. Nevertheless, the heteroleptic anion [Nb(η⁴-C₁₀H₈)₂(PMe₃)₂]⁻ (272) was obtained and used as a precursor to the niobium butadienemetalate [Nb(η⁴-C₄H₆)₃]⁻ (268, Scheme 33b). In species 272, it is believed that the PMe₃ donors are key to strengthen the metal–naphthalene bond and, consequently, provide stability to the compound with respect to the hypothetical complexes [Nb(η⁴-C₁₀H₈)₃]⁻ and [Nb(η⁶-C₁₀H₈)₂]⁻, which have remained elusive. The isolation of the unprecedented homoleptic butadiene complexes [M(η⁴-C₄H₆)₃]⁻ [M = Nb (268), Ta (269)] could serve as an inspiration to attempt the synthesis of similar compounds with other 4d and 5d transition metals in formally negative oxidation states. The [PPN]⁺ salt of the tris(naphthalene)tantalate(−I) 271 has been identified to undergo double ortho-metalation of two phenyl rings and hydrogenation of the third unit at tantalum to afford a tantalum bound 1,3-cyclohexadiene group.²⁴⁸

Scheme 33. Synthesis of Niobium and Tantalum Alkene or Arene Metalates by the Ellis Group²⁴⁷.

Girolami demonstrated that treatment of HfCl₄ (273), [TaCl₂(OMe)₃] (274), or [WCl₃(OMe)₃] (275) with ethyl lithium (LiEt) provides access to the anionic hafnium(II), tantalum(−I), and tungsten(−II) complexes [Li(tmeda)]₂[HfEt₄(C₂H₄)] (276), [Li(tmeda)]₃[TaHEt(C₂H₄)₃] (277), and [Li(tmeda)]₃[WH(C₂H₄)₄] (278), respectively (Scheme 34).²⁴⁹ The lithium organyl LiEt acted both as a reductant and an ethylene source, formed by a β-hydrogen abstraction process, as demonstrated by labeling experiments. The ethylene ligands have a significant metallacyclopropane character in all these metalates. The hydride resonance in the tantalum complex is observed at δ 1.23 ppm, while in the tungsten species the hydridic signal is observed at δ −8.21 ppm. Variable temperature NMR measurements and spin saturation transfer experiments indicate that exchange between the ethylene sites and the hydride environments for the tantalate and tungstate complexes does not take place on the NMR time scale. In the solid state, the anions 277 and 278 adopt distorted square-pyramidal geometries with the hydride group in the axial position and two of the ethylene ligands interacting with the lithium cations. The coordination environment of Hf in 276 is also distorted square-pyramidal, with the ethylene ligand in the axial position. The cations interact with the ethylene unit and with two of the ethyl groups in 276.

Scheme 34. Synthesis of Hafnium, Tantalum, and Tungsten (Ethylene)metalates by Girolami and Co-workers²⁴⁹.

2.2.5. Metallocene Anions

Cyclopentadienyl ligands and their substituted derivatives have been well-studied in organometallic chemistry. However, as discussed so far, most cyclopentadienyl metalates are heteroleptic complexes, generally CpM(alkene/arene) compounds. Until the first decade of the new millenium, very few examples of homoleptic cyclopentadienyl metalate anions had been isolated.²¹ This comes as a surprise since the electrochemical reduction of metallocenes to formally anionic species was documented decades ago.^250−257 This situation has completely changed over the last ten years, as numerous metallocene anions have become available. The remarkable development of the field has been comprehensively reviewed by Magnoux and Mills.²¹ As a result, we will focus on the most recent developments. The excellent review article by Magnoux and Mills should be consulted for earlier work, such as the synthesis and reaction chemistry of the metallocenates Na[Cp*₂Mn] and M[Cp₂Re] (M = Li, K).

In 2014, Bradley and co-workers described the first crystallographic characterization of a metallocene anion.²⁵⁸ In a previous study,²⁵⁹ the same group found that upon reduction of the Co(II)-indenyl complex [Co(η⁵-C₉H₅-1,3-(SiMe₃)₂)₂] (279) by sodium amalgam in the presence of excess vinyltrimethylsilane, instead of forming a product containing the bulky olefin, the neutral dimer [Co₂(η⁵:η⁴-C₉H₅-1,3-(SiMe₃)₂)₂] (280, Scheme 35) was obtained. It was proposed that the formation of 280 could proceed through a formally 20-valence-electron bis(indenyl)cobaltate species and, therefore, the synthesis of such an intermediate was attempted.²⁵⁸ Initially, cyclic voltammetric studies in THF indicated that the Co(II) complex 279 presented reversible one electron processes for both the oxidation to cobaltocenium and reduction to cobaltate ions (E_1/2 = −0.850 and −1.800 V, respectively, vs Fc/Fc⁺). Consequently, the 20 VE Co(I) anion [Co(η⁵-C₉H₅-1,3-(SiMe₃)₂)₂]⁻ (281, as the [Na(thf)₆]⁺ salt; Scheme 35) was isolated after chemical reduction of 279 with sodium amalgam in THF, and its structure was confirmed by crystallographic analysis. A comparison of the metric parameters of 281 and 279 shows an elongation of the Co–C bond lengths [d_CoC(279) = 2.06–2.25 Å vs d_CoC(281) = 2.09–2.46 Å], but almost unchanged C–C bond distances in the anionic compound with respect to the neutral species, suggesting a metal-based reduction event. Reactivity studies toward σ and π donors, performed on 281, as well as crossover experiments with a related indenide salt, revealed that the complex reacts by a reversible indenide ligand ejection. Mechanistically, 281 reacts via associative displacement of an indenide unit, both in the presence and in the absence of strong supporting donor ligands.²⁵⁸

Scheme 35. Synthesis of the Bis(indenyl)cobaltate 281, an Intermediate in the Formation of Dimer 280(258).

The previously observed formation²⁵⁹ of 280 was then proposed to proceed initially via associative displacement of indenide, probably through an unobserved ring slipped intermediate, which generates a 16 electron adduct [Co(η⁵-C₉H₅-1,3-(SiMe₃)₂)(L)] (L = vinyltrimethylsilane or aromatic amine). After substitution of the supporting donor ligand L by a benzo group and dimerization, 280 is formed. It was also proposed that the stabilization of the 20 VE cobaltate 281 is likely assisted by the bulkiness and less electron donating character of the silyl substituents.²⁵⁸

Other Cp-based metalates have been recently described.^260,261 As mentioned, the reduction of 3d metallocenes to their monoanions as transient species, from Cp₂V to Cp₂Ni, has been electrochemically achieved.^250−257 At extremely negative potentials and very low temperatures, some authors discussed the electrochemical reduction of ferrocene to its monoanion,^257,262 and that of cobaltocene and nickelocene to their corresponding dianions.^252,263 However, with the exception of the bis(indenyl)cobaltate 281,²⁵⁸ the products of metallocene reduction were not structurally characterized due to the highly reactive nature of the different anionic species. As recently as 2019, Malischewski reported the structural characterization of potassium salts of the decamethylmanganocene anion [Cp*₂Mn]⁻ (282; see Scheme 36a).²⁶⁰ The synthesis and electronic structures of this anion were described more than four decades ago.^253,264 Following a different synthetic protocol, [Cp*₂Mn] (283) was reduced by molten potassium (Scheme 36a) to afford the extremely air-sensitive anion 282. In the absence of a sequestering agent, the product consists of polymeric chains with strong interactions between the potassium cation and the Cp* ligand of another molecule, as confirmed in the solid-state molecular structure. The presence of a sequestering agent ([18]crown-6; see Scheme 36a) disrupts such intermolecular interactions, as shown by the X-ray characterization of [K([18]crown-6)(thf)₂][Cp*₂Mn], which displays an ion-separated structure. For the ion-separated complex (18 VE), the M–C bond lengths compare well with those of the isoelectronic permethylated ferrocene, [Cp*₂Fe] (284) [d_MnC(282) = 2.048(8)–2.088(8) Å, vs d_FeC(284) = 2.045(3)–2.053(2) Å].²⁶⁵

Scheme 36. Synthesis of Manganese-, Iron-, and Cobalt-Containing Metallocenates^260,261.

Later, Chilton, Mills, and co-workers reported the preparation of an isostructural series of 3d metallocenates containing bulky Cp‴ ligands (Cp‴ = 1,2,4-tBu₃C₅H₂).^261,262 Reduction of [Cp‴₂M] [M = Mn (285), Fe (286), Co (287)] by KC₈ at low temperatures (−78 to −40 °C) yielded the anionic complexes [K([2.2.2]cryptand)][Cp‴₂M] [M = Mn (288), Fe (289), Co (290); Scheme 36b]. The compounds are highly sensitive (thermally and air-sensitive), decomposing rapidly above −30 °C. Despite their facile decomposition, a combination of physical and theoretical methods was used for the characterization of the reduced complexes. In the solid state, the Cp_centroid–M–Cp_centroid angle for the manganese and cobalt compounds deviate from linearity by approximately 5° [angle(288) = 174.68(9)°, angle(290) = 175.96(5)°], while a much larger deviation was determined for the ferrocenate 289 [angle(289) = 169.38(11)°]. The derivatized cyclopentadienyl ligands play an important role in the kinetic and electronic stabilization of the metallocenates. Orbital analysis and Mössbauer spectroscopy indicate that the 19 VE Fe(I) ferrocenate exists in a high-spin S = 3/2 ground state, contrasting the low-spin S = 1/2 ground state observed in the formally isoelectronic cobaltocene compounds.

Recently, Hansen, Prokopchuk, and co-workers described the formation of a rare Mn(0) metalloradical anion and a mixed-valent Mn^–I–Mn^I complex, obtained upon reduction of the aminoindenyl compound [Mn(CO)₃(Ind^Pyrr)] (291, Ind^Pyrr = 1-(1H-inden-2-yl)pyrollidine).²⁶⁶ The analogous unsubstituted complex [Mn(CO)₃(indenyl)] (292) undergoes two-electron reduction to form [Mn(CO)₃(indenyl)]^2– (293). 292 and 293 react via comproportionation to afford [Mn(CO)₃(indenyl)]⁻ (294).²⁶⁷ Based on this precedent, the electrochemical behavior of 291 was investigated, and two clean cathodic redox processes were identified via cyclic voltammetry (CV).²⁶⁶ Therefore, the chemical reduction of 291 with KC₈ was carried out in the presence of [2.2.2]cryptand, yielding the Mn(0) complex [K([2.2.2]cryptand)][Mn(CO)₃(Ind^Pyrr)] (295) and the Mn(−I) carbonylmetalate [Mn(CO)₅]⁻ (37, see Scheme 37, top). The metalloradical character of 295 was confirmed by X-ray crystallography and EPR spectroscopic characterization, as well as by DFT calculations which indicate that the residual unpaired spin density is located mainly at the Mn center.

Scheme 37. Reduction of the Aminoindenyl Complex [Mn(CO)₃(Ind^Pyrr)] (291) in the Presence of Different Sequestering Agents²⁶⁶.

Contrasting with the behavior observed in the synthesis of 295, the use of [18]-crown-6 as an encapsulating agent not only led to an entirely different product but also demonstrated an influence of the order of addition of the reagents.²⁶⁶ The reaction of 291 with KC₈ (2.1 equiv), followed by addition of [18]crown-6, yielded the Mn(−I)/Mn(I) complex [K([18]crown-6)(thf)₂][(OC)₃Mn(μ–η⁴:η⁵-Ind^Pyrr)Mn(CO)₃] (296, Scheme 37, bottom) and the indenide anion salt [K([18]crown-6)][Ind^Pyrr]. In 296, both metal centers are bound to a single indenyl motif, with the Mn(−I) coordinated to the six-membered ring in η⁴ fashion. Given the disparate behavior observed with the two encapsulating agents, the influence of the order of addition of the reagents was investigated. When the reduction of 291 was carried out adding first the crown ether and then the reductant, only intractable mixtures of products were obtained and the mixed-valent compound 296 could not be identified. On the contrary, changing the order of addition of reductant and [2.2.2]cryptand did not alter the outcome of the synthesis of 295. This study underlines the importance that pairing interactions have on the outcome of the reaction, and the potential differences that arise between the electrochemical and preparative studies.²⁶⁶

In 2019, Saito and co-workers reported the isolation and structural characterization of a related anionic stannaferrocene complex, [Li(thf)₄][Fe{3,4-Me₂-2,5-(SiMe₃)₂SnC₄}₂] (297).²⁶⁸ The stannaferrocene forms when Fe(acac)₃ (298) reacts with the dilithiostannole ligand precursor 299 (Scheme 38). The compound is obtained as a stable solid under ambient conditions, albeit in low yield (15%). The planarity of the two stannole ligands was confirmed by their metric parameters in the solid state (sums of internal bond angles = 539°, on average). A spectroscopic and theoretical examination of the electronic structure of 297 gave insights on the configuration of the iron center, suggesting three resonance structures: [Fe^III(L^2–)₂]⁻, [Fe^II(L^2–)(L^–)]⁻, and [Fe^I(L^–)₂]⁻ (L = 3,4-Me₂-2,5-(SiMe₃)₂SnC₄). The ¹¹⁹Sn Mössbauer spectrum presented characteristic features (isomer shift, IS and quadrupole splitting, QS) associated with the presence of Sn(II) atoms in the stannole ligands. Furthermore, contrasting the ⁵⁷Fe Mössbauer spectrum of 297 with that of a known Fe(III) species such as ferrocenium tetrafluoroborate, suggested that the contribution of the Fe(III)-based resonance structure is negligible. A comparison with the spectrum of ferrocene and of known Fe(I) species revealed that the hyperfine parameters in these cases are in the range of those obtained for 297 and, therefore, do not give conclusive evidence on the electronic nature of the metal center (Fe^II vs Fe^I) in this compound. However, the data obtained from the Mössbauer spectra, along with scalar relativistic quantum-chemical calculations (including an analysis of Mulliken charges and spin populations) indicated that the stannole ligands should be monoanionic, while the Fe atom exists in a formal +I oxidation state with a 4s¹3d⁶ electron configuration. This oxidation state was proposed to be the result of reduction of the iron center by the dilithiostannole proligand. Cyclic voltametric studies on 297 revealed, additionally, that this anionic stannaferrocene is significantly prone to oxidation, as expected for this type of reduced compound (quasi-reversible oxidation wave at E_1/2 = −1.50 V, and irreversible oxidation wave at E_pc = −0.64 V; vs Fc/Fc⁺).²⁶⁸

Scheme 38. Synthesis of an Anionic Stannaferrocene²⁶⁸.

2.2.6. Metalates Containing Phosphorus Ligands

Anionic complexes bearing phosphorus ligands, including low-valent species, have long been known. Early examples of these include, among others, the previously discussed cobaltates [(N₂)Co(PEt₂Ph)₃]⁻ (10) [Co(PMe₃)₄]⁻ (255), or [Co(dppe)₂]⁻ (253, dppe = 1,2-bis(diphenylphosphino)ethane).^{66,67,242,243} Additionally, the Le Floch group contributed significantly to this area, by using phosphinine ligands to stabilize reduced metal centers, forming species such as homoleptic complexes of the group 4, 8, and 9 metals [M(tmbp)₃]^2– [M = Ti (300), Zr (301), Hf (302), Fe (303), Ru (304); tmbp = 4,4′,5,5′-tetramethyl-2,2′-biphosphinine] and [M(tmbp)₂]⁻ [M = Co (305), Rh (306)], and the heteroleptic compound [Mn(CO)₃(tmbp)]⁻ (307).^269−272 However, in these cases, the authors concluded that the negative oxidation states proposed for the metals are only formal, with the complexes featuring reduced forms of the biphosphinine ligand. Further examples of metalates with phosphorus ligands, reported in the period 2007–2023, are described in the following sections.

2.2.6.1. Iron

The metalate [K([18]crown-6){Cp*Fe(η⁴-C₁₀H₈)}] (201) served as the starting material for the synthesis of Cp*Fe-phosphinine complexes.^273−275 Müller, Wolf, and co-workers reported on the synthesis of the first anionic iron complex with a π-coordinated phosphinine, [Cp*Fe(η⁴-TPP)]⁻ (308, as the [K([18]crown-6)(thf)₂]⁺ salt; TPP = 2,4,6-triphenylphosphinine), obtained by exchange of the labile naphthalene ligand in 201 with the free phosphinine (Scheme 39).

Scheme 39. Synthesis and Reactivity of Cp*Fe-2,4,6-Triphenylphosphinine Complexes by Müller, Wolf, and Co-workers^273,276.

As in the polyarene complexes [Cp*Fe(η⁴-polyarene)]⁻ (201, 216),²¹⁷ DFT calculations support a d⁶ electron configuration for the iron center, in a formally zerovalent oxidation state, and corroborate the crystallographically determined structure for 308. Complex 308 undergoes hydrolysis to yield [K([18]crown-6)]{Cp*Fe(η⁴-2,4,6-triphenyl-2,3-dihydrophosphinine-1-oxide)}] (309, Scheme 39). Oxygen atom transfer to the phosphorus atom with simultaneous hydrogenation of one carbon–carbon bond of the phosphinine ligand afforded the unusual phosphinine-oxide motif. Such behavior contrasts with the previously observed formation of a 1-hydrophosphinine-1-oxide ligand in the similar compound [CpFe(η⁵-2,4,6-triphenyl-1-hydrophosphinine-1-oxide)] (310).²⁷⁷

Furthermore, reaction between complex 308 and [Cp₂Fe]PF₆ afforded the neutral dimer [Cp*Fe(η⁵-2,4,6-triphenylphosphinine)]₂ (311), which was then further oxidized using I₂ (1 equiv), to yield the cationic iron(II) compound [Cp*Fe(η⁶-2,4,6-triphenylphosphinine)]I (312, Scheme 39). Complex 312 can also be obtained by direct oxidation of 308 with I₂. Moreover, P-centered reactivity was identified for complex 312 through reaction with nucleophiles [LiBHEt₃, LiNMe₂, LiCp*, or Ga(^Dippnacnac), see Scheme 39, bottom left], affording P-substituted η⁵-phosphacyclohexadienyl complexes 313–316 [R = −H (313), −NMe₂ (314), −Cp* (315), and −GaI(^Dippnacnac) (316)]. The structural characterization of complexes 308, 309, and 311–316 highlighted the flexibility of 2,4,6-triphenylphosphinine as a ligand, varying hapticity in accordance with the electronic requirements of the metal: the ligand acts as a 4e^– donor in the anionic complexes 308 and 309, as a 5e^– donor in the neutral dimer 311 or the P-substituted compounds 313–316, or as a 6e^– donor in 312.²⁷³

A complementary protocol for the synthesis of other phosphacyclohexadienyl complexes, analogous to 313–316, is based on the direct functionalization of the anionic complex 308.²⁷⁶ This route gives access to 1-phosphacyclohexadienyls (P-substituted products) and 2-substituted phosphacyclohexadienyls (substitution on an adjacent carbon atom), as shown in Scheme 39, right.

Kinetic and thermodynamic substitution isomers were identified following protonation of compound 308 with HCl·OEt₂. At −80 °C, the kinetic product, 313, was observed after 1 h. At −30 °C, NMR spectroscopic evidence suggests the formation of a mixture of endo- and exo- conformers of the 2-hydrophosphacyclohexadienyl complex 317, as the thermodynamic products (Scheme 39). In fact, at room temperature and in the presence of HCl·OEt₂ (10 mol %), 313 slowly converts to the exo-conformer of 317, in an apparent acid-catalyzed rearrangement. A 65:35 mixture of endo-317 and exo-317 was cleanly obtained by direct reaction of 308 with isopropyl chloride (iPrCl) in THF at room temperature, without observation of 313.

In a similar vein, the 2-substituted phosphacyclohexadienyl complex 318 is isolated from the reaction of the anionic phosphinine compound 308 with Cl-catecholborane. Monitoring the reaction at −100 °C provided evidence of formation of the P-substituted intermediate [Cp*Fe(1-Bcat-PC₅Ph₃H₂)] (318-P), which upon warming transforms into the isolable compound 318. By contrast, direct reactions between 308 and the electrophiles MeI, Me₃SiCl, Ph₂PCl, or Ph₃SnCl afforded, respectively, the P-substituted analogues 319–322 [R = Me (319), SiMe₃ (320), PPh₂ (321), SnPh₃ (322)]. Thereby, it was proposed that an initial attack of the electrophile at the phosphorus atom occurs, yielding the P-substituted phosphacyclohexadienyl ligands, which, in some cases, rearrange to afford C-substituted compounds. This rearrangement was evident for the complexes endo-317, exo-317, and 318. The synthesis of the P-substituted compounds 319–322 by direct reaction between the anionic complex 308 and electrophilic reagents (Scheme 39, bottom right) is thus a more convenient protocol than that shown in Scheme 39, bottom left, which involves a minimum of two synthetic steps.²⁷⁶

Similarly, 201 reacts with the analogous chelating phosphinine 2-(2′-pyridyl)-4,6-diphenylphosphinine (L) to afford the product of naphthalene ligand displacement, [K([18]crown-6)][Cp*Fe(P,N-L)] (323, Scheme 40).²⁷⁴ However, unlike the case of the TPP derivative, 323 features a chelating σ-bonded P,N ligand, and the iron center adopts a trigonal planar geometry. The solid state ³¹P CP MAS NMR spectrum featured a single resonance at δ_iso = 121.1 ppm, supporting the coordination mode identified in the solid-state structure. Nonetheless, the ³¹P{¹H} NMR spectrum in solution showed two distinct resonances, a singlet at δ = 130.7 ppm and a broad resonance at −46.2 ppm, which were assigned to the σ-coordinated 2-(2′-pyridyl)-4,6-diphenylphosphinine and to the η⁴-bonded phosphinine compound, 323-π (Scheme 40), analogous to 308.²⁷³ DFT calculations support the assignment of the species observed in solution and indicate that the conversion between the isomeric forms 323-π and 323-σ proceeds through a barrier of 27.0 kcalmol^–1, consistent with an equilibrium at room temperature, illustrating coordinative flexibility of the ligand.

Scheme 40. Synthesis of Anionic Iron Complexes Featuring Pyridyl-phosphinine Ligands.

All anions are stabilized by [K([18]crown-6)]⁺ cations.

Addition of two equivalents of the pyridyl-phosphinine ligand L to compound 201 yielded the P–P coupled compound [K([18]crown-6)][Cp*Fe(η⁵-L₂)] (324, Scheme 40), whose structure was confirmed by X-ray diffraction analysis. The molecular structure of 324 features a dimerized ligand L₂; one-half of this ligand coordinates to the iron center in an η⁵-fashion. Addition of a second equivalent of 201 to 324 affords compound 323. DFT calculations suggest that the transformation of 323-π into 324 through an additional equivalent of ligand L is favorable by 23.5 kcalmol^–1, proceeding over a low energy barrier of 7.5 kcalmol^–1. It should be noted that the pyridyl-phosphinine ligand participates in reactions of 323-π and 323-σ with CO₂ (see section 3.3.2).²⁷⁴

2.2.6.2. Cobalt

In 2021, the Hunter group reported that the bidentate ligand cis-1,2-bis(diphenylphosphino)ethylene (dppv) provides access to a series of cobalt complexes in five sequential oxidation states.²⁷⁸ The series [Co(dppv)₂(CH₃CN)_x]_n (x = 0, 1, or 2; n = 1–, 0, 1+, 2+, 3+) constitutes a rare example of a redox series of complexes spanning the electron configurations d⁶–d¹⁰ while featuring redox-innocent ligands. The Co(−I) complex was prepared by reduction of the Co(II) compound [Co(dppv)₂]²⁺ (325) with potassium naphthalenide (3 equiv), followed by treatment with [18]crown-6, thereby providing [K([18]crown-6)][Co(dppv)₂] (326). Complex 326 exhibits a pseudo-tetrahedral geometry and is essentially isostructural with the previously reported [Co(dppe)₂]⁻ (253, Scheme 31, vide supra).⁶⁷ Moreover, a comparison of the metric parameters of the five complexes and the free ligand confirmed that the coordinated dppv motif remained almost unaltered through the range of different oxidation states (Co–P bond length, on average = 2.352 Å for Co³⁺ to 2.1165 Å for Co^–, C=C bond length = 1.319–1.334 Å, on average vs 1.334 Å in free dppv).²⁷⁹ Therefore, the ligand does not undergo redox events in this series of complexes and behaves as redox-innocent. The structure of 326 and the rest of the neutral and cationic complexes in the series correspond to the idealized geometries, in accordance with the predictions of the crystal field theory for low-spin d⁶–d¹⁰ metals.

Wolf and co-workers also investigated the classic bis(1,5-cyclooctadiene)cobaltate anion, [Co(η⁴-cod)₂]⁻ (239), as a platform for the synthesis of cobaltates with different ligands.^132,230,280 For instance, 239 reacts with two equivalents of the 1,2-diphosphetane ligand 327(281) to afford a homoleptic cobalt(III)-carborane-bridged bis(phosphanido) complex via successive oxidative addition of the P–P bonds to the Co(−I) center.²⁸⁰ Regardless of the ratio of ligand to metal used (2:1 or 1:1), the product obtained is [K(thf)₄][Co{1,2-(PtBu₂)₂C₂B₁₀H₁₀}₂] (328, Scheme 41, left). The behavior of electron-rich Co(−I) thereby resembles that of, for example, elemental lithium by promoting the reductive cleavage of the P–P bond in 327 and similar compounds.^282,283 According to the structural characterization and theoretical calculations (DFT), in compound 328 the Co(III) center (3d⁶) resides in a distorted tetrahedral geometry, with Co–P single bonds. Measurements of the magnetic susceptibility revealed that complex 328 is diamagnetic in the range of 2 to 270 K in the solid state (SQUID measurements) and at 298 K in THF-d₈ solution. Theoretical calculations suggest that the diamagnetism of 328 is explained by a low-spin singlet ground state resulting from the strong σ-donor and moderate π-donor properties of the bis(phosphanido) ligand. Treatment of 239 with phenylene-1,2-diphosphetane or tetraphenyldiphosphane afforded only intractable mixtures containing paramagnetic species.

Scheme 41. Reactivity of a Low-Valent Cobaltate toward Orthocarborane-Substituted 1,2-Diphosphetanes^132,280.

These results indicate that the carborane backbone is required to form the diamagnetic Co(III) product, providing electronic stabilization due to its electron-withdrawing properties and the attractive London dispersion forces observed in the solid state between the tBu substituents on the P atoms. Furthermore, 328 did not react with tetraphenyldiphosphane.²⁸⁰ Given the limited reaction profile observed for 328, the authors anticipated that an increase in the bulk of the ligand used would allow for the synthesis of heteroleptic complexes that could maintain labile ligands or vacant coordination sites. Thus, the mesityl-substituted diphosphetane rac-329 was treated with cobaltate 239 in the presence of [18]crown-6 (Scheme 41, right) to afford the heteroleptic compound [K([18]crown-6)(thf)][Co{1,2-(PMes)₂C₂B₁₀H₁₀}(η⁴-cod)] (rac-330).¹³² Similarly, reaction of the diphosphetane rac-329 with the Ni(0) complex [Ni(IMes)(η²-H₂C=CHSiMe₃)₂] (331, IMes = 1,3-bis(2,4,6-trimethylphenyl)imidazolin-2-ylidene) affords the neutral nickel(II) complex [Ni{1,2-(PMes)₂C₂B₁₀H₁₀}(IMes)] (rac-332).²⁸⁰

Complex rac-330 provided a suitable platform to synthesize a variety of bis(phosphanido) complexes by exploitation of the labile η⁴-cod ligand.¹³² Treating rac-330 with cyclohexyl isocyanide cleanly substitutes the cyclooctadiene ligand, yielding the bis(isocyanide) complex [K([18]crown-6)(dme)][Co{1,2-(PMes)₂C₂B₁₀H₁₀}(Cy-N≡C)₂] (rac-333; see Scheme 42, top). The characterization of this bis(isocyanide) complex by multinuclear NMR and FT-IR spectroscopy, elemental analysis, and X-ray diffraction confirmed the substitution of the labile ligand, and indicated that the bis(phosphanido) ligand acts as a strong σ-donor, making the Co(I) center more electron-rich [ν_CN(333) = 2029 and 1938 cm^–1 (see Table 1) vs ν_CN (free ligand) = 2136 cm^–1]. Furthermore, as previously observed with other low-valent cobaltate complexes (Scheme 22, vide supra),^208−210rac-330 mediated the dimerization of tert-butylphosphaalkyne to afford the diphosphacyclobutadiene complex [K([18]crown-6)][Co(η⁴-1,3-P₂C₂tBu₂){1,2-(PMes)₂C₂B₁₀H₁₀}] (rac-334; Scheme 42, bottom). The structure was confirmed through a combination of multinuclear NMR spectroscopy, elemental analysis, and X-ray diffraction. Both rac-333 and rac-334 are 16 VE species. The reactivity of rac-330 toward white phosphorus is discussed in section 3.4.3 (vide infra).

Scheme 42. Reactivity of [K([18]crown-6)(thf)][Co{1,2-(PMes)₂C₂B₁₀H₁₀}(η⁴-cod)] (rac-330) toward Cyclohexyl Isocyanide and tert-Butylphosphaalkyne.

Only the complex anions are shown, cations are omitted for clarity.

Rovis and co-workers described a Co(−I) complex featuring a perfluorinated Taddol-based phosphoramidite ligand.²⁸⁴ An interesting 1:1 complex [nBu₄NCo(CO)₃·CKphos] (335, see Figure 8) was isolated by reaction of the ligand CKphos with [nBu₄N·Co(CO)₃] (336).

Figure 8.

Representation of the structure of compound [nBu₄NCo(CO)₃·CKphos].²⁸⁴

The crystallographic analysis revealed that CKphos interacts with the metal center through two perfluorinated aryl rings and the phosphorus atom. The distorted tetrahedral geometry observed for the cobalt atom was attributed to the presence of rather strong Co–C₆F₅ interactions.^285−287 The Co–C₆F₅ bond distances and the distorted tetrahedral geometry at cobalt suggest that the two perfluoroaryls of CKphos act as Lewis acidic, Z-type ligands.

2.2.6.3. Anionic Palladium and Platinum-Phosphine Complexes

Anionic Pd⁰ and Pd²⁺ species are crucial intermediates in Heck reactions and cross-couplings. Knowledge in this field was summarized by Amatore and Jutand in the year 2000.²⁸⁸ More recently, Koszinowski investigated the phosphine-palladate complexes [(PR₃)₂PdX]⁻ [X = Br (337), nBu (338), Ph (339); PR₃ = tris[3,5-bis(trifluoromethyl)phenyl]phosphine] by means of negative-ion mode electrospray-ionization mass spectrometry, electrical conductivity measurements, and NMR spectroscopy.^289,290 The species 337–339 were generated by treating the zerovalent compound [(PR₃)₂Pd] (340) with LiBr or a Grignard reagent, RMgCl (R = nBu or Ph). This transmetalation-like reaction is supported by the electron-poor ligand, which stabilizes the formal anionic state, since electron-richer palladium precursors [L₂Pd] (L = phosphine) exhibit less tendency to form palladates. Species 337–339 react with selected aryl halides ArX′ (e.g., ethyl 4-iodobenzoate) to afford Pd²⁺ complexes of the type [(PR₃)₂Pd(Ar)XX′]⁻ [X = Br (341), nBu (342), Ph (343)]. For X = nBu or Ph, gas-phase fragmentation of 342 and 343 results in the reductive elimination of the ArX cross-coupling products. Mechanistically, these results suggest that the transmetalation step takes place before the oxidative addition, contrary to the classical concept for reactions proceeding exclusively via neutral and cationic Pd-intermediates. Other authors have also analyzed organometallic complexes and their reactions using ESI mass spectrometry, including “ate” compounds or intermediates.^291−293

In 2008, Milstein described the preparation and characterization of the monometallic anionic Pt⁰ complex Na[Pt(PCP)] [344, PCP = 2,6-(CH₂PtBu₂)₂-1-yl-C₆H₃] as well as computational and reactivity studies.²⁹⁴ The 16-electron zerovalent platinum species 344 was obtained by reducing any of the Pt²⁺ precursors [PtCl(PCP)] (345) or [Pt(PCP)][BF₄] (346) with sodium (Scheme 43). Although the solid-state molecular structure of 344 could not be determined, likely because the compound is only moderately stable in solution, it was thoroughly characterized by multinuclear NMR spectroscopy (¹H, ¹³C, ³¹P, ¹⁹⁵Pt). The NMR data, together with DFT calculations, suggest a monomeric planar T-shaped geometry in solution, instead of a dimeric [PCP–Pt–Pt-PCP]^2– arrangement. Theoretical analysis of 344 indicates that the negative charge is localized at the metal center instead of being delocalized over the Pt–C σ-bond and the π-conjugated system of the aryl scaffold. This electron-rich anionic Pt⁰ complex is a Brønsted base. For example, protonation by water affords a Pt(II)-hydride complex, [PtH(PCP)] (347). In addition, 344 is an effective electron-transfer reagent, even being capable of activating one C–F bond from hexafluorobenzene at −35 °C, with reoxidation to Pt(II) in 348.

Scheme 43. Synthesis of a Pincer-type Anionic Pt(0) Complex and Reactivity toward H₂O or C₆F₆²⁹⁴.

2.2.6.4. Molybdenum and Tungsten Complexes with Phosphinidene and Phosphinidene Oxide Ligands

Ruiz and co-workers have extensively investigated the chemistry of the anionic phosphinidene and phosphinidene oxide complexes illustrated in Figure 9.^295−308 A more extensive review of the chemistry of compounds 349 and 350 (and related compounds) was published by Ruiz in 2013, while the rich chemistry of the P₄ activation products [M₂Cp₂(μ-PCy₂)(CO)₂(μ–κ²:κ²-P₂)]⁻ [M = Mo (351), W (352)] and their corresponding methylation products [M₂Cp₂(μ-PCy₂)(CO)₂(μ–κ²:κ²-P₂Me)] [M = Mo (353), W (354)] was reviewed by Caporali, Wolf and co-workers in 2021.^149,176

Figure 9.

Mononuclear (top), homodinuclear, and heterodinuclear (bottom) anionic complexes featuring phosphinidene and phosphinidene oxide ligands.^295−308

In 2004, Ruiz had described the anionic phosphinidene oxide complex (H-DBU)[MoCp{P(O)R*}(CO)₂] (349, DBU = 1,8-diazabicyclo [5.4.0] undec-7-ene; R* = 2,4,6-tBu₃-C₆H₂; Figure 9, top), which was obtained by deprotonation and controlled oxidation of the dimer [Mo₂Cp₂(μ-H)(μ-PHR*)(CO)₄] (355) with DBU and O₂.²⁹⁵ In 349, the carbene-like phosphinidene oxide ligand R–P=O is terminally bound to the Mo center, with the complex exhibiting interesting acid/base properties owing to the presence of three different nucleophilic sites: the O, P, and Mo atoms. The reactivity of compound 349 toward various electrophiles and oxidizing agents was thoroughly studied prior to the period covered by this contribution.^295,296 In 2010, the tungsten analogue (H-DBU)[WCp{P(O)R*}(CO)₂] (350) was reported, along with a complete theoretical description of the electronic structure of anion 349, as a tool to rationalize its experimental chemical behavior toward different electrophiles.²⁹⁷ An analysis of the relevant molecular orbitals and the topology of the electron density showed that the Mo–P bond has a strong π component and can thus be described as a double bond. The oxygen atom of the phosphinidene ligand bears the highest negative charge in the molybdate anion. This electronic situation is consistent with the observed transformations against neutral electrophiles (e.g., alkyl halides), which attack at the phosphorus atom. The authors referred to this phenomenon as orbital control reactivity (Figure 9). Positively charged electrophiles or strongly polarized species such as the Meerwein salt and diethyl sulfate preferably attack at the oxygen atom; this reactivity mode is called charge control. When using other bulky electrophiles such as AuCl{P(p-tol)₃}, addition of the electrophilic fragment to the group 6 metal instead of the phosphorus atom occurs due to steric reasons (Figure 9).²⁹⁸

Besides the incorporation of formal E⁺ fragments to the O, P, and Mo atoms, Ruiz also reported the addition of a chalcogen atom (E) to the Mo–P double bond in the parent complex 349, leading to the anionic dioxo- and thiooxophosphorane complexes [MoCp(CO)₂{κE,κP-EP(O)(2,4,6-tBu₃-C₆H₂)}]⁻ [E = O (356), S (357)]. The thiooxophosphorane complex [MoCp(CO)₂{κS,κP-SP(O)(2,4,6-tBu₃-C₆H₂)}]⁻ (357) exhibits a negative charge at the oxygen atom much higher than that at the sulfur atom, while its HOMO has a large contribution from a sulfur lone pair. As a result, the sulfur atom is attacked by an electrophile under the conditions of orbital control, while the oxygen atom favors the attachment of an incoming electrophile under the conditions of charge control (Figure 9).²⁹⁹

The reactivity of [M₂(η⁵-Cp)₂(μ-PR₂)(μ-CO)₂]⁻ [M = Mo (358), W (359), R = Cy; M = Mo, R = tBu (360); Figure 9) was extensively studied with a broad variety of electrophiles, including, for instance, the nitrosyl complex [Re(η⁵-C₅H₄Me)(CO)₂(NO)]⁺, Brønsted and carboxylic acids, SnClPh₃, [AuCl(PR₃)] (R = iPr, p-tol), PbClPh₃, MeI, BnCl, PClR′₂ (R′ = tBu, Et, Cy), PCl(O)(OPh)₂, and P₄.^{176,300−302} In 2017, the related lighter analogue [Mo₂(η⁵-Cp)₂(μ-PtBu₂)(μ-CO)₂]⁻ (360) was reported, which was used to obtain new unsaturated hydride and alkyl derivatives with very distinctive chemical properties due to the presence of the bulky PtBu₂^– ligand.³⁰³ For example, complex [Mo₂Cp₂(H)(μ-PtBu₂)(CO)₂] (361) bears a terminal hydride ligand both in solution and in the solid state, in marked contrast to the conventional bridging coordination seen in all related hydrides [Mo₂Cp₂(μ-H)(μ-PR₂)(CO)₂] (362, R = Cy, Et, OEt, Ph).¹⁷⁶ Likewise, the agostic species [Mo₂Cp₂(μ–κ¹:η²-CH₃)(μ-PtBu₂)(μ-CO)] (363) is more stable than that of the related PCy₂^– compound and can be thermally dehydrogenated to give the methylidyne derivative [Mo₂Cp₂(μ-CH)(μ-PtBu₂)(μ-CO)] (364).³⁰³

The heterobimetallic species [MoW(η⁵-Cp)₂(μ-PCy₂)(μ-CO)₂]⁻ (365, Figure 9) was recently incorporated into the family of anionic unsaturated group 6 complexes, with the aim of exploring the effects of the intermetallic multiple bond on the reactivity and properties of these anions.³⁰⁴ Complex 365 shows an intermediate chemical behavior between that observed for the homonuclear analogues. The hydride and alkyl derivatives of 365 show clear heterobimetallic effects, in terms of the hydride formation preference (terminal at W) and faster dehydrogenation of the methyl ligand in comparison to the homonuclear [Mo₂Cp₂(μ–κ¹:η²-CH₃)(μ-PtBu₂)(μ-CO)] (363).

In 2016, Ruiz published the first anionic nitrosyl complex [W₂Cp₂(μ-PPh₂)(NO)₂]⁻ (366, Figure 9) featuring a metal–metal double bond.³⁰⁵ In this compound, the NO ligand is terminal and the bridging PPh₂^– scaffold was found to be crucial to stabilize both the W₂ anion and its derivatives against degradation. The reactivity of 366 toward Brønsted acids, [AuCl{P(p-tol)₃}], and elemental sulfur was explored, revealing that compound 366 displays a considerable metal-based nucleophilicity. This feature makes 366 an attractive precursor for the preparation of several ditungsten nitrosyl derivatives and the subsequent study of nitric oxide activation.^309−312

Furthermore, Ruiz and co-workers subsequently reported the complexes [MoMCp(μ-PR₂)(CO)₅]⁻ [M = Mn, R = Ph (367); M = Re, R = Cy (368); Figure 9], the first examples of organometallic anions with group 6–7 metals having M–M′ double bonds.³⁰⁶ Reactions with selected electrophiles revealed high metal-based nucleophilic reactivity, accompanied by association of solvent molecules, to provide electron-precise species. Reaction of [MoMnCp(μ-PPh₂)(CO)₅]⁻ (367) with NH₄⁺ afforded the hydride complex [MoMnCp(μ-H)(μ-PPh₂)(CO)₅(NH₃)] (369).³⁰⁷ Shortly after, the reactivity of the heavier congener [MoReCp(μ-PCy₂)(CO)₅]⁻ (368) toward simple donor ligands and p-block E–H reagents was reported.³⁰⁸ Complex 368 binds L-type donors at the Re center to afford electron-precise anions of the type [MoReCp(μ-PCy₂)(CO)₅L]⁻ [370, L = CO, HSPh, HCC(p-tol)]. The incoming ligand L generally binds in a position cis to the PCy₂^– bridge. By contrast, HPPh₂ also coordinates at rhenium but occupies the less sterically demanding trans-position to the PCy₂^– bridge. Activation of the E–H bonds only proceeds upon protonation of the corresponding anion [MoReCp(μ-PCy₂)(CO)₅L]⁻ (371), followed by thermal or photochemical dehydrogenation and/or decarbonylation.

2.2.7. Carbene Metalates

As mentioned in section 2.2.3.2, the tetraisocyano complex Fe(N₂)(CNAr^Mes2)₄ (61), reported by Figueroa and co-workers, is a masked, nitrogen-trapped, analogue of the transient Fe(CO)₄ complex.¹²³ In the search for further examples of complexes analogous to the elusive Fe(CO)₄, Smith and co-workers reported the synthesis of a series of low-valent Fe(I) and Fe(0) complexes featuring a bulky bis(carbene)borate ligand.⁷³

The diphenylbis(carbene)borate ligand Ph₂B(tBuIm)₂^– reacts with the Fe(II) precursor [FeCl₂(thf)_1.5] to the compound Ph₂B(tBuIm)₂FeCl(thf) (372) (Scheme 44a). One-electron reduction of 372 in toluene affords Ph₂B(tBuIm)₂Fe(η⁶-C₇H₈) (373, Scheme 44a). EPR and Mössbauer spectroscopic analysis of 373 support its formulation as a low-spin (S = 1/2) Fe(I) complex. Furthermore, the C_carbene–Fe bond lengths (1.981(3) and 1.975(3) Å), as provided by X-ray diffraction, revealed a stronger interaction in 373 compared to 372 (2.085(2), 2.100(2) Å). The C–C distances in the toluene ligand (on average, 1.405 Å) indicate that aromaticity is maintained, and thus, a metal-based reduction has occurred. Upon exposure to a carbon monoxide atmosphere, 373 converts to the square-pyramidal Fe(I)-carbonyl compound Ph₂B(tBuIm)₂Fe(CO)₃ (374). A reversible one-electron reduction was observed for 374 by cyclic voltammetry (−1.92 V vs Fc/Fc⁺). Chemical reduction of 374 with KC₈ (1 equiv) under a CO atmosphere in the presence of [2.2.2]cryptand afforded the diamagnetic compound [K([2.2.2]cryptand)][Ph₂B(tBuIm)₂Fe(CO)₃] (375, Scheme 44a). A comparison of the FT-IR stretching frequencies for the carbonyl ligands in 374 vs 375 [ν_CO(374) = 1987, 1900 cm^–1 vs ν_CO(375) = 1926, 1836, and 1800 cm^–1] suggests a higher degree of π-backbonding in the latter species. Together with the remaining spectroscopic data, including the Mössbauer spectra, this indicates an Fe(0) center in 375. When the reduction by KC₈ is carried out under an N₂ atmosphere instead, a CO ligand is lost and K₂[Ph₂B(tBuIm)₂Fe(CO)₂]₂ (376) is formed (Scheme 44a). The shift of the stretching bands to lower frequencies for the CO ligands [ν_CO(376) = 1838 and 1742 cm^–1] suggests reduction of the metal center. Single-crystal X-ray diffraction analysis of 376 revealed a centrosymmetric dimer featuring distorted square-planar coordination environment of the iron centers. The monomeric [Ph₂B(tBuIm)₂Fe(CO)₂]⁻ anions associate via interactions of the carbonyl ligands with the K⁺ cations.⁷³ Addition of [2.2.2]cryptand to 376 breaks the dimer, affording the four-coordinate complex [K([2.2.2]cryptand)][Ph₂B(tBuIm)₂Fe(CO)₂] (377).³¹³ Crystallographic analysis showed that compound 377 features a trigonal pyramidal iron center. The complex cocrystallized with the six-coordinate iron(II) hydride complex 378. The latter species is the result of oxidative C–H activation of a tert-butyl substituent in the bis(carbene)borate ligand. As expected, FT-IR stretching frequencies [ν_CO(377) = 1846 and 1753 cm^–1] are only slightly higher than in the dimer 376. These data indicate the presence of an Fe(0) center, while the FT-IR stretching frequencies of hydride 378 (ν_CO = 1930 and 1859 cm^–1) are significantly higher, in line with the Fe(II) oxidation state. Variable temperature IR spectroscopy revealed that complexes 377 and 378 are in equilibrium, with the Fe(II) species being predominant at lower temperatures. DFT calculations revealed that the coordinatively unsaturated complex 377 must undergo a triplet to singlet spin-state change before C–H activation occurs on the singlet surface.

Scheme 44. Synthesis of Low-Valent Fe(I) and Fe(0) Complexes Featuring a (a) Bulky Bis(carbene)borate^73,313 and (b) Tris(carbene)borate Ligand⁷⁴.

Similar reactivity was observed with the analogous tris(carbene)borate scaffold PhB(MesIm)₃^– (MesIm = 1-mesitylimidazol-2-ylidene).⁷⁴ Fe(I) and Fe(0)-tris(carbene)borate complexes were obtained by sequential reduction of the cationic Fe(II) complex [PhB(MesIm)₃Fe(CO)₃][B(C₆F₅)₄] (379). An initial reduction affords the neutral iron(I) radical PhB(MesIm)₃Fe(CO)₂ (380, see Scheme 44b), while further reduction affords the anionic complex K[PhB(MesIm)₃Fe(CO)₂] (381). Similar to compound 376, compound 381 crystallized as a potassium-bridged dimer. The tris(carbene)borate ligand acts as a strong donor in both species obtained. The structural and spectroscopic data for compound 380 revealed that it adopts a low-spin (S = 1/2) configuration, as in the case of the Fe(I) complex 373. The expected increase in electron density in 380 with respect to 379 was evidenced by the shorter Fe–CO distances observed in the X-ray data (d_FeCO = 1.771(5) and 1.791(5) Å vs average d_FeCO(379) = 1.821 Å) along with low carbonyl stretching frequencies in the FT-IR spectra [ν_CO(379) = 2100, 2040 cm^–1 vs ν_CO(380) = 1956, 1886 cm^–1; see Table 1]. The π-backbonding interaction is even more pronounced in the anionic complex 381 [average d_FeCO(381) = 1.729(4) Å; ν_CO(381) = 1812, 1728 cm^–1; see Table 1].⁷⁴

Smith and co-workers additionally described a related anionic bis(carbene)borate iron complex, featuring a dinitrogen ligand.³¹⁴ The ferrate compound is accessed by a two-step synthetic protocol which involves the alkylation of complex Ph₂B(tBuIm)₂FeCl(thf) (372)⁷³ with LiCH₂tBu (1 equiv) to obtain the tricoordinate compound Ph₂B(tBuIm)₂Fe(CH₂tBu) (382), and subsequent reduction with KC₈ in the presence of [2.2.2]cryptand under a nitrogen atmosphere to afford [K([2.2.2]cryptand)][Ph₂B(tBuIm)₂Fe(N₂)(CH₂tBu)] (383, Scheme 45). The four-coordinate complex 383 features a low frequency FT-IR band attributed to the coordinated dinitrogen ligand (ν_N≡N = 1897 cm^–1).

Scheme 45. Synthesis of an Anionic Iron Bis(carbene)borate Featuring a Dinitrogen Ligand, and Its Ligand Substitution Reaction³¹⁴.

Treatment of 383 with diphenylacetylene substitutes the N₂ ligand to give [K([2.2.2]cryptand)][Ph₂B(tBuIm)₂Fe(PhC≡CPh)(CH₂tBu)] (384). The strong backdonation from the iron center in 384 is evidenced by the elongation of the C–C bond distance (1.281(7) Å) in the coordinated alkyne ligand. Complex 383 catalyzes the isomerization of 1-hexene to 2-hexene. The authors proposed that the isomerization of the olefin substrate occurs by the allyl mechanism.^315,316 The combined experimental and computational study revealed that species with two different spin states (S = 3/2 and 1/2) are involved in the operating mechanism. This two-state (spin) reactivity provides resistance toward common catalyst poisons to the catalyst.

In 2014, Bertrand, Peters, and co-workers reported the synthesis of two-coordinate, formally Fe(0) and Co(0) complexes [M(CAAC)₂] [M = Fe (385), Co (386); CAAC = cyclic (alkyl)(amino)carbene].⁷⁵ CAAC ligands are strong σ-donors and efficient π-acceptors, known to be redox-active.^317,318 Reaction of the corresponding M(II) salt precursor with the CAAC ligand 387 (2 equiv) followed by reduction with sodium amalgam (1 equiv), afforded the three-coordinate complexes [MCl(CAAC)₂] [M = Fe (388), Co (389)]. Complexes 388/389 reacted with NaBAr^F₄ (BAr^F₄ = tetrakis[3,5-bis(trifluoromethyl)phenyl]borate) as a chloride scavenger to obtain cationic species [M(CAAC)₂]BAr^F₄ [M = Fe (390), Co (391)], which were further reduced to yield the neutral, two-coordinate compounds 385 and 386 (Scheme 46 shows compound 385). The cationic (390/391) and neutral (385/386) complexes have a moderately bent structure in all the cases, with angles ranging from 165.4 to 170.5°.⁷⁵

Scheme 46. Reduction of a Low-Valent and Low-Coordinate Fe(0)-CAAC Complex to Form an Fe(−I)-Dinitrogen Complex^75,76.

Later, Peters reported that the two-coordinate Fe(0) compound [Fe(CAAC)₂] (385) coordinates the dinitrogen molecule at low temperatures (T < – 80 °C).⁷⁶ Evidence of coordination was obtained via variable temperature UV–vis spectroscopy since the absorption spectrum of 385 presents drastic, but reversible, changes below −80 °C. This behavior might be indicative of the formation of a three-coordinate dinitrogen complex, proposed to be [(N₂)Fe(CAAC)₂] (Scheme 46), though such a species was not isolated. Nonetheless, further reduction of 385 with KC₈ in the presence of [18]crown-6 afforded the formally Fe(−I) complex [K([18]crown-6){(N₂)Fe(CAAC)₂}] (392, Scheme 46). X-ray crystallographic data for this compound revealed that a further elongation of the C–N bonds in the coordinated carbene ligands has occurred upon formation of the Fe(−I) compound (on average: d_CN(385) = 1.37 Å vs d_CN(392) = 1.40 Å), reflecting the increased spin density at the metal center.^75,76 The stretching band (FT-IR, ν_N≡N = 1850 cm^–1; see Table 2) of the coordinated dinitrogen ligand in the metalate 392 suggests that this motif presents some degree of activation. Magnetic susceptibility and EPR measurements (1.9 μ_B in C₆D₆, S = 1/2 in the ground state) are consistent with a highly reduced, formally Fe(−I) center. The presence of an activated N₂ ligand served as a starting point to pursue functionalization studies (see section 3.2.3).⁷⁶

In 2021, Krämer, Young, and co-workers reported the stabilization of iron centers in the range of formal oxidation states Fe(II), Fe(I), and Fe(−I) using a PC_carbeneP pincer ligand scaffold.⁷⁷ In the search for a readily accessible iron pincer complex, this group synthesized an Fe(0) PC_carbeneP pincer compound via dehydration of an isolated α-hydroxylalkyl hydrido complex and evaluated its reactivity toward a breadth of reagents and its redox behavior. The pincer complex [Fe(PC_carbeneP)(CO)₂] (393, see Scheme 47) reacts with KC₈ in benzene to afford the formally Fe(−I) compound K[Fe(PC_carbeneP)(CO)₂] (394).

Scheme 47. Synthesis of a Formal Fe(−I) PC_carbeneP Pincer Complex and Its Reactivity toward Hydrogen Atom Transfer (HAT)⁷⁷.

FT-IR analysis of 394 indicated the presence of a significantly reduced iron center, according to the stretching bands of the carbonyl ligands [ν_CO(394) = 1826 and 1767 cm^–1 vs ν_CO(393) = 1961 and 1894 cm^–1], whereas the EPR spectrum suggested the formation of a carbon centered radical. Since 394 is an extremely reactive compound, it was not possible to further characterize it in solution. However, it was found that 394 abstracts hydrogen atoms from aliphatic sources (such as THF, MeCN, 9,10-dihydroanthracene, and NEt₃; see Scheme 47) to form a PC_alkylP ferrate complex, [K([18]crown-6)][Fe(PC_alkylP)(CO)₂] (395). Indirect confirmation of the structure of 394 was therefore achieved by characterization of the hydrogen atom transfer (HAT) product 395 via X-ray diffraction (Scheme 47). Furthermore, DFT calculations support the presence of a markedly reduced iron center with a significant carbene radical character. Calculated Fukui parameters also suggested the presence of an increased nucleophilic character on the carbene C atom, while the iron center was more electrophilic.

G. de Ruiter and co-workers described the synthesis of the anionic pincer monohydride iron complex Na[(PC_NHCP)Fe(H)(N₂)] (396, PC_NHCP = 2,10-(tBu)₂-4,8-(PiPr)₂C₁₁H₄N₂; Scheme 48) and its high catalytic activity in the isomerization of olefins.³¹⁹ The two-step synthesis of 396 comprises the reduction of the previously reported dichlorido complex [(PC_NHCP)Fe(Cl)₂] (397)³²⁰ with KC₈ under a nitrogen atmosphere to afford the neutral complex [(PC_NHCP)Fe(N₂)₂] (398). The latter species then reacts with NaBHEt₃ to afford the monohydride compound 396. ¹H NMR spectroscopic analysis confirmed the formation of the hydride ligand at the iron center (δ = −10.85 ppm (t), ²J_PH = 55.1 Hz). Crystallographic characterization of 396 showed a dimer in which two sodium atoms link the two iron fragments by interactions with the dinitrogen ligands. These interactions, along with the presence of the electron-rich metal center, account for the slightly activated dinitrogen ligands evidenced by the N–N bond distances (1.155(6) Å). Complex 396 is a rare example of a well-defined anionic Fe(0) hydride species not supported by π-acidic carbonyl or isocyanide ligands. The striking catalytic activity observed in the isomerization of alkenes with complex 396 (TONs up to ≥160,000, TOFs up to 6,600 h^–1) contrasts with that of the known dihydride compound [(PC_NHCP)Fe(H)₂(N₂)] (399), described by the same group, which proved inactive despite effectively mediating other transformations.^319,321 Similar to the case of complex [K([2.2.2]cryptand)][Ph₂B(tBuIm)₂Fe(N₂)(CH₂tBu)] (383),³¹⁴ computational studies indicated that a two-state reactivity (from the low-spin singlet to the high-spin triplet) might be responsible for the catalytic performance. Catalyst 396, however, operates via an alkyl-type mechanism.^315,316

Scheme 48. Synthesis of ferrate Complex Na[(PC_NHCP)Fe(H)(N₂)] (396)³¹⁹.

In 2020, independent reports by the groups of Deng, and Tejel and Ciriano described the synthesis of three-coordinate group 9 metalates featuring N-heterocyclic carbene ligands.^322,323 Reduction of [(IMes)Co(dvtms)] (400, IMes = 1,3-dimesitylimidazol-2-ylidene, dvtms = divinyltetramethyldisiloxane) or a mixture of [RhCl(cod)]₂ (401) and IPr (IPr = 1,3-bis(2′,6′-diisopropylphenyl)imidazol-2-ylidene) with KC₈ (in the presence of [18]crown-6, see Scheme 49, a and b, respectively), afforded, instead of the desired compound [K([18]crown-6)][(IMes)Co(dvtms)] or its rhodium analogue, the products of single C–N bond cleavage (dearylation) of an aryl substituent in the NHC ligands. These compounds were characterized as the Co(0) N-aryl imidazolate complex [K([18]crown-6)(μ-C₃H₂N₂Mes-κC²,κN³)Co(dvtms)] (402) and the Rh(I) species [K([18]crown-6)(THF)][(cod)₂Rh₂(μ-C₃H₂N₂(Dipp)-κC²,κN³)₂(μ-Dipp)₂K] (403). The authors proposed that the dearylation of the NHC ligands in 402 and 403 is possibly mediated by cobalt(−I) and rhodium(−I) metal centers, formed as intermediates in the reduction of the precursor complexes 400 or 401/IPr. Such activation by the highly reduced metal center becomes more evident in the case of 403, given that the aryl substituents remain coordinated to the Rh(I) centers.

Scheme 49. Anionic Cobalt and Rhodium Complexes Derived from NHC Ligands by Deng and Co-workers and Tejel, Ciriano, and Co-workers^322,323.

It was proposed that a similar cobalt intermediate formed from the undetected Co(−I) compound [(IMes)Co(dvtms)]⁻, via intramolecular oxidative addition of the C–N bond, followed by homolytic cleavage of the Co–C(aryl) bond to generate the isolable Co(0) complex 402 and a mesityl radical. The formation of the latter was confirmed by detection of mesitylene by gas chromatography (GC). The observed mesitylene is the product of hydrogen atom abstraction from the solvent by the mesityl radical.

For a deeper study of the behavior of the highly reduced Co(−I) and Rh(−I) metal centers, further examples of the desired complexes were synthesized. Analogous treatment of [(ICy)Co(dvtms)] (404, ICy = 1,3-dicyclohexylimidazol-2-ylidene) with KC₈ in the presence of [2.2.2]cryptand yielded the formal Co(−I) complex [K([2.2.2]cryptand)][(ICy)Co(dvtms)] (405, Scheme 49a). Similarly, dimer [RhCl(dvtms)]₂ (406) reacted with IPr and KC₈ to afford either the Rh(0) complex [(IPr)Rh(dvtms)] (407) or the formal Rh(−I) species [K([18]crown-6)][(IPr)Rh(dvtms)] ([K([18]crown-6)][408], see Scheme 49b), depending on the relative amount of reductant used. Neither 405 nor [K([18]crown-6)][408] undergoes C–N bond cleavage, even under heating conditions (50 or 70 °C for the cobalt or rhodium complexes, respectively). The authors proposed that, in the case of the cobalt complex, the formation of 405 responded both to the weaker π-accepting character of the ICy ligand and to the inability of the substituents on the NHC scaffold to form arene–metal interactions, thought to play an important role in the early activation (dearylation/intramolecular oxidative addition) of the C–N bond by the reduced metal center. Deng and co-workers also proposed that the lower electronegativity of cobalt in contrast to rhodium and the different π-accepting nature of the COD ligand versus that of the divinylsiloxane might explain the distinct behavior observed between compounds 405/[K([18]crown-6)][408] and 402/403. Altogether, the collective evidence supports M(−I) mediated C–N bond oxidative addition processes for the dearylation reactions observed.³²²

Tejel and Ciriano had previously reported the formal Rh(−I) compound K[Rh(IPr)(dvtms)] (K[408], Scheme 49c) by reduction of the Rh(I) dimer [Rh(μ-Cl)(IPr)(dvtms)]₂ (409, IPr = 1,3-bis(2,6-diisopropylphenyl)imidazolyl-2-ylidene).³²³ The solid-state molecular structure of K[408] showed that the Rh center has a trigonal-planar geometry, where the metal is bound to the NHC ligand and both C=C bonds in dvtms. It is noteworthy that all the carbon atoms in the Rh coordination sphere are essentially in the molecular plane. This is a consequence of increased π-backdonation, as revealed by the elongation of the C=C bond distances compared to those of the uncoordinated C=C units in the dimer 409 (ca. 1.43 vs 1.30 Å). NMR spectroscopy also supports the increased π-backdonation evidenced in the solid state. Considering this, the Rh(−I) diene complex K[408] can alternatively be described as the corresponding Rh(I) metallacyclopropane resonance structure.

Complex K[408] undergoes NHC ligand exchange by PPh₃ forming K[Rh(PPh₃)(dvtms)] (410), while salt metathesis with [AuCl(PPh₃)] (146) gave [(IPr)(dvtms)RhAu(PPh₃)] (411, Scheme 49c). NMR spectroscopic data, along with NBO analysis, indicate a higher contribution of the Rh^–I-Au^I resonance form than the Rh⁰-Au⁰ form to the ground state of 411, which is therefore a rare example of a TM complex with two closed-shell d¹⁰ metals.

Kennedy reported the family of anionic nickel(0) complexes bearing bidentate NHC–pyridone ligands, and their use in hydroboration catalysis.³²⁴ Reactions of [Ni(cod)₂] (240) with the corresponding ligand in a 1:2 ratio gave the series of complexes [(κ²-C,N-^ArNHCPyO)Ni(cod)]⁻ [Ar = Mes (412), Dep (413), Dipp (414)] as the [K([18]crown-6)]⁺ salts (see Scheme 50). The 1,5-cyclooctadiene ligand in 412–414 is labile, undergoing ligand displacement by MeCN to yield the tricoordinate 16-electron complexes [(κ²-C,N-^ArNHCPyO)Ni(η²-MeCN)]⁻ [Ar = Mes (415), Dep (416), Dipp (417)] as the [K([18]crown-6)]⁺ salts. In the solid state, the [K([18]crown-6)]⁺ counterions in 412, 413, and 415 form a contact ion pair with the O atom of the pyridonyl unit, thereby preventing bridging coordination modes and multimetallic aggregation. The MeCN ligand in 415 has an elongated C–N bond (1.235(2) Å), reflecting the significant backbonding from the electron-rich nickel center. The series of (κ²-C,N)Ni⁰ complexes were evaluated as catalysts for hydroboration reactions (see section 3.1.6).

Scheme 50. Synthesis of Anionic Ni(0) Complexes Featuring Bidentate NHC–Pyridone Ligands³²⁴.

2.2.8. Alkyl, Amido, and Imido Complexes

2.2.8.1. Monodentate Alkyl, Silyl(aryl)amido, and Imido Complexes

The Sc²⁺ complexes [K(L)][Sc{N(SiMe₃)₂}₃] (418, L = [2.2.2]cryptand; 419, L = [18]crown-6) and [Cs([18]crown-6)][Sc{N(SiMe₃)₂}₃] (420) were synthesized by Evans and co-workers, upon reduction of [Sc{N(SiMe₃)₂}₃] (421) with elemental K and Cs (Scheme 51).^325,326 These compounds are the first crystallographically characterized Sc²⁺ complexes. The presence of Sc²⁺ ions was also confirmed by their eight-line EPR spectra arising from the I = 7/2 ⁴⁵Sc nucleus. It is interesting to note that the complexes do not readily react with N₂ even though reduction of 421 under dinitrogen affords the crystallographically characterized complex {K([2.2.2]cryptand)}₂{[{(Me₃Si)₂N}₃Sc]₂[μ–η¹:η¹-N₂]} (422) with rare end-on (N=N)^2– ligands.³²⁵ The reactivity of [K([18]crown-6)][Sc{N(SiMe₃)₂}₃] (419) with CO₂ is discussed in section 3.3.2.

Scheme 51. Synthesis of Mononuclear Scandate(II) Bis(silyl)amide Complexes^325,326.

The first example of a linear two-coordinate iron(I) complex, [K([2.2.2]cryptand)][Fe{C(SiMe₃)₃}₂]₂ (423) was reported by Long and co-workers in 2013.^327,328 The possibility of accessing the reduced complex 423 was identified upon studying the electrochemical behavior (CV) of the Fe(II) precursor, [Fe{C(SiMe₃)₃}₂]₂ (424), for which a reversible reduction event (E_1/2 = −1.82 V, vs Fc/Fc⁺) was observed. Based on these data, the chemical reduction of 424 with KC₈ in the presence of [2.2.2]cryptand was carried out, yielding [K([2.2.2]cryptand)][423]. Crystallographic analysis showed that anion 423 exhibits almost perfect linearity [C–Fe–C angle: 179.2(2)°].³²⁷ Compound 423 exists in a triplet spin ground state (S = 3/2). Due to its high symmetry, it features a large magnetic anisotropy with a high barrier for magnetic relaxation (U_eff = 226(4) cm^–1) even in the absence of an applied magnetic field and a magnetic blocking temperature of 4.5 K. Information on the magnetization dynamics of both 423 and 424 was obtained through Mössbauer spectroscopic studies, which agreed well with the corresponding calculated hyperfine parameters.³²⁸

As mentioned in section 2.2.4.2, silylamido complexes such as Cr[N(SiMe₃)₂]₂(thf)₂ (166) can be used for the synthesis of metalate complexes, e.g., [Cr₂(naphthalene)₂]⁻ (165).¹⁹⁵ However, a series of low-valent and low-coordinate anionic M(I)-(aryl)silylamido or M(I)-bisilylamido species (M = Cr–Ni) have also been described.^329−337 These are generally obtained by reduction of the corresponding two-coordinate M(II) precursors with KC₈ or alkali metals,^338,339 in most cases in the presence of crown ethers or cryptands.^329−336

In 2013, Power described the formation of a tetranuclear chromium-silyl(aryl)amido [Cr{N(SiMe₂CH₂)Dipp}₂Cr]₂(thf) (425) complex, obtained in an attempt to synthesize a two-coordinate Cr-silyl(aryl)amido species. The ligand failed to stabilize the low-coordinate species by undergoing C–H activation at a methyl substituent of the -SiMe₃ group, leading to the tetra-nuclear chromium compound 425 (see Figure 10).³³⁸

Figure 10.

Representation of the structure of the tetranuclear compound [Cr{N(SiMe₂CH₂)Dipp}₂Cr]₂(thf) (425).³³⁸

Later, in 2014, Tilley and co-workers reported on the synthesis of a series of chromium compounds in the formal oxidation states +I, +II, and +III, stabilized by the bulkier silyl(aryl)amido ligand (SiiPr₃)(Dipp)N^– (Dipp = 2,6-diiso-propylphenyl).³²⁹ The mononuclear Cr(II) compound [Cr{N(Dipp)SiiPr₃}₂] (426) was obtained by a salt metathesis of CrCl₂ with 2 equiv of K[N(Dipp)SiiPr₃}₂] in THF. Cyclic voltammetry performed on 426 revealed a reversible Cr(I/II) couple (E_1/2 = −2.04 V, vs Fc/Fc⁺) and, therefore, its chemical reduction with 1.1 equiv of KC₈ was attempted, affording [K(dme)₄][Cr{N(Dipp)SiiPr₃}₂] (K[427], Scheme 52a). K[427] reacts with tetrabutylammonium bromide to afford the Bu₄N⁺ salt [Bu₄N][427]. The solution magnetic moments of both salts (Evans method, 5.2 and 5.7 μ_B, respectively) are consistent with the metal centers in a high-spin d⁵ configuration.

Scheme 52. Synthesis of Low-Valent M(I)-(Aryl)silylamido Complexes of the 3d Metals from Chromium to Nickel^329−333.

Oxidation of 426 by I₂ yielded the cationic Cr(III) compound [(I)Cr{N(Dipp)SiiPr₃}₂] (428), which could be further functionalized.³²⁹ Likewise, the same group reduced the Ni(II) precursor [Ni{N(SiMe₃)Dipp}₂] (429)³⁴⁰ with KC₈ (1.1 equiv) to access the anionic complex K[Ni{N(SiMe₃)Dipp}₂] (K[430]) in toluene at −30 °C (Scheme 52b).³³⁴ As with the chromium analogues,³²⁹ the ligand scaffold proved useful to stabilize three different formal oxidation states at the nickel center, Ni(I/II/III). Fully reversible Ni(I/II) (E_1/2 = −1.28 V, vs Fc/Fc⁺) and Ni(II/III) (E_1/2 = +0.18 V, vs Fc/Fc⁺) redox couples were identified in the cyclic voltammogram of 429. X-ray crystallographic characterization of K[430] revealed an almost perfectly linear N–Ni–N geometry (178.05(9)°) and that the potassium counterion is coordinated by two aryl rings in η⁶-fashion. Subsequent exchange of the potassium cation by tetrabutylammonium (NBu₄⁺) afforded the ion-separated analogue [NBu₄][430]. According to the analysis of the calculated molecular orbitals in the latter, a considerable nucleophilic character was expected for [NBu₄][430] due to the double occupation of its antibonding (dz²-derived) orbital. As a result, reaction of K[430] with methyl iodide at low temperatures (−78 °C to r.t.) yields the product of two-electron oxidative addition, the T-shaped Ni(III) compound [(Me)Ni{N(SiMe₃)Dipp}₂] (431). Attempts to access a Ni(III) species by oxidation of the Ni(II) complex 429 (e.g., with Ag(I) reagents) led to intractable mixtures or no reaction. The formal oxidation states proposed for the anionic Ni(I) complex and for the Ni(III) oxidative addition product are supported by DFT calculations and EPR spectroscopic analysis.³³⁴

Additional examples of (aryl)(silyl)amido complexes [K([18]crown-6)][M{N(Dipp)SiMe₃}₂] [M = Fe (432/433), Co (434), Ni (435)]³³³ or [KM{N(Dipp)SiR₃}₂] [M = Cr, R = iPr (436); M = Mn (437), Fe (438), Co (439, toluene solvate), R = Me]³³² have been reported by different groups. Power, Long, and co-workers obtained 432–435 by reduction of the corresponding M(II) precursors [M = Fe (440), Co (441), Ni (429)]^{338,340−342} in the presence of [18]crown-6.³³³ Compound 432 was obtained as a diethyl ether-cation adduct ([K([18]crown-6)(Et₂O)₂]⁺, 432, Scheme 52c, right), which undergoes loss of Et₂O upon storage at ambient pressure and temperature, under vacuum or after dissolution in noncoordinating solvents (pentane, toluene) to generate the [K([18]crown-6)]⁺ salt 433. The cobalt and nickel compounds 434 and 435 are perfectly linear, while the iron(I) complexes 432 and 433 have a slightly bent structure in the solid state [N–M–N angles: 171.99(6) (432) and 172.67(7)° (433)].³³³

Similar complexes 436–438 were obtained by Werncke and co-workers (Scheme 52d). The characterization by single-crystal X-ray diffraction revealed that these species are associated by contact ion pairing in the solid state, which results in the formation of 1D-coordination polymers with a zigzag arrangement for 437 and 438, while the polymer 436 is more linear.³³² By contrast, the cobalt compound 439 (Scheme 52d) does not form an aggregate in the solid state. Instead, the potassium cation is sandwiched by aryl rings and additionally η²-coordinated by a toluene molecule.³³² The molecular structures are slightly bent [N–M–N angles 177.44(10)° (436), 165.56(6)° (437), 170.59(5)° (438), 178.43(8)° (439)], with a greater deviation from linearity for the manganese and iron compounds, 437 and 438, than for cobalt. NMR studies revealed the presence of an ion-pairing between potassium and the complex anion, which depends on the nature of the solvent: while in weakly/noncoordinating solvents (e.g., toluene-d₈) the ion-pairing seems to persist in solution, ion-separated species appear to be formed in polar coordinating solvents such as THF-d₈. These findings are corroborated by the crystallographic characterization of the monomeric adducts 436·3THF, 438·Et₂O, 438·2DMAP (DMAP = p-dimethylaminopyridine), which are structurally similar to compound 439. In these, polymer formation is prevented by the coordination of additional donor molecules, although the potassium cation remains in the “coordination pocket” formed by two aryl rings of the amido ligands. Furthermore, the addition of a crown ether prior to reduction of the M(II) precursor with lithium or sodium affords ion-separated compounds [AM(chelate)][Fe{N(Dipp)SiMe₃}₂] [AM = alkali metal = Li, chelate = [12]crown-4 (442); AM = Na, chelate = [18]crown-6 (443)] (Scheme 52e). In the absence of crown ether, only decomposition or the formation of oxidized products was observed. These observations suggest that the coordination of lithium or sodium cations strongly influences the stability of these bis(amido)metalates(I).³³²

Anionic silylamido complexes of general formula [K(chelate)][M{N(SiMe₃)₂}₂] [M = Cr (444), Mn (445, dimeric), Fe (446), Co (447), Ni (448); chelate = [18]crown-6 or [2.2.2]cryptand] (see Scheme 53) were described by the team of Bontemps, Sabo-Etienne, and Werncke.^330,331,336 Compounds 444, 446, and 447 were prepared by reduction of M(II)-silylamido-Lewis base adducts [M{N(SiMe₃)₂}₂(L)_n] [M = Cr (449), L = thf, n = 2; Fe (450), L = PCy₃ or thf, n = 1; Co (451), L = thf, n = 1] in the presence of a crown ether or cryptand (Scheme 53a). In the solid state, the molecules reside on crystallographic inversion centers. As a result, perfectly linear structures are found within an eclipsed conformation of the substituents on the silylamido ligands.

Scheme 53. Synthesis of M(I)-Silylamido (M = Cr, Mn₂, Fe, Co, Ni) Complexes^330,331,336.

For the manganese species, an analogous reduction procedure on [Mn{N(SiMe₃)₂}₂] (452) failed to give access to the expected monomeric Mn(I) complex, affording the dianionic dimer [K([18]crown-6)_1.5]₂[Mn{N(SiMe₃)₂}₂]₂ (445), featuring a Mn–Mn bond instead (Scheme 53b). Stabilization of a low-coordinate Mn(I) complex was attempted by using the more encumbering (aryl)silylamido ligand (SiMe₃)(Dipp)N^–. Gratifyingly, the two-coordinate complex [K([18]crown-6)][Mn{N(Dipp)SiMe₃}₂] (453) was accessible after reduction of the Mn(II) precursor [Mn{N(Dipp)(SiMe₃)}₂] (454) with KC₈ in the presence of [18]crown-6 (see Scheme 52c, left, above). Crystallographic analysis revealed that the orientation of the substituents on the nitrogen atoms changes from trans in the neutral complex 454 to cis in the reduced compound 453, with one isopropyl group of each ligand pointing toward the phenyl ring of the opposing ligand. Complex 453 features a bent structure (N–M–N angle 167.12(14)°).³³⁰

The nickel complex [K(chelate)][Ni{N(SiMe₃)₂}₂] (448, Scheme 53c, chelate = [2.2.2]cryptand, or [18]crown-6) proved more difficult to access than its analogues 444–447 with Cr, Mn, Fe, and Co.³³⁶ Initial attempts to reduce the Ni(II) precursor Li(thf)_x[Ni{N(SiMe₃)₂}₃] (455) with KC₈ in toluene afforded only traces of the crystalline dimer [K(toluene)][Ni₂{N(SiMe₃)₂}₃] (456, Scheme 53c, down left). By contrast, reducing 455 with KC₈ in the presence of a sequestering/chelating agent in Et₂O afforded the desired compound 448 (Scheme 53c). However, despite giving access to 448, it was difficult to control the ratio of the reactants due to the tendency of 455 to disproportionate in nonpolar solvents, the sensitivity of the Ni-containing intermediates, and difficulties in determining the THF contents of the precursor, which varies because of decoordination. Therefore, the product was usually obtained with various contaminants, which included the tris(disilyl)amido complex [K([18]crown-6)][Ni{N(SiMe₃)₂}₃] (457). This problem was not solved by using a different sequestering agent ([2.2.2]cryptand). An alternative approach yielded a reliable method for the synthesis of pure 448 in good yields. Displacing the phosphine ligands in the adduct [Ni{N(SiMe₃)₂}(PPh₃)₂] (458) with K{N(SiMe₃)₂} affords 448 as pure [K([18]crown-6)]⁺ and [K([2.2.2]cryptand)]⁺ salts in good yields (Scheme 53c). Analysis of the spectroscopic and magnetic properties indicated that the open-shell complex was best described as an S = 1/2 system. NMR and EPR spectroscopic data of 448 suggested the presence of different conformations in solutions, which may result from distortion of linearity or the suppression of rotation of the silylamido ligands along the main axis. This dynamic behavior of the complexes is likely caused by ion-pairing interactions.

Moreover, the linear or quasilinear N–M–N arrangement of the M(I)-disilylamido complexes results in interesting magnetic properties such as a large magnetic anisotropy and slow relaxation of the magnetization. For example, [K(chelate)][Fe{N(SiMe₃)₂}₂] (446) exhibits large magnetic moments at ambient temperature μ_eff = 5.12 and 5.14 μ_B in solution, which are close to the free ion value (d⁷, 5.19 μ_B).³³¹ These data suggested a large magnetic contribution from unquenched orbital angular momentum. In line with this hypothesis, variable-frequency, variable-temperature magnetic susceptibility measurements revealed effective barriers to magnetic relaxation of U_eff = 43 and 64 cm^–1, and relaxation times of τ₀ = 4.9 × 10^–6 and 8.9 × 10^–6 s for compound 446 when chelated by [18]crown-6 or [2.2.2]cryptand, respectively.

An attempted alternative synthesis of 444–447 was later described,³⁴³ consisting of the reduction of a series of trigonal halido bis(silylamido) M(II) metalates, from manganese to cobalt, by KC₈. However, attempts to reduce the [NBu₄]⁺ salts of the complexes led only to partial formation of the linear metal(I) silylamides, while degradation of the metal complex and decomposition of the [NBu₄]⁺ cations was also observed. By contrast, reduction of [K([18]crown-6)][M^IIBr{N(SiMe₃)₂}₂] (459, M = Fe, Co) afforded the expected [M{N(SiMe₃)₂}₂]⁻ products, albeit in low yield. Similar results were obtained from the trigonal halido (aryl)silylamido metalates(II) [M^IIBr{N(Dipp)SiMe₃}₂]⁻.³⁴³ Therefore, the reduction of the M(II) bis(silylamido) or bis(aryl)silylamido precursors remains the most efficient reaction pathway for the synthesis of the M(I) anionic compounds known to date.

It is worth mentioning that bipyridine adducts of anionic M(I) silylamido complexes of the formula [K([18]crown-6)][M{N(SiMe₃)₂}₂(bipy)] [M = Cr (460), Mn (461), Fe (462), Co (463)] are also accessible. These complexes were prepared by addition of 2,2′-bipyridine to the anionic M(I) silylamides 446 and 447, or additionally with in situ reduction of the neutral homoleptic M(II) precursors 449 and 452 by KC₈ (Scheme 54).³⁴⁴ Complexes 460–463, along with the zinc analogue [K([18]crown-6)][Zn{N(SiMe₃)₂}₂(bipy)] (464), could also be accessed by direct reduction of the neutral heteroleptic M(II) precursors M{N(SiMe₃)₂}₂(bipy) [M = Cr (465), Mn (466), Fe (467), Co (468), Zn (469)]. However, it should be noted that X-ray diffraction analyses of 460–463 and further physical measurements (UV–vis spectra, magnetic susceptibility, and electrochemical measurements) suggest the presence of metal centers in the +II oxidation state and 2,2′-bipyridine radical anions in each case as observed for other anionic bipyridine complexes (vide infra), and these compounds are therefore not discussed in greater detail.^{2,83,198−201,266,342−347}

Scheme 54. Synthesis of Anionic Heteroleptic Bipyridine-silylamido Complexes of the 3d Metals³⁴⁴.

Reactivity studies evaluated the behavior of the silyl(aryl)amido complexes [KM{N(Dipp)SiMe₃}₂] [M = Mn (437), Fe (438), Co (439)] and of the hexamethyldisilazanide complexes [K([18]crown-6)][M{N(SiMe₃)₂}₂] [M = Cr (444), Mn (445), Fe (446), Co (447)] toward alkynes (Scheme 55).^332,351 For instance, [18]crown-6 adducts [K([18]crown-6)][M{N(Dipp)SiMe₃}₂] of 437 and 438 form adducts of the type [M{N(Dipp)SiMe₃}₂(η²-RC≡CR)]⁻ (470, M = Mn, R = Ph, Et; 471, R = Ph) with diphenylacetylene and 3-hexyne (Scheme 55a, right).^332,351 The molecular structures of these complexes feature side-on coordinated alkyne ligands. The alkyne ligands in 470 are labile.³⁵¹ Analogous reactivity was observed for 444–447 (Scheme 55b), affording compounds of general formula [K([18]crown-6)][M{N(SiMe₃)₂}₂(η²-RC≡CR)] [M = Cr (472), Mn (473), Fe (474), Co (475)], although some combinations of metal and alkyne resulted in deviating reactivity and products (vide infra). The authors concluded that, for this series (Scheme 55b), the chromium and manganese derivatives bind the alkyne ligand more strongly and more covalently and, consequently, the C≡C bond is more activated in those complexes, in contrast to the iron and cobalt analogues.³⁵¹

Scheme 55. Reactivity of (a) M(I)-Bis(silylamide) (as [K([18]crown-6)]⁺ Salts) and (b) Neutral and Ion-Separated M(I)-Silylamido Complexes by the Group of Werncke^332,351.

The chromium compounds shown on Scheme 55b (left) illustrate the different reactivity observed for the hexamethyldisilazanide complexes of the studied early transition metals.³⁵¹ The side-on coordinated alkyne complex 472 is formed by reaction of 444 and bis(trimethylsilyl)acetylene (btmsa). By contrast, the reaction of 444 with diphenylacetylene afforded the alkyne-bridged compound [K[18]crown-6}][Cr₂{N(SiMe₃)₂}₂{μ-N(SiMe₃)₂}(μ-1κC¹,2κC²–PhCCPh)] (476). The formation of 476 was attributed to partial oxidation of the metal atoms, with concomitant 2-fold reduction of the alkyne as a result of strong metal-to-alkyne ligand backbonding.³⁵¹

Treating 445 with btmsa similarly yielded [K[18]crown-6}][Mn₂{N(SiMe₃)₂}₂{μ-N(SiMe₃)₂}(μ–η²-Me₃SiCCSiMe₃)] (477), which features a reduced alkyne ligand and an oxidized metal framework, whereas analogous treatment with diphenylacetylene afforded the side-on complex 473. The latter species, however, reacts with additional equivalents of 445, via a series of C≡C activation processes, to generate first the dinuclear compound [K([18]crown-6]₂[Mn{N(SiMe₃)₂}₂(μ–κ¹:κ¹-PhCCPh)] (478), and then, in the presence of further substrate, the heteroleptic complex [K([18]crown-6)][Mn(η⁶-C₆Ph₆)(η²-C₂Ph₂)] (479) accompanied by formation of [Mn{N(SiMe₃)₂}₃]⁻ (480) as a byproduct (see Scheme 55b, right down). The alkyne thereby undergoes coordination, 2e-reduction, and cyclotrimerization during the transformation of complex 473 into 479. These processes involve a disproportionation of 445 as well as ligand exchange and metal-mediated substrate trimerization. Unambiguous determination of the oxidation state in 479 was inhibited by a lack of pure analytical samples. The elongated C–C bond of the alkyne [1.346(7) Å] according to X-ray diffraction data led the authors to suggest that 479 could be described as a manganese(I) metallacyclopropene compound. However, its depiction as a formally manganese(−I) alkyne complex was also discussed, which would account for redox disproportionation leading to the formation of the manganese triamide 480.³⁵¹

Analogously, the more encumbering neutral (contact ion pair) compounds [KM{N(Dipp)SiMe₃}₂] also engage in metal-mediated trimerizations of diphenylacetylene, probably as the result of redox disproportionation of the corresponding metal complex. For the manganese complex 437, the characterization of the unusual species 481 and 482 (Scheme 55a, left) was achieved. Such species are probably intermediates en route to the cyclotrimerization of PhC≡CPh.³³² In the case of the iron complex 438, the trimerization afforded the previously reported anion [Fe(η⁶-C₆Ph₆)(η²-C₂Ph₂)]⁻ (192, vide supra), analogous to 479.²¹¹

For the bis(silylamide) complex [Co{N(SiMe₃)₂}₂]⁻ (447) and the more sterically encumbered precursor 439, a weak and reversible coordination of the alkyne motif to the metal center, attributed to the anionic nature of the precursors,^332,351 was identified for all the substrates (RC≡CR with R = Ph, SiMe₃, Et, and nPr). Despite such weak coordination, the reaction product between 447 and diphenylacetylene, [K([18]crown-6)][Co{N(SiMe₃)₂}₂(η²-PhC≡CPh)] (475) was isolated and crystallographically characterized. Computational analysis of the hexamethyldisilazanide/alkyne compounds [K([18]crown-6)][M{N(SiMe₃)₂}₂(η²-RC≡CR)] supports the structural observations indicating a partial reduction of the alkyne ligands by the metal centers. Consequently, the compounds were described as metal(II) complexes with covalently bound alkyne ligands bearing some radical character (formal 1e^– reduction of the alkyne ligands).³⁵¹

Further reactivity studies showed that the quasilinear complexes [M{N(SiMe₃)₂}₂]⁻ [M = Fe (446), Co (447)] behave as reductants toward ketones, aldehydes, and imines, generating the corresponding M(II) stabilized radical anions.³⁵² Examples of the latter are the anionic ketyl complexes [K(chelate)][M(bp){N(SiMe₃)₂}₂] [M = Fe (483), Co (484); bp = benzophenone, chelate = [18]crown-6 or [2.2.2]cryptand], which were evaluated in terms of their ability to engage in radical-like reactivity, specifically in hydrogen atom abstraction. Whereas the iron(II) complex 483 did not react with the H atom donor 1,4-cyclohexadiene (1,4-CHD), 484 promoted its dehydrogenation to benzene.

Furthermore, the reaction of 484 with an excess/of 1,4-CHD formed not only benzene (see Scheme 56) but also a bis(1,3-cyclohexadiene)cobaltate(−I) complex, [Co(η⁴-1,3-C₆H₈)₂]⁻ (485), and the tris(silylamido)cobalt(II) complex [Co{N(SiMe₃)₂}₃]⁻ (486).³⁵² The structure of 485 is analogous to the bis(butadiene)cobaltate 248 and to the bis(anthracene)cobaltate 14 (vide supra).⁸⁵ Moreover, the authors proved that the ketyl ligand in 484 is not required to obtain 485 since it was also independently synthesized by reactions of 447 (3 equiv) with either 1,3- or 1,4-CHD (see Scheme 56). The reaction also produces the Co(II) complex 486, thereby representing a disproportionation of the Co(I) complex 447 into Co(II) and Co(−I). Together, the independent synthesis of 485 from 447 and the reactivity observed with the ketyl radical complex 484 indicated that the role of the latter involved promotion of the H atom abstraction process.³⁵²

Scheme 56. Formation of a Bis(1,3-cyclohexadiene)cobaltate(−I) Complex by Two Different Reaction Pathways, Described by the Group of Werncke³⁵².

In 2015, Tilley and co-workers demonstrated the potential of a low-valent and low-coordinate heteroleptic carbene/silyl(aryl)amido Fe(I) complex as a precatalyst in the trimerization of alkynes.³³⁵ The neutral compound [Fe{N(Dipp)(SiMe₃)}(IPr)] (487, IPr = 1,3-bis(2,6-diisopropylphenyl)imidazolin-2-ylidene, Dipp = 2,6-diisopropylphenyl) was obtained via a one-pot reaction, by protonating in situ generated [Fe{N(Dipp)SiMe₃}₂] (440),³⁴¹ adding the carbene ligand and finally reducing the formed species with KC₈ (Scheme 57). Further reduction of neutral 487 with KC₈ yielded a rearranged product, K[(η⁶-IPr)Fe{N(Dipp)(SiMe₃)}] (488). Here, the carbene ligand coordinates to the metal atom as a π-bonded arene ligand, with the C_carbene-atom interacting with the potassium cation. ⁵⁷Fe Mössbauer spectroscopic analysis confirmed that 487 is an Fe(I) compound and that the reduction occurs at the metal center, making product 488 an Fe(0) species. Additionally, 487 was found to catalyze the cyclotrimerization of both internal and terminal alkynes at catalyst loadings of 2–5 mol % under mild conditions to afford substituted arenes.³³⁵

Scheme 57. Synthesis of a Heteroleptic Carbene/Silyl(aryl)amido Iron(I) Complex and Its Reduction to Afford the Fe(0) Complex 488(335).

2.2.8.2. Amido Ligands with Pendant Donors

Gambarotta reported low-valent titanium and vanadium complexes featuring σ-/π-bonded pyrrolide-based ligands.^192,193 A tripyrrolide ligand served as platform for the synthesis of a mixed-valent anionic titanium complex.¹⁹² Treatment of [TiCl{2,5-[(C₄H₃N)CPh₂]₂[C₄H₂N(Me)]}] (489) with potassium in toluene, followed by recrystallization from DME, afforded complex [K(dme)₂][{2,5-[(C₄H₃N)CPh₂]₂[C₄H₂N(Me)]}Ti(μ,η⁶:η⁶-C₇H₈)Ti[{2,5-[(C₄H₃N)CPh₂]₂[C₄H₂N(Me)]}]·toluene (490, Scheme 58a) which has an inverse sandwich-type structure. The dianionic polydentate ligand framework in 490 has three pyrrole rings, one of which is π-coordinated to each titanium center. The metric parameters in the crystal structure point toward the presence of a mixed valent complex. Furthermore, the bridging toluene ligand is significantly distorted, suggesting increased metal-to-ligand backbonding. In turn, the magnetic moment (μ_eff = 1.73 μ_B) indicates that the species has one unpaired electron. These findings were corroborated by DFT calculations and 490 was proposed to be a formally Ti(I)/Ti(II) complex. By reducing 489 with potassium in the presence of trans-stilbene, a mixture of two diamagnetic complexes was obtained after crystallization, [K(dme)][Ti{2,5-[(C₄H₃N)CPh₂]₂[C₄H₂N]}(η²-trans-PhHC=CHPh)] (491) and [Ti{2,5-[(C₄H₃N)CPh₂]₂[C₄H₂N(Me)]}(η²-trans-PhHC=CHPh)] (492, Scheme 58). The tripyrrrolide ligand has lost the methyl substituent in 491, through an unknown mechanism, thereby becoming trianionic. The elongated C–C bond distance (1.443(6) Å) of the coordinated trans-stilbene ligand reflects the increased electron density and metal-to-ligand backbonding in 491. According to the authors, 491 can be considered either a Ti(II)- or a Ti(IV)-species. Later, the same group achieved the reduction of N₂ to nitride at tripyrrolide Ti(III) neutral complexes.³⁵³

Scheme 58. Synthesis of Anionic Titanium or Vanadium Complexes Featuring σ-/π-Bonded Pyrrolide-Based Ligands^192,193.

A related dipyrrolidephenyl ligand scaffold supported the V(III) or V(II) complexes [(dme)VCl{1,3-[(C₄H₃N)CMe₂]₂(C₆H₄)}] (493) or [(thf)₃V{1,3-[(C₄H₃N)CMe₂]₂(C₆H₄)}] (494), respectively.¹⁹³ Compounds 493 and 494 reacted with the appropriate amount of KH to the low-valent anionic dinuclear V(I) complex [{(thf)₂K}V{1,3-[(C₄H₃N)CMe₂]₂(C₆H₄)}]₂ (495, Scheme 58b). The vanadium atoms in 495 are connected through bridging interactions with the central phenyl ring of the ligand framework, and a weak V–V bond. The two phenyl rings are significantly distorted, indicating strong covalent bonding between each vanadium atom and the ring π-system. Theoretical calculations indicated that, despite these interactions, the central phenyl ring has a net charge of almost zero, and therefore, the distortion does not affect the formal oxidation state of the metal centers. Consequently, the authors concluded that, in 495, the metal atoms are formally in the d⁴ V(I) configuration. A striking contrast in behavior was observed between complex 494 and the analogous vanadium species with the tripyrrolide ligand: whereas 494 is inert toward N₂ coordination, the related neutral tripyrrolide-vanadium complex mediated nitrogen fixation and cleavage.³⁵⁴

Recently, Fortier and co-workers reported that the reduction of the dimeric titanium compound [Cl₂Ti(μ-NIm^Dipp)]₂ (496) (NIm^Dipp = [1,3-bis(Dipp)imidazolin-2-iminato]⁻) by 4 or 6 equiv of KC₈ generates the dinuclear compounds [(μ-N-η⁶-Im^Dipp)Ti]₂ (497) and {[(Et₂O)₂K](μ-N-μ–η⁶:η⁶-Im^Dipp)Ti}₂ (498), respectively (Scheme 59).³⁵⁵ X-ray structural analyses suggested that the η⁶-bound Dipp groups in both compounds had undergone two-electron reduction, lending formal oxidation state assignments of Ti(III)/Ti(III) for neutral 497 and Ti(II)/Ti(II) for anionic 498. However, reactivity of these species with small organic molecules revealed reducing capabilities beyond their formal oxidation states.

Scheme 59. Synthesis and Reducing Capabilities of the Dinuclear Anionic Complex {[(Et₂O)₂K](μ-N-μ–η⁶:η⁶-Im^Dipp)Ti}₂ (498).

The reaction of 498 with 4 equiv of AdN₃, followed by addition of [18]crown-6, allowed isolation of the tris(imido) complex [K([18]crown-6)(thf)₂]{[(Im^DippN)Ti(NAd)](μ-NAd)₂K[Ti(NIm^Dipp)]} (499), consistent with the 6-electron reduction of 3 equiv of AdN₃ with loss of N₂. Additionally, reaction of 498 with 4 equiv of cyclooctatriene (COT) yielded a polymeric species of repeat unit [(Im^DippN)(η⁴-COT)-Ti(μ–η⁴:η⁶-COT)K(thf)(μ–η⁵:η⁴-COT)Ti(NIm^Dipp)(μ–η⁴:η⁴-COT)K(thf)₂]_n (500). Inspection of the C–C bonds in the coordinated COT rings of the solid-state structure revealed either localized -ene dianionic character or charge delocalization consistent with an overall 8-electron oxidation of 498. Thus, the authors proposed that, despite the formal assignment of its metal centers as Ti(II), compound 498 acts as a masked anionic Ti(0)/Ti(0) species. Related experiments implicated the neutral compound 497 as a similarly masked Ti(I)/Ti(I) species.

It is worth mentioning that Khusniyarov, Meyer, Mindiola, and co-workers have isolated highly reduced, dinuclear dipyrrolide pyridine Fe complexes with the ligand ^tBupyrr₂py^2– (^tBupyrr₂py^2– = 2,6-bis((3,5-ditert-butyl)pyrrol-2-yl)pyridine), which feature nonlinearly bridged dinitrogen ligands.³⁵⁶ Dianionic and even tetraanionic species with Fe(μ₂-η¹:η¹-N₂)Fe cores were isolated. Spectroscopic and quantum-chemical studies suggest that these compounds contain high-spin Fe(II) centers, even in the case of the tetraanionic species, and reduced ligand scaffolds or dinitrogen units. The chemistry of these species is, therefore, out of the scope of this review.

Lichtenberg, Grützmacher, and co-workers described the synthesis and redox properties of a low-valent anionic [Fe₄N₄]⁻ heterocubane cluster stabilized by an imido-olefin ligand.³⁵⁷ Upon reaction between complex [Fe(NCy₂)₂] (501) and the H₂Ntrop ligand (trop = 5H-dibenzo[a,d]cyclohepten-5-yl) at room temperature in toluene, the neutral heterocubane [Fe₄(Ntrop)₄] (502) assembles (Scheme 60). Cluster 502 can be reduced by reaction with either Cp*₂Co or a sodium amalgam to afford [Cp*₂Co][Fe₄(Ntrop)₄] (503) or [Na(thf)₄][Fe₄(Ntrop)₄] (504), respectively (Scheme 60). Reaction of the anionic [Fe₄N₄]⁻ cluster with [Cp*₂Co]I yielded compound 502 again. Cluster 502 has a heterocubane structure in which each iron atom coordinates an olefin of one trop unit in the deprotonated ligand Ntrop^2–, and the general structural parameters are maintained upon reduction. The neutral cluster 502 possesses a 56-valence-electron Fe₄ unit.

Scheme 60. Formation of Iron/Nitrogen Heterocubane Clusters by Lichtenberg, Grützmacher, and Co-workers³⁵⁷.

Cyclic voltammetry of 502 (THF, 23 °C) revealed four redox events between −1.72 and −3.59 V (vs Fc/Fc⁺) that were assigned to four consecutive formal Fe^II/Fe^I redox couples and indicate that [Fe₄N₄]ⁿ exists in five different oxidation states (n = 0, 1–, 2–, 3–, 4−). Therefore, the cluster 502 can store up to four electrons. Analysis of the electronic structure of the reduced species revealed that, in the ground state, the unpaired electron is localized predominantly at one iron center and fluctuates with increasing temperatures. Due to the chemical reversibility of the [Fe₄(Ntrop)₄]/[Fe₄(Ntrop)₄]⁻ couple, the authors evaluated the use of this system as an electron-transfer catalyst for C–C bond couplings.

In 2008, Tejel and de Bruin discovered an exceptional redox asymmetric dinuclear Rh^–I–Rh^I complex bearing a doubly deprotonated bis(2-picolyl)amine (bpa) scaffold and two norbornadiene (nbd) ligands.³⁵⁸ The two-electron mixed-valence Rh^–I–Rh^I complex [{Rh(nbd)}₂(μ-bpa-2H)] (505) was obtained by deprotonation of the homovalent Rh^I–Rh^I compound [{Rh(nbd)}₂(μ-bpa-1H)]Cl (506) or by reaction of the dimer [Rh(nbd)(μ-OMe)]₂ (507) with bis(2-picolyl)amine (Scheme 61a). After double deprotonation of bpa, an electronic reorganization takes place, formally reducing Rh(I) to Rh(−I) and oxidizing bpa-2H to the neutral imine PyCH₂N=CHPy (Py = pyridine). Species 505 can be seen as a rhodate(−I) anion for which the counterion is the second rhodium atom, formally a Rh(I) center. Indeed, the distinct nature of the two Rh centers was confirmed both in the solid state and in solution. In the molecular structure of 505, the Rh(−I) atom is in an almost tetrahedral geometry while the Rh(I) center is, typically, square planar. The bridging PyCH₂N=CHPy moiety binds to Rh(I) as a σ-imine through the central N donor and to Rh(−I) as η²-imine, rather than forming a rhoda-aza-cyclopropane.³⁵⁹ In solution, the nbd bound to Rh(I) is almost static whereas the nbd fragment bound to the tetrahedral Rh(−I) is dynamic, as observed by ¹H NMR spectroscopy. Moreover, EXSY revealed that a 1,3-prototropic shift of one of the methylene protons to the CH=N group takes place, suggesting that electrons are pumped from one metal to the other via the ligand, exchanging their formal oxidation state. The authors did not rule out the possibility of having a singlet diradical structure with a monoanionic ligand and Rh⁰–Rh^I centers, as an alternative electronic description for 505.³⁵⁸

Scheme 61. Synthesis of a Family of Mixed-Valent (bpa) M^–I–M^I (M = Rh, Ir; bpa = Bis(2-picolyl)amine) Complexes^{358,360−362}.

Later, the same authors further examined this chemistry by using cod (cod = 1,5-cyclooctadiene) instead of nbd and extending the study to iridium.³⁶² By mixing [M(cod)(bpa)]⁺ [M = Rh (508), Ir (509)] and [M(cod)(bpa-2H)]⁻ [M = Rh (510), Ir (511)], the related complexes [{M(cod)}₂(μ-bpa-2H)] [M = Rh (512), Ir (513)] were obtained. As with 505, the related complex 512 can be described as a mixed-valent Rh^–I–Rh^I species, exhibiting both very similar spectroscopic and structural features. Interestingly, in the case of iridium, the reaction mixture [Ir(cod)(bpa)]⁺ (509) plus [Ir(cod)(bpa-2H)]⁻ (511) affords the Ir^–I–Ir^I analogue [{Ir(cod)}₂(μ-bpa-2H)] (513) in solution and one equivalent of free bpa, which further reacts with the dinuclear complex via reprotonation and electronic rearrangement to yield the neutral amido mononuclear compound [Ir(cod)(bpa-H)] (514, Scheme 61b).

The preparation of the related Rh-amido complex [Rh(cod)(bpa-H)] was unsuccessful and led instead to the mixed-valence species 512 and free bpa. The thermodynamic differences in [{M(cod)}₂(μ-bpa-2H)] [M = Rh (512), Ir (513)] were explained in terms of the lower stability of the M(−I) oxidation state for iridium as compared to rhodium. This was further evidenced by the fact that the heterodinuclear Rh(−I)–Ir(I) compound [Rh(cod)(μ-bpa-2H)Ir(cod)] (515) can be readily isolated upon metal exchange between 512 and 513. In 515, the Rh(−I) center is located in the tetrahedral compartment of the μ-bpa-2H ligand and the Ir(I) atom lies in the square planar compartment (Scheme 61c).

This family of anionic amido-olefin complexes reacts with molecular oxygen resulting in the monooxygenation of the bpa-2H scaffold instead of the diolefins, as would be anticipated.^{358,360−362} Another example of mixed-valence Rh species was reported by Goswami and McGrady in 2010 with an azopyridine ligand. This presumed Rh(–I)–Rh(III) complex exhibits a significant rhoda-aza-cyclopropane character; thus, the formal oxidation states appear in fact to lie in between Rh^I–Rh^III and Rh^III–Rh^III.³⁶³

2.2.9. Metalates Featuring β-Diketiminato Ligands

Bill, Hoffman, Holland, and co-workers reported β-diketiminate-supported iron(I) complexes of relevance to biochemistry.^364,365 Reduction of the iron(II) hydride [(^tBunacnac)Fe(μ-H)]₂ [516, ^tBunacnac = CH[C(tBu)N(2,6-iPr₂C₆H₃)]₂] by a slight excess of KC₈ yielded the dimeric iron(I) hydride K₂[(^tBunacnac)FeH]₂ (517, see Scheme 62a).³⁶⁴ Furthermore, sequestration of the potassium cation was effected by chelating agents such as [18]crown-6 and [2.2.2]cryptand to give the monomeric, three-coordinate ion-separated complexes [K(chelate)][(^tBunacnac)FeH] [chelate = [18]crown-6 (518), [2.2.2]cryptand (519)].

Scheme 62. Synthesis of (a) the Fe(I) Hydride Complex K₂[(^tBunacnac)FeH]₂ (517) and (b) the Fe(I) Sulfide Complexes [AM]₂[{(^Me3nacnac)Fe}₂(μ-S)] (521, 522)^364,365.

Magnetic measurements, Mössbauer spectroscopy, and DFT calculations of 517–519 were consistent with high-spin S = 3/2 iron(I) centers in each case. Calculations suggested that π-backbonding into the β-diketiminate assists the ligand in supporting the very electron-rich metal center. The solid-state structures of 518 and 519 each showed a terminal Fe–H moiety, and ENDOR analysis revealed a near-axial anisotropic coupling tensor (T) expected for a terminal hydride ligand for 518. Similar to this hydride example, reduction of the sulfide-bridged diiron(II) complex [{(^Me3nacnac)Fe}₂(μ-S)] [520, ^Me3nacnac = MeC[C(Me)N(2,6-iPr₂C₆H₃)]₂] by KC₈ or Na afforded the corresponding diiron(I) metalate [AM]₂[{(^Me3nacnac)Fe}₂(μ-S)] [AM = Na (521), K (522)] (Scheme 62b).³⁶⁵ The assignment of the Fe(I) oxidation states was supported by crystallography, Mössbauer spectroscopy, magnetic susceptibility measurements, and DFT calculations. The stability of the low-valent iron complex was attributed to steric protection of the Fe–S–Fe core, and favorable interactions of this anionic core with the cations. Unusual linear μ-sulfido bridges were observed in the crystal structures of 521 and 522, possibly due to geometric restraints from the π-interactions between the cations and aryl groups. The alkali metal cations underwent exchange in solution, with ¹H NMR spectroscopic monitoring of a solution of 521 and 522 revealing the formation of a third species, assigned as the mixed-cation complex.

Arnold and co-workers have extensively studied the synthesis and reactivity of rhenium metalates featuring β-diketiminato ligands.^10,366−370 The anionic Re(I) complex Na[(η⁵-Cp)Re(nacnac)] (Na[523]; nacnac = N,N′-bis(2,6-diisopropylphenyl)-3,5-dimethyl-β-diketiminate) displays a rich chemistry which includes, among other reactions, N₂ coordination and trapping and bond formation, as well as serving as a precursor to other metal complexes, including heterometallic compounds. The variety of reactions mediated by 523 is based on its ability to act as a base, nucleophile, or reductant.

The chemistry of this reactive Re(I) β-diketiminate cyclopentadienide and related species was comprehensively reviewed.¹⁰ Therefore, we will only highlight some of the key transformations exerted by either 523 or some of its derivatives. Na[523] was obtained by reduction of the oxo rhenium(V) cation [ORe(η⁵-Cp)(nacnac)]⁺ (524) with sodium.³⁶⁶ Protonation of metalate Na[523] with [Et₃NH][BPh₄] afforded the terminal hydride [(η⁵-Cp)ReH(nacnac)] (525, see Scheme 63), whereas its oxidation with AgOTf afforded a rare neutral open-shell Re(II) complex [(η⁵-Cp)Re(nacnac)] (526).³⁶⁶ The unsaturated Re(II)-d⁵ complex 526 activated H₂, providing a different route to access the terminal hydride 525, by formal H atom transfer (Scheme 63).³⁶⁷ In addition, 526 activates P₄ at room temperature to yield a Re(V) species, [(η⁵-Cp)Re(η³-cyclo-P₃)(nacnac)] (527). The reducing character of Na[523] was demonstrated with a variety of substrates. For example, Na[523] reacted with ZnCl₂ to afford the tetranuclear compound [(η⁵-Cp)Re(nacnac)Zn]₂ (528) and 526.³⁶⁶ The molecular structure of 528 exhibits a nearly linear Re(I)–Zn(I)–Zn(I)–Re(I) arrangement (173.563(16)°) with a Zn₂²⁺ core flanked by datively bound anionic [(η⁵-Cp)Re(nacnac)]⁻ fragments on each side (Scheme 63). Moreover, by treating Na[523] with Me₃SiCl under a nitrogen atmosphere at −78 °C, the Re(III) complex [(η⁵-Cp)Re(nacnac)(N=NSiMe₃)] (529) was obtained (Scheme 63).³⁶⁹ The structure of 529 features a silyldiazenide ligand, the product of N₂ functionalization at the rhenium metalate Na[523]. The authors demonstrated that the N₂ trapping reactivity is highly dependent on ion pairing interactions since a similar silylation using the ion-separated complex [Na(benzo[12]crown-4)₂][523] afforded only small amounts of product. A related outcome was obtained from Na[523] and [MCl(cod)]₂ [M = Rh (401), Ir (530)] under a nitrogen atmosphere.³⁶⁸ The products are heterobimetallic complexes [(η⁵-Cp)Re(μ-nacnac)(μ-N₂)M(η⁴-cod)] [M = Rh (531), Ir (532)] featuring a bridging diazenido ligand end-on coordinated to the rhenium center and the group 9 atom (Scheme 63). The N≡N stretching vibrations [ν_N≡N(531) = 1767 cm^–1 and ν_N≡N(532) = 1740 cm^–1] indicate that the N₂ ligand is significantly activated, allowing for its subsequent functionalization. In turn, rhenium metallotetrylenes [(η⁵-Cp)Re(E[PhC(NtBu)₂])(nacnac)] [E = Si (533), Ge (534), Sn (535); Scheme 63 shows the Si and Ge derivatives] were obtained by salt metathesis between Na[523] and amidinate-supported tetrylenes, ClE[PhC(NtBu)₂].³⁷⁰ For the silicon and germanium analogues, the reaction yielded complexes with short Re–E multiple bonds, as a result of π-interactions with the tetrel atoms. By contrast, the tin derivative has a Re–Sn single bond formed by σ-donation to the Re atom, in a σ-metallotetrylene arrangement, and retains its electron lone pair.

Scheme 63. Reactivity of the Re(I) Metalate Na[(η⁵-Cp)Re(nacnac)] (Na[523]).

2.2.10. Redox-Active Diimine Scaffolds

Diimine ligands are redox-active N,N′-donors with a N=C–C=N backbone. This class of ligands includes various diimine scaffolds, e.g., bis(imino-acenaphthene) (BIAN), o-benzoquinonediimine (1,2-phenylenediamine dianion), 2,2′-bipyridine, 1,4-diazabutadiene, and imino-pyridine.³⁷¹ The chemistry of BIAN ligands and their use in metal-catalyzed reactions have been recently reviewed.³⁷² The interest in this type of ligands, particularly in combination with low-valent iron, is reflected by a large number of contributions.^373−381 Nonetheless, although there have been numerous contributions to the field in recent years, only selected examples have resulted in the formation of anionic complexes.

2.2.10.1. α-Diimine Ligands

Wolf and co-workers described the synthesis and characterization of anionic BIAN-iron complexes, which were used as precatalysts in hydroboration reactions.³⁷⁶ Reduction of the [(^DippBIAN)FeBr₂] (536, ^DippBIAN = 1,2-bis(2,6-diisopropylphenylimino)acenaphthene) precursor by stoichiometric KC₈ in the presence of cod afforded [K([18]crown-6)(thf)_0.5][(^DippBIAN)Fe(η⁴-cod)] ([K([18]crown-6)(thf)_0.5][537], see Scheme 64a). A similar complex [Li(thf)(^DippBIAN)Fe(η⁴-cod)] (Li[537]) is also accessible by ligand exchange from the formally Fe(−II) homoleptic complex [Li(dme)₂]₂[Fe(cod)₂] ([Li(dme)₂]₂[16], Scheme 64a), albeit in lower yield. The mechanism of the latter reaction, which involves an oxidation of the Fe complex [Li(dme)₂]₂[16] by one electron, is not well-understood. An analysis of the X-ray structural data for Li[537] suggests the presence of a dianionic ^DippBIAN^2– ligand as indicated by the metric data of the ^DippBIAN moiety. Elongated C–N [1.385(4) Å] and shortened C–C [1.388(5) Å] bond lengths are observed in comparison with the free ^DippBIAN molecule and its alkali metal salts (cf. C–N 1.39 Å and C–C 1.40 Å, respectively, for Na₂[^DippBIAN]).^371,382 The proposed presence of a doubly negatively charged ^DippBIAN^2– ligand implies an oxidation state of +I with a d⁷ configuration for the Fe atom. Attempts to synthesize a 2,5-norbonadiene (NBD) analogue of 537 by the same reduction pathway from 536 were unsuccessful. Instead, the isolated compound [K([18]crown-6)(thf)₂][(^DippBIAN)Fe(C₁₄H₁₆)] (538, C₁₄H₁₆ = 2,2′-bi(bicyclo[2.2.1]heptane-5,5′-diene-3,3′-diyl) contains a bis(norbornenediyl) ligand derived from the C–C coupling of two nobornadiene molecules (Scheme 64a, bottom right). Additionally, the bis(anthracene)ferrate(−I) complex 52 reacted with DippBIAN to give only the homoleptic compound [K([18]crown-6)(thf)₂][Fe(^DippBIAN)₂]^DippBIAN the homoleptic compound [K([18]crown-6)(thf)][Fe(^DippBIAN)₂] (539) regardless of the metal-to-ligand ratio used (Scheme 64b). All the complexes were structurally characterized, and their physical and electronic properties were analyzed by spectroscopic methods (NMR, EPR, ⁵⁷Fe Mössbauer, UV–vis), magnetic susceptibility measurements, and theoretical calculations (DFT, CASSCF). These data provided evidence to conclude that complex salts of 537 have a low-spin ground state, while complex 538 features an intermediate-spin Fe(III) center, and complex 539 is an Fe(II) species in a quartet ground state. Complexes 537–539 were evaluated as precatalysts in hydroboration reactions; these results are discussed in section 3.1.6.³⁷⁶

Scheme 64. Synthesis of Anionic Fe-^DippBIAN Complexes by the Wolf Group³⁷⁶.

Heteroleptic cobaltate complexes, bearing α-diimine and alkene/arene ligands, were also described.^383−387 Using the Co^– source [K(thf)][Co(η⁴-cod)₂] (239), Wolf and co-workers synthesized the [K(OEt₂){Co(^ArBIAN)(η⁴-1,5-cod)}] [^ArBIAN: ^DippBIAN = 1,2-bis(2,6-diisopropylphenylimino)acenaphthene (540), ^MesBIAN = 1,2-bis(2,4,6-dimethylphenylimino)acenaphthene (541); 1,5-cod = 1,5-cyclooctadiene]^383,384 and [K([18]crown-6)(thf)_1.5][Co(PHDI)(η⁴-1,5-cod)] [542, PHDI = bis(2,6-diisopropylphenyl)phenanthrene-9,10-diimine)]³⁸⁵ (see Scheme 65), which are closely related to 537 discussed earlier. Again, X-ray diffraction analysis suggested the presence of BIAN^2– ligands in 540 and 541 [av. C–N 1.381 Å (540), 1.374 Å (541); C–C 1.383(3) Å (540), 1.377(5) Å (541)]

Scheme 65. Synthesis of α-Diimine Cobaltates by the Wolf Group^383−385.

The related phenanthrene-9,10-diimine complex 542 was not crystallographically characterized, though the ¹H and ¹³C NMR data compared well with those of the ^DippBIAN complex 540.³⁸⁵ This observation contrasts with those for the previously reported complex [Ni(^DippBIAN)(η⁴-1,5-cod)] (543), which is isoelectronic with 537 and 540.³⁸⁸ Nonetheless, the structural data indicate the presence of a BIAN^– ligand. The presence of a more reduced BIAN^2– ligand in 537 and 540 may be attributed to the lower effective nuclear charge of iron and cobalt in comparison with nickel.³⁸³ Catalytic applications of complexes 540 and 541 are described in section 3.1.

Reactivity studies of the diimine cobaltates 540 and 541 revealed that the BIAN ligand mostly serves as a spectator ligand, whereas the cyclooctadiene ligand is readily displaced by other molecules.^383,389 For instance, the reaction of [K(thf){Co(^MesBIAN)(η⁴-1,5-cod)}] (541) and 1,2-bis(diphenylphosphino)-orthocarborane (544) affords a 13-vertex closo-cobaltacarborane cluster compound 545 (Scheme 66). The product is a result of substitution of the 1,5-cod ligand and subsequent polyhedral expansion of the carborane framework. In the solid-state molecular structure, the 13-vertex cobaltacarborane cluster anion forms as a contact ion pair with the [K(dme)₃]⁺ counterion. Shorter C–N (av. 1.332(2) Å) and C–C distances (1.420(2) Å) in the BIAN ligand framework with respect to 540 and 541 indicate the presence of a monoanionic BIAN^– ligand in 545. This notion is supported by DFT molecular orbital analysis of the complex. DFT calculations, the crystallization of intermediates and spectroscopic observations (specifically ³¹P{¹H} NMR reaction monitoring and ESI-MS studies), enabled the authors to propose a mechanism for the formation of 545. An initial electron transfer from 541 to 544 is followed by coordination of the resulting dianion 544^2– to the cobalt center. Oxidation and isomerization of the resulting complex gives 545.³⁸⁹

Scheme 66. Synthesis of the Anionic 13-Vertex Closo-cobaltacarborane Cluster 545 from the (BIAN)cobaltate 541(389).

2.2.10.2. Pyridine-diimine Ligands

Chirik and Roşca independently reported anionic iron complexes with pyridine-diimines (PDIs) and closely related pyrimidine and pyrazinediimines.^136,390,391 Cyclic voltammetry of the previously reported [(^DippPDI)Fe(CO)₂] (546)³⁹² showed reversible one-electron oxidation and reduction events.¹³⁶ The accessibility of these products was later confirmed by the preparation of the cationic compound [(^DippPDI)Fe(CO)₂][BAr^F₄] (547), obtained by oxidation of 546 with [Cp₂Fe][BAr^F₄], and of the anionic complex [Na([15]crown-5)][(^DippPDI)Fe(CO)₂] (548, Scheme 67), synthesized by reduction of 546 with 0.5% Na(Hg) in the presence of [15]crown-5 (1.2 equiv, Scheme 67). Complex 548 is highly air-sensitive and prone to thermal decomposition. FT-IR spectroscopy showed two carbonyl stretching bands (ν_CO = 1935 and 1863 cm^–1) at lower wavenumbers in comparison with 546 (ν_CO = 1974 and 1914 cm^–1). These data are consistent with the expected increase in electron density at the reduced iron center. DFT calculations indicated that the anionic complex 548 features a strongly reduced, diradical dianionic bis(imino)pyridine ligand³⁹³ and is thus best described as a low-spin Fe(I) complex with a high degree of covalency.¹³⁶

Scheme 67. Reduction of Complexes [LFe(CO)₂] (L = PDI, P^PymDI, P^PzDI; top) and [(^AdP^PymDI)Fe(N₂)] (bottom)^{136,390,391,402}.

By synthesizing carbonyl complexes analogous to 546 featuring either the pyrimidine- [^RP^PymDI: R = tert-butyl (tBu), adamantyl (Ad)] or pyrazine-based (P^PzDI) ligands, Roşca and co-workers demonstrated that the redox potentials of such complexes reflect the differences in π-acidity of the central heterocycle of the ligand.³⁹¹ Compound [(^tBuP^PymDI)Fe(CO)₂] (549) was obtained from the free ^tBuP^PymDI ligand and Fe(bda)(CO)₃ (550, bda = benzylideneacetone), whereas displacement of the dinitrogen ligand in [(^AdP^PymDI)Fe(N₂)] (551) by carbon monoxide generates the adamantyl-substituted compound [(^AdP^PymDI)Fe(CO)₂] (552). Similar to the previous report on 546 (vide supra),¹³⁶549 can also be chemically oxidized with [Cp₂Fe][BAr^F₄], yielding [(^tBuP^PymDI)Fe(CO)₂][BAr^F₄] (553). Likewise, reduction of 549 with KC₈ in the presence of [18]crown-6 affords [K([18]crown-6)][(^tBuP^PymDI)Fe(CO)₂] (554) (see Scheme 67, top).³⁹¹ Magnetic susceptibility measurements performed in solution with the Evans NMR method indicate that a one-electron reduction process has occurred (doublet state, S = 1/2, μ_eff = 1.7 μB). Similar synthetic protocols were used in an attempt to obtain the oxidized and reduced derivatives of the P^PzDI complex [(P^PzDI)Fe(CO)₂] (555). However, whereas a metal-centered oxidation led to [(P^PzDI)Fe(CO)₂][BAr^F₄] (556),³⁹⁰ the reduction of 555 with KC₈ in the presence of [18]crown-6 afforded a ligand-based radical which dimerizes to form a dinuclear compound, 557, with an additional C–C bond (see Scheme 67). In fact, the one-electron reduction of the three complexes [LFe(CO)₂] (546, 549, and 555, L = PDI, ^RP^PymDI, and P^PzDI, respectively) is ligand-based in each case, irrespective of the nature of the heterocyclic support. By contrast, one-electron oxidation is a metal-centered process, affording a metalloradical.^136,391 Contrasting behavior was observed for the reduction of the dicarbonyl complex [(^AdP^PymDI)Fe(CO)₂] (552) and that of the related dinitrogen complex [(^AdP^PymDI)Fe(N₂)] (551). While in both cases the one-electron reduction is ligand-based, for 552 a doubly reduced ligand state is preferred, initiating a metal-to-ligand electron transfer, which results in an Fe(I) center which is antiferromagnetically coupled with a ^AdP^PymDI^2– diradical. Reduction of 551 affords the anionic complex [Na([18]crown-6)(thf)₂][(^AdP^PymDI)Fe(N₂)] (558, Scheme 67, bottom), in which the ^AdP^PymDI ligand stores three electrons, while the metal center appears to attain an Fe(II) oxidation state. This example illustrates the striking redox non-innocent behavior of the ^AdP^PymDI ligand.³⁹¹

Chirik and co-workers reported additional examples of anionic manganese, iron, or cobalt complexes featuring PDI ligands.^{136,394−400} The complexes [(^DippPDI)Mn(CO)₂]⁻ (559), [(^DippPDI)Co(N₂)]⁻ (560), and [(^iPrAPDI)Co(N₂)]⁻ (561) were synthesized by reduction of the appropriate neutral precursors with either sodium amalgam or sodium naphthalenide.^395−397 According to an analysis of the metric parameters, 559–561 feature dianionic PDI ligands (Figure 11). The electronic structures of 559–561 differ from that in [(^DippPDI)Fe(CO)₂]⁻ (548),¹³⁶ in which the ligand was reduced to the diradical dianionic form.³⁹³ Further examples of anionic complexes are the alkyl- and aryl-substituted compounds [(^DippPDI)Fe(N₂)(CH₂CMe₃)]⁻ (562), [(^DippPDI)Fe(N₂)(p-C₆H₄-R)]⁻ [R = H (563), Me (564)], which feature ferrous metal centers and doubly reduced PDI scaffolds.^398,399562–564 are reminiscent of the previously reported bis(imino)pyridine iron methyl complex, [Li(thf)₄][(^DippPDI)FeMe] (565) by Gambarotta and co-workers.⁴⁰¹

Figure 11.

Anionic manganese or cobalt complexes featuring reduced PDI ligands.^395−397

Gambarotta and Budzelaar investigated the reduction of PDI-iron dinitrogen complexes.⁴⁰³ The reaction of the Fe(II) complex [(^DippPDI)FeCl₂] (566) with 3 equiv of NaH formed the ion-paired species [Na(thf)(^DippPDI-H)Fe(N₂)] ([Na(thf)][567]), in which one methyl group from the PDI backbone is deprotonated (Scheme 68a). Increasing the relative amount of reductant led to mixtures of products. By reaction between 566 and either sodium or sodium hydride, a species equivalent to [Na(thf)][567] with different pairing interactions ([Na(OEt₂)₃][567]) was obtained along with the more reduced product [(^DippPDI)Fe(μ-N₂)Na{Na(thf)₂}] (568), in which the ligand remained neutral (Scheme 68a). Further increasing the amount of NaH to 2 equiv yielded another mixture of products, consisting of the neutral deprotonated compound [(^DippPDI-H)Fe(N₂)] (569), proposed to be the precursor of 567, and the dinuclear [{(^DippPDI)Fe(μ-N₂)}₂Na{Na(thf)₂}₂] (570, Scheme 68a). Complexes 567–570 were characterized by X-ray crystallography, IR spectroscopy, elemental analysis, and the determination of their magnetic moments in solution. More detailed spectroscopic characterization is required to further substantiate their structural and bonding characteristics.

Scheme 68. Reduction of (PDI)Fe-Dinitrogen Complexes by Gambarotta and Chirik^400,403.

Complexes 567–570 feature only slightly activated dinitrogen ligands according to the stretching vibrations (ν_N≡N = 1868–2159 cm^–1), with the most weakened N≡N motifs present in the complexes with the highest degree of reduction, i.e., 568 (1899 cm^–1) and 570 (1868 cm^–1). Complex 568 was regarded as featuring an iron center in the negative divalent state. In general, it was concluded that the reduction processes affected mainly the metal center, and only to some extent the ligand scaffold.

Indeed, the Chirik group found evidence of the reduction of the PDI scaffold at iron-dinitrogen complexes up to a trianionic state.⁴⁰⁰ To avoid ion pairing interactions like those reported by Gambarotta and Budzelaar,⁴⁰³ the reduction reactions were carried out in the presence of sequestering agents ([15]crown-5 or [18]crown-6). Reduction of neutral compound [(^DippPDI)Fe(N₂)] (571) and the phenyl-substituted analogue [(^DippBPDI)Fe(N₂)₂] (572, BPDI = 2,6-(2,6-iPr₂–C₆H₃N=CPh)₂C₅H₃N) by sodium naphthalenide afforded the salts [Na(chelate)(thf)₂][(^DippPDI)Fe(N₂)] (573, chelate = [15]crown-5, [18]crown-6) and [Na(thf)_n][(^DippBPDI)Fe(N₂)] (574, Scheme 68b). In both cases, the reduction events were ligand-based, and the electronic configuration of 573 and 574 was described as featuring intermediate spin ferrous centers, i.e., [(PDI^3–)Fe^IIN₂]⁻. The phenyl-substituted compound 574 underwent further reduction, yielding the dianionic complex [Na(thf)₂]₂[(^DippBPDI)Fe(N₂)] (575, Scheme 68b). In this case, reduction is a metal-based process. The ligand framework in 575 remains in the trianionic state, while the intermediate-spin ferrous center was reduced to a low-spin Fe(I). Thus, three ligand-based electron transfer events occur upon reduction in the series of complexes, before the metal atom engages in redox chemistry.

2.2.10.3. Bipyridine Ligands

Bipyridine ligands are well-known to be particularly susceptible to redox activity in metal complexes. In his 2006 review, Ellis comments on the exclusion of various anionic bipyridine complexes of the form [M(bpy)₃]ⁿ⁻ from the discussion, where metal centers had originally been assigned negative oxidation states.² These assessments were rendered erroneous by subsequent characterization that identified dianionic bipyridine ligands as responsible for the anionic character of the complexes.²⁶⁹ In more recent years, experimental and computational investigations into the electronic structures of anionic bipyridine complexes have been reported,^345,346 notably by Wieghardt and co-workers.^347−350 For instance, in 2015, the Wieghardt group used a combination of X-ray crystallographic data, UV–vis spectroscopy, magnetochemistry, and broken-symmetry density functional theory to study the electronic structures of all the redox states of the bis- and tris-substituted 2,2′-bipyridine-nickel complexes, including anionic species.³⁵⁰ [Ni(bpy)₂] (576) was also prepared according to modified literature methods and its crystal structure determined for comparison.^404,405 Cyclic voltammetry of the neutral complex [Ni(bpy)₂] (576) proved that its reduction to the monoanionic species [Ni(bpy)₂]⁻ (576^–), and even to the dianionic [Ni(bpy)₂]^2– (576^2–), was possible. The structures of the reduced anionic species were optimized using DFT. After analysis of the experimental and theoretical data, it was concluded that the reductions are ligand-centered, and therefore, the nickel atom in none of these complexes is in the Ni(0) oxidation state or lower. The anionic species [Ni(bpy)₂]⁻ features two identical π-radical anions and can be more accurately described as [Ni^I(bpy^•–)₂]⁻, i.e., a Ni(I) (d⁹) complex with three unpaired electrons which are antiferromagnetically coupled with the ligand radicals. As a result, an S = 1/2 ground state is observed. Likewise, the tris(bipyridine) complex [Ni(bpy)₃]⁻ (577^–) was formulated as [Ni^II(bpy^•–)₃]⁻.³⁵⁰

Like in the previous example, detailed analyses of the contributions of Wieghardt and co-workers generally led to the conclusion that reduction events tend to occur at the bipyridine ligands.^347−350 Thus, the complexes feature metal centers retaining common oxidation states (Fe^II, Zn^II, Cr^III, V^II, etc.).^347−349 A similar phenomenon had been observed within the series of “anionic” iron-substituted bpy complexes [Fe(^Rbpy)₃]⁻ (186, vide supra),^201−204 and with the cobalt compound [Co(bpy^•–)(η⁴-cod)]⁻ (261).⁸⁵ Other examples^347−349 will not be subject to discussion in this present review due to its focus on metalates featuring low-valent metal centers.

2.2.10.4. Diazadiene Ligands

Grützmacher and Lichtenberg have developed iron systems featuring diolefin-diazadiene ligands such as trop₂dad (trop = 5H-dibenzo-[a,d]cyclo-hepten-5-yl, dad = diazadiene) and its saturated analogue trop₂dae (dae = N–CH₂–CH₂–N; Scheme 69).^379−381 Reaction of the trop₂dad ligand with a suitable Fe(II) precursor, [FeBr₂(thf)₂] (Scheme 69a), affords the mononuclear Fe(II) compound [FeBr₂(trop₂dad)] (578), which can be reduced by sodium or sodium hydride to yield the ferrate(I) complex [Na(thf)₃{Fe(trop₂dad)}] (579).³⁷⁹ Likewise, attempts to generate metal alkyl complexes by reaction of 578 with LiCH₂SiMe₃ resulted in nucleophilic attack on the imine moieties with reduction of the metal center, yielding the low spin Fe(I) complex 580.³⁸⁰

Scheme 69. Synthesis of Low-Valent (a) trop₂dad- and (b) trop₂dae-Iron Complexes by the Group of Grützmacher and Lichtenberg^379−381.

The crystallographic characterization of 579 indicates that the ligand is present as a diazadiene dianion (trop₂dad)^2–, featuring a short C–C bond distance (1.368(3) Å) and two elongated C–N bonds (N–C 1.384(3) and 1.375(3) Å),⁴⁰⁶ and shows the formation of a contact ion pair with the cation Na(thf)₃⁺. Compound 579 was used as a building block, in a 2:1 ratio with [FeCl₂(thf)_1.5], to form the trinuclear iron cluster [Fe₃(trop₂dad)₂] (581, Scheme 69a). X-ray diffraction analysis performed on single crystals of 581 showed that the three iron centers are in an almost perfect linear arrangement through metal–metal bonds (Fe–Fe 2.6340(8)–2.6410(8) Å). ⁵⁷Fe Mössbauer spectroscopy and magnetic susceptibility measurements along with the X-ray crystallographic data suggest that the terminal Fe atoms are in a low spin Fe(I) state, while the central metal atom of the Fe₃ chain appears to be in a high-spin Fe(II) state.³⁷⁹ Oxidation of 579 with ferrocenium hexafluorophosphate (FcPF₆; Scheme 69a, bottom) in the presence of a donor ligand L afforded the five-coordinate species [Fe(trop₂dad)(L)] [L = thf (582), MeCN (583), PPh₃ (584), P(OMe)₃ (585)]. Conversely, reaction of compounds 582–585 with the mild reducing agents NaBH₄ or NaOtBu regenerates compound 579, illustrating the chemical reversibility of the oxidation reaction (Scheme 69a).³⁸⁰ By comparing the X-ray structural data of 582–585 with that of 578, bearing a neutral trop₂dad ligand, and that of the parent compound 579, featuring a dianionic (fully reduced) (trop₂dad)^2– moiety,³⁷⁹ it was concluded that all of the complexes 582–585 contain monoanionic radical (trop₂dad)^•– ligands. The electronic structures of these compounds are thus best described as open shell singlets with the metal center in a low spin Fe(I) oxidation state which is antiferromagnetically coupled to the monoanionic ligand radical. This is corroborated by DFT calculations, ⁵⁷Fe Mössbauer spectroscopy, and magnetic susceptibility measurements. Therefore, the oxidation of 579 is a ligand-centered process, where one electron is removed from the redox-active ligand.

The chemically noninnocent ligand trop₂dae is the hydrogenation product of trop₂dad.⁴⁰⁷ Coordination studies show that trop₂dae is also able to stabilize low-valent Fe complexes. Ferrates [M(solv)_n{Fe(trop₂dae)}] [M = Li, solv = Et₂O, n = 2 (586); M = Na, solv = thf, n = 3 (587); Scheme 69b] were synthesized from [FeCl₂(thf)_n] and trop₂dae in the presence of MCH₂SiMe₃, which acts both as a base and as a reductant. Subtle structural differences were observed in the molecular structures of the lithium and the sodium salt. Both complexes present a distorted planar coordination geometry about the iron center; whereas in 587, the sodium cation is asymmetrically coordinated to the trop₂dae unit, in 586, the lithium cation coordinates both nitrogen atoms. ⁵⁷Fe Mössbauer spectroscopy revealed a higher electron density in the Fe s orbitals for 587, which is in line with Li⁺ acting as a stronger Lewis acid.

A series of (α-diimine)(polyarene)cobalt complexes with the ligand N,N′-bis(2,6-diisopropylphenyl)butane-2,3-diimine (^Dipp2dmdad) was described by Yang and co-workers in 2015.³⁸⁷ This series included neutral complexes [(^Dipp2dmdad)Co(η⁴-anthracene)] (588), [(^Dipp2dmdad)Co(μ–η⁴:η⁴-naphthalene)Co(^Dipp2dmdad)] (589), and [(^Dipp2dmdad)Co(μ–η⁴:η⁴-phenanthrene)Co(^Dipp2dmdad)] (590) and the anionic species [{Na₂(Et₂O)₂}{(^Dipp2dmdad)Co(μ–η³:η³-pyrene)Co(^Dipp2dmdad)}] (591) and [{Na₂(Et₂O)₃}{(^Dipp2dmdad)Co(η³-pyrene)}] (592). All of those compounds have three potential redox-active sites: the metal center, the α-diimine ligand, and the polyarene moiety.

The dinuclear complexes 589 and 590 bear some analogy to anthracene- and naphthalene-bridged bis(cyclopentadienyliron) compounds previously described by the groups of Wolf and Tatsumi (see Scheme 27).^221,222 The salts 591 and 592 were obtained by treating the dimeric precursor [(^Dipp2dmdadCo)₂] (593) first with pyrene (in 1:1 or 1:2 stoichiometric ratio, see Scheme 70) and subsequently with metallic sodium. X-ray diffraction studies, EPR spectroscopy, magnetic susceptibility measurements, and DFT calculations indicate the presence of a closed-shell dianionic (^Dipp2dmdad)^2– ligand. Therefore, the sodium-mediated reduction is ligand (α-diimine, polyarene, or both)-based. In 592, the pyrene ligand has also undergone reduction to some extent, as evidenced by the crystallographically determined C–C bond lengths and DFT calculations. Consequently, this mononuclear species features an antiferromagnetically coupled monoradical form of the pyrene ligand. Thus, the experimental data indicate that the cobalt center is in the formal Co(I) oxidation state, with a neutral pyrene ligand in 591 and a monoanionic one in 592.³⁸⁷

Scheme 70. Heteroleptic (α-Diimine)(polyarene)cobalt Complexes Described by Yang and Co-workers³⁸⁷.

The Grützmacher group has described 4d or 5d metalates featuring diamine and diazadiene ligands.^407−415 Grützmacher and Trincado demonstrated the versatility of several Ru-diazadiene-olefin complexes in hydrogenation reactions of organic substrates, as well as dehydrogenation reactions of light alcohols or formic acid.^407−415 For these Ru compounds, the ligand architecture plays a crucial role during the hydrogenative or dehydrogenative transformations in which they are involved, and it is supposed to be both redox and chemically non-innocent. The coordination environment is well-suited to stabilize the Ru center in low oxidation states, including anionic species (Scheme 71). An early example of this chemistry is the anionic hydride complex [K(dme)₂][RuH(trop₂dad)] (594, Scheme 71a), which can be formally described as any of the three resonance forms [Ru²⁺(trop₂dad^2–)], [Ru⁺(trop₂dad^·–)], or [Ru⁰(trop₂dad)] due to the noninnocence of the diazadiene ligand scaffold (the hydride ligand and potassium atom have been omitted for clarity).⁴⁰⁷ Reaction of 594 with water generated the neutral diimino compound [Ru(trop₂dad)] (595), with elimination of KOH and H₂.

Scheme 71. Ruthenium Complexes Featuring (a) the trop Family of Diazadiene Ligands and (b) the Diazadiene-dct Ligands^407−415.

Although the form [Ru⁰H(trop₂dad)]⁻ was initially considered as an adequate description of 594, it was later proved, by means of multinuclear NMR spectroscopy, X-ray crystallographic data, and computational studies that the structure of this complex is consistent with a Ru²⁺ ion bearing a 1,2-enediamide chelate, i.e., the resonance form [Ru²⁺(trop₂dad^2–)] has a greater contribution to the ground-state of 594.^408−410 Changes in the ligand backbone were shown to dramatically influence both the reactivity in the hydrogenation/dehydrogenation reactions and the formal description of the electronic nature of complexes themselves. For instance, if the 1,2-diaminoethene unit that connects the trop groups in trop₂dad is replaced by the saturated 1,2-cyclohexane bridge, trop₂dach, a four-coordinate ruthenate complex can be obtained under reaction conditions like those used for 594. The species [K([18]crown-6)(thf)][Ru(trop₂dach-H)] (596, Scheme 71a) was characterized as a zerovalent ruthenium center bearing a singly deprotonated trop₂dach ligand. In solution, this amido-amino-ruthenium complex undergoes intramolecular metal–ligand N–H addition/elimination leading to a transient diamido-ruthenium-hydride species [RuH(trop₂dach-2H)]⁻ (597), as supported by NMR and DFT analysis.⁴¹¹

This phenomenon might be of mechanistic relevance in the methanol/water dehydrogenation mediated by 594. If the backbone length in trop₂dad is increased by one =CH– fragment, forming a β-diketiminate-type chelate trop₂ipa (ipa = iminopropenamide), the four-coordinate salt K[Ru(trop₂ipa)] (598) can be isolated under reaction conditions similar to those for 594. In 598, a formally monoanionic β-diketiminate ligand is coordinated to a Ru⁰ center, which readily reacts with water to yield the Ru(II)-hydride complex 599 with elimination of KOH (Scheme 71a). Ruthenate 598 and its parental zerovalent complex [Ru(trop₂dap)] (600, dap = 1,3-diaminopropane), bearing a fully saturated ligand backbone, were used as catalysts in dehydrogenation and hydrogenation reactions of several organic substrates. Ruthenate 598 promotes the dehydrogenative coupling (DHC) reaction between benzyl alcohol and water in basic media to afford the corresponding carboxylic acid salt with release of molecular hydrogen.⁴¹² In connection with the conversion of aqueous basic methanol or formaldehyde solutions into H₂ and carbonate catalyzed by 595,⁴¹³ the Grützmacher group developed a modular version of such a zerovalent complex by separating the two dad and olefin motifs while maintaining the same coordination environment around the metal center. With this idea, trop₂dad was conceptualized as a Me₂dad fragment and a dct scaffold (Me₂dad = 1,4-dimethyl-diazabuta-1,3-diene; dct = dibenzo[a,e]cyclooctatetraene). Under reaction conditions like those used to obtain 594, the modular diazadiene-olefin approach led to the tetrameric anionic complex K[Ru₄(μ-H)(Me₂dad)₂(dct)₄] (601, Scheme 71b) instead of the hypothetical K[RuH(Me₂dad)(dct)].⁴¹⁴ The tetramer can be reductively cleaved to the dianionic dimer K₂[Ru₂(Me₂dad)(dct)₂] (602) or reductively protonated and then deprotonated to the anionic bridging hydride K[Ru₂(μ-H)(Me₂dad)(dct)₂] (603, see Scheme 71b). The anionic dimers 602 and 603 and the tetramer 601 are in principle related by proton and electron transfer processes, mimicking the behavior of hydrogenases. In fact, the neutral dihydride [Ru₂H(μ-H)(Me₂dad)(dct)₂] (604), regarded as the thermodynamic sink within the [Ru₂(Me₂dad)(dct)₂] family, converts H₂ to protons and electrons (Rauchfuss test for hydrogenase activity), and catalyzes the selective hydrogenation of vitamins K₂ and K₃ to their corresponding hydroquinones without affecting the C=C double bonds. Moreover, 604 catalyzes the reduction of nitrous oxide (N₂O) with light alcohols as hydrogen source, forming molecular nitrogen and carboxylates.⁴¹⁵ The combination of the noninnocent Me₂dad ligand with the four electron π-donor and π*-acceptor dct scaffold has been essential to stabilize low-valent ruthenium centers and to achieve the observed hydrogenase- or alcohol dehydrogenase-like catalytic activity.

2.3. Stabilization of Low-Valent Metalates by Ion-Pairing and Element–Element Bonding

2.3.1. Ion-Pairing in Low-Oxidation State Organyl Complexes

The countercation is an essential feature of every metalate compound. In many cases, the cation is an alkali metal such as Li⁺, Na⁺, or K⁺, which is sequestered by crown ether or cryptand molecules to prevent ion-pairing and facilitate crystallization of the resulting ion-separated compound. Ion-pairing interactions can occur in the absence of a cation sequestering agent. While studies by Fürstner,²¹⁴ Hevia,^416−418 Mountford,^{137,138,419,420} Crimmin,^421−424 Wolf,⁴²⁵ Peters,^{139−141,426−430} and Lu,^78,431−435 among other contributions, illustrate that ion-pairing is a frequent phenomenon, only a handful of the investigations explicitly address the influence of such interactions on their reactivity. However, from these investigations it has become clear that the countercation may have profound influence on the reaction patterns, stoichiometric or catalytic, of metalates. Indeed, modification of the countercations might provide an opportunity for developing finely tuned systems.⁴³⁶ Evidence of counterions assisting in catalytic reactions has already been reported,^437−439 and whenever the anion/cation pair is metal based, these systems can be formally considered as (hetero)bimetallic.^440,441

This section collects the most relevant examples demonstrating the importance of ion-pairing effects in transition metalate chemistry. However, it should be noted that the role of the countercation is always worth considering. As such, further examples of countercation effects can be found in the preceding section 2.2.

While studying iron-catalyzed cross-coupling reactions, Fürstner and co-workers found that iron salts FeX_n [X = Cl, acac (acac = acetylacetonate); n = 2, 3] are reduced by Grignard reagents bearing alkyl groups susceptible to β-hydride elimination to afford clusters of the formal composition [Fe(MgX)₂]_n (605).^{214,442−446} The authors proposed that, in the reduction leading to the heterometallic species 605, the resulting iron center is in a formally −II oxidation state, reaching a d¹⁰ electron configuration. According to the authors, this type of compound might be sufficiently nucleophilic to promote cross-coupling reactions. Furthermore, the group also reported that the catalytic behavior of clusters 605 can be mimicked by using the known lithium ferrates [Li₂(TMEDA)₂{Fe(η²-C₂H₄)₄}] (15), [Li(dme)]₂[Fe(cod)₂] ([Li(dme)]₂[16]),⁸ [Li(tmeda){CpFe(C₂H₄)₂}] (193), [Li(dme){CpFe(cod)}] (194), or [Li(tmeda){CpFe(C₂H₄)₂}] (196, see Scheme 23, above), with the Fe(−II) ferrate 15 being the most efficient precursor.^{212−214,447} Based on several mechanistic studies, it was proposed that both for the in situ generated clusters 605 and the aforementioned electron-rich lithium ferrates 15, [Li(dme)]₂[16], and 193–196, a structural resemblance might be anticipated: the short intermetallic Fe–Li interactions influence the reactivity patterns of the iron species involved in the studied catalytic cross-coupling reactions.²¹⁴

The effect of the intermetallic bonds was demonstrated by investigating ligand exchange reactions of the coordinated olefins (ethylene or 1,5-cod) in ferrates 193 and 194 by, for instance, 1,3-butadiene.⁴⁴⁸ The substitution of the olefin ligands proved difficult for the studied lithium ferrates 193 and 194, even under forcing conditions. Therefore, to facilitate the ligand exchange, the authors performed a transmetalation reaction of the lithium cation on these compounds by the less electropositive zinc cation (see Scheme 72). It was proposed that a reduction in the ionic character of the interaction would cause a decrease in the metal-to-ligand back-donation (ferrate → coordinated olefin), thus reducing their bond strength, and facilitating ligand exchange. The heterotrimetallic compound Zn[CpFe(C₂H₄)₂]₂ (606) was formed by the reaction of 193 with ZnCl₂. Its solid-state structure confirms the transmetalation, and the intermetallic contacts between the metal centers. Ligand-exchange of ethylene by an excess of 1,3-butadiene was achieved with 606, although a reaction temperature of 60 °C to facilitate complete substitution was required to obtain the butadiene complex Zn[CpFe(1,3-C₄H₆)]₂ (607). Treatment of the butadiene complex with lithium in the presence of TMEDA afforded the desired, and otherwise inaccessible, lithium(butadiene)ferrate, [Li(tmeda){CpFe(1,3-butadiene)}] (608) with concomitant deposition of metallic zinc. Similar treatment of 194 afforded the analogous compound Zn[CpFe(cod)]₂ (609). The authors anticipated that the heterometallic complexes might be of interest as precatalysts in a variety of transformations, through reactivity of the intermetallic contacts.⁴⁴⁸

Scheme 72. Facilitating Ligand-Exchange of Ethylene by 1,3-Butadiene by Transmetalation to Form Zinc-Ferrates⁴⁴⁸.

Nonetheless, despite various complexes having been identified and proposed as reaction intermediates, the exact role that the highly reduced iron species play in catalysis is not entirely clear. Other researchers have, therefore, revisited the evidence^436,449,450 and contributed to this area.^447,451,452 For instance, in 2011, Wolf and co-workers studied the catalytic cross-coupling of alkyl electrophiles with aryl Grignard reagents mediated by a series of metalate and other low-valent iron complexes.⁴⁴⁷ The studied catalysts included the Fe(−II) compound [Li₂(TMEDA)₂{Fe(η²-C₂H₄)₄}] (15), the “iron Grignard” compound [Cp(dppe)FeMgBr(thf)₂] (610, dppe = 1,2-bis(diphenylphosphino)ethane)⁴⁵³ and other organoiron complexes with formal oxidation states ranging from Fe(−I) to Fe(III). While compound 15 remained the most active precursor, the labile (anthracene)ferrates(−I) [Fe(η⁴-C₁₄H₁₀)₂]⁻ (52) and [Fe(η⁴-C₁₄H₁₀)(η⁴-cod)]⁻ (183) were similarly competent, and the “iron Grignard reagent” 610 hardly showed any catalytic activity under the studied conditions. Therefore, the presence of metal–metal bonds did not enhance the catalytic behavior in this case, and the authors remained doubtful about whether the putative intermetallic iron Grignard compounds were the true catalysts in iron-catalyzed cross-couplings. Furthermore, it was concluded that, whereas the oxidation state of the iron center had little influence on the catalytic performance, a labile coordination environment was crucial for high catalytic activity.⁴⁴⁷

In 2015, Bedford reviewed the available mechanistic data for iron catalyzed cross-coupling reactions and concluded that scant evidence exists to support the participation of highly reduced (subzero-valent) iron species in the catalytic cycle.⁴⁴⁹ It was argued that while Fe(−I) and Fe(−II) complexes, such as 15, can function as precatalysts, this has no bearing on the oxidation state of the active species.⁴⁴⁹ Bedford and co-workers instead proposed that the species accounting for the catalytic activity in iron-catalyzed cross-coupling reactions are homoleptic three-coordinate Fe(II)-“ate” complexes, of the type [FeR₃]⁻ [R = mesityl³⁴⁵ (611, as the [Mg₂Br₃(thf)₆]⁺ salt) benzyl (612, as the [Mg₂Cl₂(OTf)(thf)₃]⁺ salt); see Scheme 73]. Complexes 611 and 612 are observed under catalytically relevant conditions, and obtained by reaction of FeCl_n (n = 2, 3) with Grignard reagents (in the presence of an excess of tetramethylethylene-1,2-diamine, TMEDA, for complex 611). Although neutral [FeR₂(TMEDA)] [R = Mes (613), Bn (614)] complexes were also observed in these investigations, the authors demonstrated that the anionic 611 and 612 complexes undergo cross-coupling with electrophiles significantly faster than 613/614. For coupling reactions of smaller aryl Grignard reagents, the formation of zerovalent nanoparticles was observed, whereas EPR spectra of the catalytic reaction mixture suggest the presence of low-valent Fe(I) species. Consequently, the formation of catalytically relevant species with oxidation states below Fe(II) cannot be completely excluded for reactions with smaller R substituents.⁴⁵¹ Neidig and co-workers identified a compound related to 611 and 612 under catalytically relevant conditions, the Fe(II)-“ate” compound [Mg(NMP)₆][FeMe₃]₂ (615, NMP = N-methylpyrrolidone), as the major species in a system catalyzed by Fe(acac)₃ (298) and MeMgBr, using NMP as a cosolvent. 615 is highly active and selective in the production of cross-coupled products.⁴⁵² Furthermore, this group also isolated and characterized the small anionic cluster [MgCl(thf)₅][Fe₈Me₁₂] (616), obtained from FeCl₃ and MeMgBr in THF.⁴⁵⁴ This compound is only slightly active in cross-coupling reactions by direct reaction with the electrophile, but in the presence of additional MeMgBr promotes the coupling reaction rapidly and selectively.

Scheme 73. σ-Organyl Ferrates Relevant in Cross-Coupling Reactions Catalyzed by Iron Precursors, Reported by Bedford and Co-workers⁴⁵¹.

In this context, Koszinowski and co-workers also investigated the chemistry of organoferrate complexes,^455−457 demonstrating that cluster 616 can also be accessed by treating Fe(acac)₃ (298) with MeMgX (4 equiv; X = Cl, Br) in THF solution, and studied its reactivity through electrospray ionization (ESI)–mass spectrometry. The authors reported an average oxidation state of 1.4 for the iron centers in 616. In the presence of TMEDA, the signal for 616 practically disappears. Similar heteroleptic organoferrate clusters Me_12–nFe₈Ar_n^– (n = 1–5, Ar = Ph, Tol), with identical average oxidation states to 616, were prepared. The spectrometric identification of 616 corroborates the previous results by Neidig and co-workers⁴⁵² and further supports the possible involvement of this type of cluster ion in iron-catalyzed cross-coupling reactions.⁴⁵⁶

Further ESI–mass spectrometry investigations in solution, using a variety of iron salts (Fe(acac)₃, FeCl₃, FeCl₂, Fe(OAc)₂) and Grignard reagents RMgX (R = Me, Et, Bu, Hex, Oct, Dec, Me₃SiCH₂, Bn, Ph, Mes, 3,5-(CF₃)₂-C₆H₃; X = Cl, Br), led to the spectrometric identification of different mono- or polynuclear organoferrates with metal centers mainly in the +II or +III oxidation states.⁴⁵⁷ For R = Ph, only small amounts of cluster [Ph₇Fe₄]⁻ (617), featuring iron centers in an average oxidation state of 1.5, were identified.⁴⁵⁵ In the presence of TMEDA, the alkylferrates displayed enhanced stability, and it was possible to identify the transient low-valent complex [Bu₂Fe^I]⁻ (618), although most of the species still corresponded to organoferrates in the+II or +III oxidation states. In view of the observation of species 618, the authors proposed that the formation of the Fe(III) alkylferrates might involve an oxidative addition to the low-valent Fe(I) intermediate.

Low-valent organoferrates were also observed upon treating [FeCl₂(dppbz)₂] (619, dppbz = 1,2-bis(diphenylphosphino)benzene) with EtMgCl. In this case, minor amounts of Fe(−I) and Fe(0) ferrates were identified.⁴⁵⁷ In turn, treatment of 619 with PhMgCl (4 equiv) allowed identification of species in the −I, 0, and +I oxidation states, including dinuclear [Ph₃Fe₂(dppbz)]⁻ (620) and mononuclear [Fe(dppbz)]⁻ (621), [PhFe(dppbz)]⁻ (622), and [Ph₂Fe(dppbz)]⁻ (623). The dinuclear species [Ph₂Fe₂(dppbz)]⁻ (624), having an average oxidation state of 0.5, was also observed.⁴⁵⁵ The observation of the various low-oxidation Fe complexes was attributed to the additional stabilization provided by the bidentate phosphine ligand dppbz, which lowers the electron density of the ferrate species through π-backbonding and coordinative saturation of the iron center. Since most of the identified Fe(III) species underwent reductive elimination processes, the authors proposed that these could be considered as likely intermediates in iron-catalyzed cross-coupling reactions.⁴⁵⁷

Hevia and co-workers have reported several investigations on the chemistry of organonickelate complexes and their implications in catalytic cross-coupling reactions.^{416−418,458,459} Monitoring of an NMR scale reaction between equimolar amounts of [Ni(cod)₂] (240) and PhLi revealed only half of the nickel complex reacted. This suggested the formation of a 1:2 Ni:PhLi species, subsequently identified as [Li₂(thf)₄Ph₂Ni(cod)] (625). By changing the concentration of the reaction, it was possible to identify a minor species with 1:1 ratio of Ni and PhLi [Li(thf)₂PhNi(cod)] (626). This ion-paired species is in equilibrium with an ion-separated one, which in turn dissociates one or both the coordinated cod olefin units to afford compound 625 and partially regenerates complex 240. Preparative reactions between 240 and PhLi resulted in three lithium (cod)nickelates, which were preferentially isolated depending on the ratio between the reactants and the crystallization conditions.⁴¹⁶ Treatment of 240 with PhLi in a 1:2 ratio in the presence of a donor afforded the corresponding lithium nickelates [Li₂(donor)_xPh₂Ni(cod)] [625, donor: THF (x = 4), TMEDA (x = 2), PMDETA (x = 2); Scheme 74, left, shows only the THF derivative], as previously observed on an NMR scale. The complexes are extremely air- and moisture-sensitive. Characterization through X-ray diffraction showed a trigonal planar environment for the Ni center. The C=C bond distances reflect the different coordination environment of the cyclooctadiene ligand: the coordinated C=C bond is significantly elongated (d_CC = 1.446(2)–1.452(2) Å) with respect to the uncoordinated C=C bond (d_CC = 1.321(5)–1.327(2) Å). The elongation of the coordinated C=C bond is attributed to the higher electron density on the Ni(0) center and, therefore, increase in backbonding to the C=C moiety. Compound 625 suffers a slow and reversible loss of half an equivalent of COD in solution to yield the bridged hexanuclear complexes [Li₄(donor)_4xPh₄Ni₂(cod)] [627, donor: THF (x = 2), TMEDA (x = 1); Scheme 74, top left, shows only the THF adduct]. The shift toward nickelate 625 in the presence of excess COD is accompanied by formation of the nickel precursor 240. Also, increasing the Ni:PhLi ratio (mimicking catalytic cross-coupling conditions) led to the isolation of a bridged octanuclear 3:1 lithium nickelate from the reaction mixture, namely [Li₆(thf)₈Ph₆Ni₂(cod)] (628, Scheme 74, bottom left). NMR monitoring of such reactions revealed, however, that the trinuclear compound 625 was the main component of the reaction mixture.

Scheme 74. Lithium (cod)Nickelates and Polyheterometallic Organonickelates by Hevia and Co-workers^416,417.

Further investigations helped to shed some light on the role that the lithium nickelates play on the Ni-catalyzed cross-coupling of aryl ethers. Catalytic, stoichiometric, and kinetic experiments of aryl ether cross-coupling reactions were performed. In a model reaction of cross-coupling of PhLi and 2-methoxynaphthalene, catalyzed by [Ni(cod)₂] (240, 5 mol %), solvent and donor effects were observed for the series of lithium nickelates. For instance, when THF was the solvent of choice, the reaction returned only products of ortho-lithiation of the aryl ether, instead of the cross-coupled compound.⁴¹⁶

Likewise, by using a solvated PhLi aggregate, [PhLi(donor)]_n, in combination with 240, the immediate formation of the trinuclear lithium nickelate 625 was observed.⁴¹⁶ Upon addition of the substrate 2-methoxynaphthalene to the mixture, it was possible to identify the cross-coupling and homocoupling products. Furthermore, the hexanuclear and octanuclear complexes 627 and 628 proved competent to promote the cross-coupling reactions at equal catalyst loadings. These observations, along with additional spectroscopic and experimental evidence, led to the conclusion that heterometallic Ni(0)-“ate” complexes play an important and cooperative role in the activation of the aryl ether substrate and, therefore, in the catalytic cycle. Both the nucleophilic Ni center and the Lewis acidic lithium cations seem to be involved in such activation of the substrate. Two mechanistic proposals were asserted based on the combined data. Starting from complex 240, the addition of [PhLi(donor)]_n can afford any of the observed 1:1 (626) or 1:2 nickelate species 625. The latter seems to be favored, however, since the 1:1 nickelate was not routinely observed. Nonetheless, knowing that these two nickelates are related by equilibrium in the presence of [PhLi(donor)]_n, both were considered to be involved in the mechanism. Therefore, the authors proposed both an anionic and a dianionic reaction pathway. In the anionic pathway, complex 240 and [PhLi(donor)]_n form compound 626, which in turn coordinates the substrate, oxidatively adds the aryl–OMe bond with elimination of LiOMe, followed by reductive elimination of the cross-coupling product and regeneration of complex 240. In the dianionic pathway, nickelate species 625 forms rapidly in the early stages of the reaction, and directly coordinates the substrate to engage in the additional steps of the catalytic reaction.

In further investigations,⁴¹⁷ Campos and Hevia revisited the structure of the homoleptic lithium nickelate “Li₃NiPh₃(solv)₃”, reported by Taube and co-workers over four decades ago.^460−462 Based exclusively on its ¹³C NMR spectroscopic characterization, this species was thought to adopt an extremely rare hexagonal planar coordination environment around the Ni atom which would be composed of three Li⁺ cations and three phenyl ligands. The two initial synthetic routes proposed by Taube were reexamined. One of these involves the displacement of a COD ligand in [Ni(cod)₂] (240) by PhLi, which, as described above, affords three different lithium nickelates 625–628, depending on the stoichiometric ratio between the reactants and on the crystallization conditions.⁴¹⁶ Similarly, treating precursor 240 with [(PhLi·Et₂O)₃LiBr] (prepared from PhBr and tBuLi)⁴⁶³ afforded [Li₂(Et₂O)₄(LiBr)Ph₂Ni]₂(cod) (629, Scheme 74, top right), which contains cocomplexed LiBr.

None of these experiments suggested the presence of a planar complex. While treatment of the Ni(0) precursor with donor-free PhLi (5 equiv) led to the formation of a deep red solution, as previously described by Taube, NMR spectroscopic characterization of this solution did not match the previous reports.^460−462 The choice of solvent influences the type of product obtained. If the reaction is carried out in THF, the major product will correspond to the nickelate 627. By using Et₂O (Scheme 74, middle right), it was possible to isolate a crystalline solid in low yields, which corresponded to [{Li₃(Et₂O)₂Ph₃Ni}₂(μ,η²:η²-C₆H₄)] (630), containing a benzyne ligand bridging the Ni centers. Additionally, treating 240 with 3 equiv of the THF adduct [PhLi(thf)]₄ in toluene afforded the related bridging benzyne complex [{Li₃(thf)₂Ph₃Ni}(μ,η²:η²-C₆H₄){Li₂(thf)₃Ph₂Ni}] (631, Scheme 74, bottom right). 629–631 were characterized via X-ray diffraction and NMR spectroscopy. DFT calculations were also carried out on the electronic structure of 630, indicating that this is best described (lowest-energy state) as a closed-shell singlet, which agrees well with the crystal structure and the diamagnetic behavior observed. Additionally, the calculations confirmed that there is no metal–metal bond, which explains the long Ni–Ni distance (d_NiNi = 2.7117(8) Å) observed in the crystal structure. The phenyl carbanions in 630 act as strong σ-donors, increasing the electron density of the metal centers. NBO analysis suggested that the backdonation from a Ni d orbital to a π* orbital of the bridging C₆H₄ unit is, therefore, particularly strong. The latter corresponds then to an in situ generated, π-accepting and formally reduced C₆H₄^2– motif, which alleviates the extremely high electron density at the nickel centers. Altogether, the results from these investigations confirm that the initially proposed species “Li₃NiPh₃(solv)₃” is in fact consistent with the dinickel structure 630.

Attempts to replicate the observed chemistry with PhNa led to different results, suggesting a countercation effect.⁴¹⁷ Depending on the Ni:PhNa ratio, either the sodium nickelate [Na₂(solv)₃Ph₂Ni(cod)]₂ (632, 2:1 ratio) or [Na₃(solv)₃Ph₂(C₈H₁₁)Ni(cod)]₂ (633, 3–5 equiv of NaPh) was obtained (Figure 12). Although the sodium nickelates do not feature the benzyne dianion ligand observed in 630, the authors suggested that a lithium-containing analogue of 632 might be involved in the formation of 630. Complexes 627 and 630 were evaluated as catalyst precursors in Csp²–Csp³ Kumada coupling, Buchwald–Hartwig coupling, and cross-coupling of aryl ethers (for 630).⁴¹⁷ For the Kumada coupling, the nickelates performed better than precursor 240, which might indicate that similar Ni-“ate” species might be involved in this type of catalytic transformation.

Figure 12.

Sodium organonickelates by the group of Campos, Hevia, and co-workers.⁴¹⁷

Later, Hevia and co-workers reported the formation of planar Ni(0) complexes supported by organolithium ligands.⁴¹⁸ Reactions of [Ni(cod)₂] (240) with lithium aryl-acetylides in the presence of TMEDA (slight excess) yielded a family of homoleptic lithium nickelates of general formula [Li₃(TMEDA)₃Ni(C≡C–Ar)₃] [Ar = Ph (634), p-Tol (635), 3-thienyl (636); see Scheme 75). As with the lithium nickelates illustrated in Scheme 74,^416,417 the solvent choice is critical for the synthesis of the complexes. In the absence of additional donors, neither THF nor Et₂O allowed for the spectroscopic characterization or isolation of the nickelates, and the presence of a sequestering agent (crown ether) led only to decomposition.

Scheme 75. Synthesis of Tri-lithium Nickelates Derived from Aryl-acetylides⁴¹⁸.

Stabilization of the cation by solvation with TMEDA as a donor was, therefore, deemed crucial, allowing for the reaction to be carried out in Et₂O. The nature of the acetylide ligand also influenced the type of compound formed as, for instance, the use of the more electron-rich Me₃Si–C≡CLi led to the formation of the polynuclear cluster [Li₆(TMEDA)_3.5Ni₂(C≡C-SiMe₃)₆]₂ (637, Scheme 75, bottom) instead of an analogue of 634–636. X-ray diffraction analysis of 634 (Ar = Ph) suggested that the nickel center in the complex is located in a planar coordination environment, surrounded by three lithium cations and three acetylide ligands in an alternating fashion. The C–Ni–Li angles (59.9°, on average) indicate a hexagonal planar geometry around nickel, whereas the Ni–Li distances (average d_NiLi = 2.50 Å) and the Li–C distances (average d_NiLi = 2.37 Å), along with a Hirshfeld surface analysis, show close Ni–Li and Li–C contacts.

A theoretical analysis of the bonding situation (Quantum Theory of Atoms in Molecules, QTAIM) indicated that bond critical points exist for all Ni–C bonds, but not for the Ni–Li interactions, indicating the absence of covalent Ni–Li bonds.⁴¹⁸ In fact, the noncovalent interaction index (NCI) revealed that the latter are repulsive in nature. By contrast, the Li–C contacts are attractive, and several H···C bond contacts between the methyl groups of the TMEDA donor and carbon atoms of the acetylide ligands, were identified. These were assigned as London dispersion (van der Waals) interactions, known to assist in the stabilization of compounds and to ease in their isolation.⁴⁶⁴ Such observations would explain the difficulties in isolating the nickelates by using THF or Et₂O as solvents in the absence of TMEDA.

An analysis of the charge distribution in the complex also suggested that the acetylide ligands are negatively charged and that the overall charge transfer occurs from the Ni atom toward the ligands due to substantial metal-to-ligand backbonding. The transition metal would then formally be a neutral Ni(0) center. The lithium cations are then electrostatically attracted to the acetylide units. Altogether, the bonding analysis indicated that, instead of hexagonal planar organonickel complexes, the geometry of 634 corresponds to trigonal planar Ni(0) species stabilized by dispersion interactions between the Li-TMEDA motifs and the acetylide ligands. Compound 634 failed to react with phosphine ligands (PCy₃ or PEt₃) but reacted with iodobenzene (1 equiv) to afford the hexanuclear cluster 638 (Scheme 75, top), diphenylacetylene, and lithium iodide. The molecular structure of 638 features a side-on coordinated diphenylacetylene ligand, with the metal centers adopting a pseudo-trigonal planar environment composed of the bridging alkyne ligand and two terminal phenylacetylide ligands. Attempts to react 638 with additional iodobenzene, or with PhC≡CLi to regenerate 634, were unsuccessful. The authors concluded that, since 634 cannot be regenerated, the system is not suitable to perform catalytic Sonogashira-type cross-coupling reactions.

2.3.2. Transition-Metal Magnesium Compounds and Related Heavier Alkaline Earth Metal Complexes

In comparison with the wealth of alkali metal (lithium, sodium, and potassium) metalates, alkaline earth (AE) metal salts of low-oxidation transition metalate anions, are relatively scarce. These heterobimetallic species can be formed either by interaction between reduced transition metal precursors and suitable magnesium species or from TM salts in common oxidation states and Grignard reagents or Mg(I) compounds, taking advantage of their reducing character.^465−467 As previously mentioned, a recent review by Xu and co-workers comprehensively discusses Mg- or Zn-heterometallic complexes.²⁴ Considering this, we will herein highlight the contributions including the chemistry of low-valent transition metals, also known as magnesium metalates, and related heavier analogues.

A handful of magnesium-transition-metal compounds (sometimes termed ‘inorganic Grignard reagents’) were reported by the groups of Felkin, Green, and Jonas prior to the period covered by this review.^453,468,469 The first examples of these, [Cp(dppe)FeMgBr(thf)₂] (610) and [Cp₂MoHMg(C₆H₁₁)(μ-Br)₂Mg(Et₂O)]₂ (639), were synthesized almost 50 years ago (Figure 13).^453,468 The magnesium cobaltates [CpCo(C₂H₄)PhMgBr(tmeda)] (640) and [CpCo(η³-C₃H₅)MgBr(thf)₂] (641) were described by Jonas and co-workers in 1986 (Figure 13).⁴⁶⁹

Figure 13.

‘Inorganic Grignard reagents’ by Felkin, Green, and Jonas.^453,468,469

Similar to these early examples, Jones and co-workers reported a two-coordinate complex featuring an unsupported manganese–magnesium bond and reactivity of an ‘inorganic Grignard reagent’.⁴⁷⁰ Reduction of the manganese(II) halide complex [L*Mn(thf)(μ-Br)]₂ (642) bearing the silyl amido ligand L*, −N(Ar*)(SiiPr₃) (Ar* = 2,6-{CHPh₂}₂-4-iPr–C₆H₂), with the magnesium(I) dimer [(^Mesnacnac)Mg]₂ (643, ^Mesnacnac = [(^MesNCMe)₂CH]⁻) in 1:2 ratio afforded [L*MnMg(^Mesnacnac)] (644, Scheme 76). According to X-ray crystallographic analysis, in 644, the two-coordinate Mn(0) center is in a bent arrangement (N–Mn–Mg = 160.85(9)°). The Mn–Mg bond length (2.8244(13) Å) is within the sum of the covalent radii of Mg and high-spin Mn (3.02 Å). Magnetochemical studies suggested that 644 has a S = 5/2 ground state, corresponding to the expected high-spin Mn(0) center in 4s²3d⁵ configuration. Theoretical calculations performed on a simplified model in which the isopropyl groups have been truncated to methyl groups, 644′, support the proposed occupation of the five, nonbonding Mn 3d orbitals by a single electron each, and indicate that the Mn–Mg bond has an effective bond order (EBO) of 0.97, essentially originating from the 4s and 3s orbitals on both Mn and Mg centers. Given the small differences in electronegativity of both metals (1.55 for Mn vs 1.31 for Mg, according to the Pauling scale), the Mn–Mg bond was expected to have a covalent character, and to be formed of Mn(0) and Mg(II) atoms. The ‘inorganic Grignard reagent’ behavior of 644 was demonstrated via transfer of the “L*Mn” fragment. For instance, 644 reacts with the Mn(II) dimer [L′Mn(thf)(μ-Br)]₂ (645, L′ = −N(Ar’)(SiMe₃) (Ar′ = 2,6-{CHPh₂}₂-4-Me-C₆H₂), less sterically hindered than its analogue 642, affording the unsymmetrically substituted manganese(I) compound [L*MnMnL′] (646). Similarly, 644 reacted with a chromium(II) complex, forming a bis(amido)manganese mixed valent (Mn(II)–Cr(0)) species believed to result from an internal redox process on a Mn–Cr bonded intermediate, analogous to 646.

Scheme 76. Synthesis of a Two-Coordinate Mn(0)–Mg(II) Complex by Jones and Co-workers⁴⁷⁰.

Kaltsoyannis and Mountford have systematically investigated the reactivity of transition metal compounds with a variety of alkaline earth precursors.^{137,138,419,420} In 2011, this group reported on the reaction between the known ferrate K[(η⁵-Cp)Fe(CO)₂] (K[2]) and alkaline earth metal precursors stabilized by β-diketiminate ligands. Reaction of K[2] with [(^Dippnacnac)CaI(thf)₂]₂ [647, ^Dippnacnac = HC{C(Me)N(2,6-C₆H₃iPr₂)}₂; see Scheme 77a] yielded a dimeric species, [(^Dippnacnac)Ca(μ-OC)₂Fe(η⁵-Cp)(thf)₂]₂ (648),^13,137 as observed in the solid state. 648 featured Fe–CO···Ca isocarbonyl interactions (see Table 1 for ν_CO stretching frequencies) and no apparent Ca–Fe bond. Furthermore, the synthesis of a Ca/Fe heterometallic species [Ca(thf)₃{(μ-OC)Fe(η⁵-Cp)(CO)}₂]₂ (649, Scheme 77b), and the Yb/Fe analogue [Yb(thf)₃{(μ-OC)Fe(η⁵-Cp)(CO)}₂]₂ (650) was reported, which were formed by reductive cleavage of the dimeric precursor [(η⁵-Cp)Fe(CO)₂]₂ (651) with calcium or ytterbium amalgam (Ca/Hg or Yb/Hg). The molecular structure of 649 is composed of a centrosymmetric dimer formed via an isocarbonyl linkage to calcium and shows two unsupported calcium-transition metal bonds (d_CaFe = 3.0185(6) Å). The ytterbium analogue is isostructural to 649. By contrast, attempting an analogous reaction between 651 and magnesium amalgam, followed by crystallization from THF, formed [(thf)₄Mg(μ-OC)₂Fe₂(η⁵-Cp)₂(CO)₂] (652), in which no direct metal–metal bond exists (Scheme 77b). The six-coordinate Mg²⁺ center is connected to the coordination sphere of each iron center only through an isocarbonyl linkage. The natures of the electronic structures of 649 and its ytterbium analogue 650 were studied using theoretical calculations (DFT, molecular orbital, energy decomposition, and atoms-in molecules analyses), and the M–Fe interactions were found to be predominantly electrostatic. Moreover, these analyses explained the formation of 652. The stabilization energy for the coordination of THF to Mg²⁺ is more favorable than the energy required for the Mg–Fe interaction, leading to the formation of the isocarbonyl linkage. Reaction of either 649 or its Yb-analogue 650 with MeI yielded [(η⁵-Cp)Fe(Me)(CO)₂] (653), consistent with the complexes being sources of anion 2.¹³⁷ Subsequent studies revealed that, while compound [(thf)₄Mg(μ-OC)₂Fe₂(η⁵-Cp)₂(CO)₂] (652, crystallized from THF) lacks metal–metal bonds, performing the same reductive cleavage, but crystallizing the product from benzene, yields the centrosymmetric dimeric complex [(thf)Mg{Fe(η⁵-Cp)(CO)₂}₂]₂ (654, Scheme 77b), featuring two Mg–Fe bonds.^137,419 The two Mg–Fe distances in 654 (2.6112(5) and 2.5629(5) Å) are comparable to that for [Cp(dppe)FeMgBr(thf)₂] (610, 2.593(7) Å).⁴⁵³ The strontium derivative “Sr[Fe₂(η⁵-Cp)₂(CO)₂]₂”(655) was also reported, although without structural authentication due to its high sensitivity.⁴¹⁹ Additionally, the authors found that treating complex 652 with hexamethylphosphoramide (HMPA, 4 equiv) yielded the separated ion pair [Mg(HMPA)₄][(η⁵-Cp)Fe(CO)₂]₂ (656), whose structure was confirmed via X-ray diffraction.

Scheme 77. Interaction of Ferrates with Alkaline Earth Metal Species to Obtain Heterometallic Fe-AE (and Lantanide) Complexes^137,138,419.

Similar to the formation of [(^Dippnacnac)Ca(μ-OC)₂Fe(η⁵-Cp)(thf)₂]₂ (648), treatment of K[2] with [(^Dippnacnac)MgI(thf)] (657) affords [(^Dippnacnac)MgFe(η⁵-Cp)(CO)₂(thf)] (658, Scheme 77c).¹³⁸ X-ray diffraction analysis revealed the presence of unsupported Mg–Fe bonds (d_MgFe = 2.6326(4) Å), which are shorter than the sum of the covalent radii of the metals (Σ = 2.73(10) Å), and comparable to those in complexes 610 and 654, described earlier.

The higher stretching frequencies of the carbonyl ligands in complex 658 (ν_CO(658) = 1926, 1857 cm^–1; see Table 1) compared to K[2] (ν_CO = ca. 1865, 1788 cm^–1) reflect the changes in electron density at the transition metal upon formation of the metal–metal bond. As with compound 649, 658 reacted with MeI to afford 653 and regenerated 657, after addition across the Mg–Fe bond. Similarly, di(p-tolyl)carbodiimide (TolNCNTol) inserts into the Mg–Fe bond, affording [(^Dippnacnac)Mg{(TolN)₂CFe(η⁵-Cp)(CO)₂}] (659). Although complex 659 was not crystallographically characterized, its structure was authenticated by DFT computations along with spectroscopic and analytical data of the isolated product. The increase in stretching frequencies in 659 (ν_CO(659) = 2010, 1956 cm^–1), with respect to those in 658, indicates significant charge transfer from the transition metal center to the (TolN)₂C unit. Changes in the Mulliken charge for both complexes are consistent with the observations in infrared spectroscopy since the Fe center in 659 (−0.040) is substantially less negative than in the parent compound 658 (−0.164).¹³⁸

Complementary results on the interaction between alkaline earth metal cations and lanthanide cations with the cobaltate anion [Co(CO)₃(PR₃)]⁻ (660, R = Cy) were reported in 2015.⁴¹⁹ A series of products [Mg{Co(CO)₃(PCy₃)}₂(thf)]₂ (661), [Ca{Co(CO)₃(PCy₃)}₂(thf)₂]_∞ (662), [Sr{Co(CO)₃(PCy₃)}₂(thf)₃]_∞ (663), and [Ba{Co(CO)₃(PCy₃)}₂(thf)₆}] (664) was generated by treatment of [Co(CO)₃(PCy₃)]₂ (665) with amalgams of alkaline earth metals (AE = Mg, Ca, Sr, Ba). Representations of the structures of 661–663 are shown in Figure 14a. Of these, complex 661 features two Co–Mg bonds as in 654, 662 presents only one Co–Ca bond per transition metal center, 663 has a long Co–Sr distance and bears one side-on bound CO ligand, while 664 lacks any metal–metal interaction. Based on the experimental evidence and a systematic theoretical analysis of the bonding situation in these complexes, the authors suggested that, going down the group of the alkaline earth metals, the AE–TM linkage becomes weaker.

Figure 14.

Representation of the structures of (a) Co-AE heterobimetallic complexes and (b) Fe or Co complexes featuring amidinate- or guanidinate-magnesium metalloligands by Mountford and co-workers.^419,420

In fact, theoretical calculations revealed that the AE–Co bonding becomes progressively more ionic going down the group. Nonetheless, by comparing the behavior of the series of complexes derived from cobaltate 660 with the previous results from ferrate 2 or with carbonylcobaltate [Co(CO)₄]⁻ (95), it was concluded that 660 forms inherently stronger AE–TM bonds than the other metalates studied. The HOMO in 660 is less stabilized (by 0.39 eV) than in 95, accounting for the increased nucleophilicity of the former. In summary, although the AE–TM interactions are essentially ionic in general, their strength is favored by a larger charge/size ratio (stronger for Mg–TM bonds, weaker for Sr–TM bonds). However, as observed for complexes 652, 654, and 656, Lewis bases (e.g., THF, (μ-OC)TM η¹-isocarbonyl groups) compete with the formation of the metal–metal bonds, even to the extent of forming ion-separated species.⁴¹⁹

Similar studies were carried out using amidinate or guanidinate magnesium complexes, which were treated with the potassium salts of the transition metal anions [(η⁵-Cp)Fe(CO)₂]⁻ (2) and [Co(CO)₃(PR₃)]⁻ (660, R = Cy or Ph).⁴²⁰ Reactions with the amidinate magnesium compounds afforded the heterobimetallic complexes [{MesC(NR)₂}(thf)Mg{Fe(η⁵-Cp)(CO)₂}] [R = Dipp (666), Mes (667); Figure 14b) and [{MesC(NR)₂}(thf)Mg{Co(CO)₃(PCy₃)}] [R = iPr (668), Dipp (669), Mes (670); Figure 14b). Single crystal X-ray diffraction studies of 668 and 669 confirmed the presence of Mg–Co bonds in the solid state (d_MgCo = 2.525 Å, on average).

Analogous reactions with the guanidinate derivatives yielded the series of complexes [{Me₂NC(Ndipp)₂}Mg{Fe(η⁵-Cp)(CO)₂}]₂ (671) and {Me₂NC(Ndipp)₂}Mg{Co(CO)₃(PR₃)} [R = Cy (672), Ph (673)], shown in Figure 14b (middle and right). The molecular structure of 671 features two Mg–Fe bonds (2.5279(4) Å) and isocarbonyl linkages, and its dimeric solid-state structure is maintained in solution, according to diffusion NMR spectroscopic measurements. The cobalt complexes consist of discrete Mg–Co bonds without forming isocarbonyl bonds, based solely on infrared spectroscopic data, since no structural authentication was available.⁴²⁰

Crimmin and co-workers reported the unsolvated analogue of complex 658, [(^Dippnacnac)MgFe(η⁵-Cp)(CO)₂] (674), which was obtained via an improved protocol consisting of the reduction of dimer 651 with [(^Dippnacnac)Mg]₂ (675, Ar = Dipp; 1 equiv) in hydrocarbon solvents, therefore allowing for the exclusion of coordinating solvents such as THF.⁴²¹ The solid state molecular structure of 674 presents a shorter Mg–Fe distance (d_MgFe = 2.5190(7) Å) than the solvated analogue 658 (d_MgFe(658) = 2.6326(4) Å),¹³⁸ and significantly shorter than in the parent compound 675 (d_MgMg = 2.8457(8) Å).⁴⁶⁵ Theoretical calculations (NBO analysis) confirmed the polar nature of the Fe–Mg bond, with negative and positive charges located on the Fe (−0.41) and Mg (+1.66) centers, respectively. The nucleophilic character of 674 was demonstrated through reactivity with fluoroarenes.

By reaction of compound 674 with the aromatic substrates in a 2:1 ratio (Scheme 78), aromatic substitution of a series of electron-deficient fluorinated aromatic compounds was achieved. Although the efficiency of the substitution was evidently dependent on the nature of the fluoroaromatic compound (perfluorobenzene and perfluorotoluene gave only small amounts of product), the reaction yielded the fluoroaryl-substituted complexes [(η⁵-Cp)Fe(CO)₂(C₆F₄R)] (R = F (676), CF₃ (677), 2-py (678); Scheme 78). Structural authentication was reported for the 2-pyridyl derivative, 678. Furthermore, a trimetallic reaction side product 679 was also characterized, which provided insight into the unusual 674:fluoroarene = 2:1 stoichiometry. Single crystal X-ray diffraction revealed a triangular complex consisting of two Mg(nacnac) units bridged by a fluoride anion, which are further connected to the (η⁵-Cp)Fe(CO)₂ moiety via isocarbonyl linkages. It was proposed that this side product originates from a molecular magnesium fluoride formed as a transient intermediate, which reacts with an additional equivalent of starting material 674. Further insights into the mechanism of nucleophilic aromatic substitution were gained from theoretical calculations. Two mechanistic pathways were considered, a concerted S_NAr mechanism and a stepwise S_NAr pathway with a low energy Meisenheimer intermediate. The analysis of the calculated reaction barriers suggests that a stepwise S_NAr route is likely in operation and could be competitive with a concerted process identified for apolar M–M bonds. The charge separation between the two metals and polarization of the Mg–Fe bond in 674 were proposed as crucial factors favoring the S_NAr mechanism leading to products 676–678. The importance of charge separation to achieve a regioselective C–F functionalization was further demonstrated by the reaction of ferrate Na[(η⁵-Cp)Fe(CO)₂] (Na[2]) with 2-(pentafluorophenyl)pyridine. This reaction affords the same reaction product, 678, as observed when 674 is used as a reagent.^13,421

Scheme 78. Reactivity of [(^Dippnacnac)MgFe(η⁵-Cp)(CO)₂] (674) with Fluoroarenes, By the Group of Crimmin⁴²¹.

A magnesium–iron complex analogous to 674 and related to 658,^138,421 [(L)MgFe(η⁵-Cp)(CO)₂] [680, L = CH₂{PPh₂N(SiMe)₃}₂], featuring a phosphinimine based ligand, was prepared similarly from the corresponding magnesium iodide complex and Na[(η⁵-Cp)Fe(CO)₂] (Na[2]).⁴⁷¹ Predominantly ionic bonding between the magnesium and iron centers was inferred from the stretching frequencies of the carbonyl ligands (ν_CO = 1924, 1860 cm^–1), and corroborated by DFT calculations.

Crimmin and co-workers additionally reported the Rh–Mg complex [(^Dippnacnac)Mg(μ-H)₂Rh(η⁵-Cp*)(SiEt₃)] (681), which was described as a complex of a cationic magnesium fragment coordinated by a rhodate anion (NPA charge at Rh = −0.99).⁴⁷²

Wolf and co-workers synthesized magnesium cobaltates related to 674 in an attempt to determine the influence of the countercation on the stability and reactivity of alkene metalate complexes.⁴²⁵ As described, complexes bearing magnesium-transition metal bonds had long been known, but most of these were carbonyl complexes. The strong d−π* back-donation in such complexes weakens the metal–metal interaction. With the use of the bis(cyclooctadiene) complex K[Co(cod)₂] (K[239]), it was possible to synthesize highly reactive magnesium cobaltates and study their reactivity. A strong influence of the Mg²⁺ cation on the structure and their reactivity pattern was observed in the resulting complexes. Magnesium cobaltates [(^Arnacnac)MgCo(η⁴-cod)₂] [^Arnacnac = [CH(ArNCMe)₂]⁻; Ar = Mes (682), Dep (683), Dipp (684)] were synthesized via salt metathesis of [(^Arnacnac)MgI(OEt₂)] [Ar = Mes (685), Dep (686), Dipp (687)] with the potassium cobaltate K[239] (see Scheme 79). In the solid state, these complexes form contact ion pairs, as confirmed through X-ray diffraction analysis, with the formal oxidation state of the cobalt center unchanged, Co(−I).⁴²⁵

Scheme 79. Synthesis of Magnesium Cobaltates by Wolf and Co-workers⁴²⁵.

DFT calculations performed on 683 (population analyses, analyses of intrinsic bond orbitals, and atoms-in-molecules analyses) provided further insights on the nature of the interaction between both metal centers. NBO analysis shows no Lewis or non-Lewis shared orbitals between the magnesium and cobalt atoms. The Wiberg bond index for the Mg–Co bond is 0.027, and the atoms in molecules analysis reveal no bond critical point between the Mg and Co centers. Therefore, the combined theoretical analysis of the bonding in 683 indicated that the magnesium–cobalt interaction is highly electrostatic, and consistent with a cobalt atom more negatively charged than the magnesium atom. NMR scale studies on the interaction of the magnesium cobaltates toward donor/nondonor solvents showed that the contact ion pair structure is maintained in solution in nondonor solvents such as toluene, while the ion pair separates in donor solvents such as THF. The solvent-separated ion pair [(^Depnacnac)Mg(thf)₃][Co(η⁴-cod)₂] (683·thf) was obtained by crystallization from THF/n-hexane, and its molecular structure was determined via single-crystal X-ray diffraction. The formation of 683·thf is reversible (Scheme 80), as demonstrated after drying 683·thf in vacuo and redissolving it in C₆D₆. Furthermore, the ion-pairing has a direct influence on the reactivity and selectivity of the magnesium cobaltates toward phosphaalkynes. Upon reaction of 683 with tert-butylphosphaalkyne, tBuC≡P, in THF or toluene, different products are accessible, illustrating the influence of the ion-pairing on the reactivity of the magnesium cobaltates. While in THF the reaction with tBuC≡P afforded the previously reported homoleptic bis(1,3-diphosphacyclobutadiene)cobaltate,^208,473 [Co(P₂C₂tBu₂)₂]⁻ (189, as a [(^Depnacnac)Mg(thf)₃]⁺ salt), in toluene the heteroleptic sandwich complex [(^Depnacnac)MgCo(P₂C₂tBu₂)(η⁴-cod)] (688) was obtained (Scheme 80). Since this heteroleptic complex bears a labile cod ligand, further functionalization was achieved by reaction with white phosphorus (P₄) (section 3.4.3).⁴²⁵

Scheme 80. Reversible Ion-Pairing in a Molecular Magnesium Cobaltate Complex and Subsequent Reactivity of Both Species toward Phosphaalkynes.

Moreover, the heterobimetallic complex [(Cp*NiP₂C₂tBu₂)Co(η⁴-cod)] (689) was obtained after 688 reacted with Cp*Ni(acac) (see Scheme 81a). This complex results from the insertion of the “Cp*Ni” fragment into the diphosphacyclobutadiene ring.⁴²⁵ A similar insertion had previously been observed by reaction of the homoleptic 1,3-diphosphacyclobutadiene complex 189 with [Ni₂Cp₃]BF₄ (Scheme 81b). This reaction initially affords an intermediate that proved too thermally unstable to allow for its isolation. This intermediate rapidly transforms into compound 690, in which the Ni center is σ-coordinated by one of the P-atoms of a diphosphacyclobutadiene rings in 189. Intermediate 690 slowly transforms into the 1,4-diphospha-2-nickelacyclopentadiene complex 691, via insertion of the [CpNi]⁺ fragment into a P–C bond of anion 189. Furthermore, 691 rearranges to the thermodynamically more stable 1,3-isomer 692 upon heating.⁴⁷⁴

Scheme 81. Synthesis of Diphospha-nickelacyclopentadiene Species from 1,3-Diphosphacyclobutadiene Species from 1,3-Diphosphacyclobutadiene-Cobaltate Complexes^425,474.

[(^Depnacnac)MgCo(P₂C₂tBu₂)(η⁴-cod)] (688) also served as a starting material for the preparation of bimetallic tetrel(II) cobaltate complexes, namely [Ar^Dipp2TtCo(P₂C₂tBu₂)(η⁴-cod)] [Tt = Ge (693), Sn (694), Pb (695); Ar^Dipp2 = C₆H₃-2,6{C₆H₃-2,6-iPr₂}₂; Scheme 82], and the homoleptic tin(II) complex Sn[Co(P₂C₂tBu₂)(η⁴-cod)]₂ (696).⁴⁷⁵ Complexes 693–695 were cleanly obtained by reaction of the magnesium cobaltate 688 with terphenyl tetrel halides, [Ar^Dipp2Tt(μ-X)]₂ (697, Tt = Ge, X = Cl; Sn, Pb, X = Br), either at −80 °C for the germanium derivative or at room temperature for the tin and lead compounds. X-ray crystallographic characterization of 693–695 revealed that the tetrel atoms are in pseudo-two coordinate environments ligated by the terphenyl-substituent and one of the P atoms of the diphosphacyclobutadiene ligand, and that no covalent metal–metal bonds are apparent between the tetrel atoms and the cobalt atom. Tt–Co distances are longer than the sum of their covalent radii (d_GeCo = 2.8434(6) Å, d_SnCo = 2.9534(5) Å, d_PbCo = 3.0318(6) Å, vs Σr_GeCo = 2.70 Å, Σr_SnCo = 2.89 Å, Σr_PbCo = 2.96 Å). These observations agree well with the quantum chemical calculations performed (DFT and atoms-in-molecules analyses). Likewise, reaction between 688 and Sn(acac)₂ afforded the homoleptic complex 696 (Scheme 82). ³¹P{¹H} NMR spectroscopic studies of the reaction of 688 with Pb(acac)₂ indicated the formation of the analogous lead(II) complex Pb[Co(P₂C₂tBu₂)(η⁴-cod)]₂ (698). However, all attempts to isolate 698 were unsuccessful, due to its thermal instability, which resulted in the deposition of a black precipitate, assumed to be metallic Pb. The crystallographic characterization of 696 showed that, similar to the structures of 693–695, the cobalt centers interact only weakly with the tin atom. The reactivity of complexes 693–696 toward white phosphorus was evaluated (see section 3.4.3).⁴⁷⁵

Scheme 82. Synthesis of Tetrel Cobaltates from the Magnesium Cobaltate [(^Depnacnac)MgCo(P₂C₂tBu₂)(η⁴-cod)] (688)⁴⁷⁵.

The chemistry of heterometallic magnesium–nickel hydride compounds has also been investigated.^476,477 Zhao, Maron, and Xu reported a dihydride-bridged heterotrimetallic complex bearing a tridentate β-diketiminato ligand with a pendant phosphine group.⁴⁷⁶ Treating [Ni(cod)₂] (240) with the magnesium-alkyl complexes [(NNP)MgR] [R = nBu (699), Cy (700), tBu (701); NNP = [CH₃C(2,6-iPr₂C₆H₃N)CHC(CH₃)(NCH₂CH₂PPh₂)] in a 1:2 molar ratio afforded, in all cases, the trinuclear Mg–Ni–Mg complex 702 (see Scheme 83a) with elimination of the corresponding alkene and formation of two bridging hydride ligands. The NNP ligand in 699–701 adopts a different coordination mode upon reaction with the Ni(0) precursor, with the pendant phosphine arm dissociating from the magnesium atom and binding the nickel center in 702, as verified by crystallographic characterization. Structural motifs similar to that in 702 were previously observed by this group at nickel or palladium, using a zinc-based reagent bearing the same NNP tridentate ligand.⁴⁷⁸ The bridging hydride ligands in 702 result from nickel induced β-H elimination from the alkyl groups in the magnesium reagents. This was confirmed by treating a benzyl analogue of 699–701 with complex 240. No reaction was observed, even when conducted at 60 °C, due to the lack of β-hydrogen atoms in the benzyl group. Additional information on the formation of cluster 702 was gained from the reaction of complex 700 with 240 in a 1:1 molar ratio, allowing for the formation of the hydride-bridged heterobimetallic compound 703 (see Scheme 83a) with elimination of cyclohexene.

Scheme 83. Synthesis of (a) Hydride Bridged Heterobi- And Trimetallic Mg–Ni Complexes and (b) a Heterotrimetallic Nickelaspiropentane Featuring Magnesium-Based Metalloligands^476,477.

Crystallographic analysis of complex 703 confirmed that the nickel center retains a cyclooctadiene ligand and is coordinated by the pendant phosphine group of the NNP ligand. Further addition of one equivalent of 700 to 703 led to elimination of cyclohexene and cyclooctadiene and resulted in the heterotrinuclear cluster 702. The calculated mechanism of formation of 702 supports the experimental observation of species 703 and its additional transformation leading to 702, which occurs initially via coordination of the magnesium reagent to the nickel precursor, and later by alkyl transfer from the magnesium to the nickel center. NBO analysis suggests that the Ni–H–Mg motifs in this compound are best described as three-center two-electron bonds. Additionally, the reactivity of 702 toward isocyanides, carbodiimides and alkynes was studied. While isocyanides reversibly coordinate to the nickel center and induce rearrangement of one bridging hydride, the hydride moiety of 702 undergoes insertion of carbodiimides to give amidinate ligands coordinated to magnesium, or dehydrogenation of phenyl acetylene to give alkynyl ligands bridging the magnesium and nickel centers.⁴⁷⁶

Later, Maron, Xu, and co-workers reported a dimagnesium nickelate, a heterotrimetallic planar nickelaspiropentane complex, stabilized by magnesium-based metalloligands.⁴⁷⁷ The reaction of 240 with the β-diketiminato magnesium monoalkyl compound (^Dippnacnac)MgEt (704, ^Dippnacnac = [CH(ArNCMe)₂]⁻; Ar = Dipp) in a 1:2 ratio afforded the heterotrimetallic complex [{(^Dippnacnac)Mg}₂Ni(C₂H₄)₂] (705, see Scheme 83b). The formation of 705 contrasted with that of complex 702: whereas β-H elimination from the alkyl groups in 699–701 generated a thermally stable hydride-bridged complex, in the reaction leading to 705, β-hydrogen elimination from the ethyl group at the magnesium metalloligand should generate a transient hydride species responsible for the hydrogenation of the cyclooctadiene ligands to cyclooctene at the nickel(0) center. The same process also generates the ethylene ligands coordinated to the transition metal in 705. The rare metalla-bis-cyclopropane structure was determined by X-ray diffraction analysis, revealing a strictly linear Mg–Ni–Mg moiety (bond angle = 180°) and a planar hexagonal coordination geometry around nickel (sum of angles = 360°). Furthermore, the C–C bond length of the ethylene ligands indicates the presence of a single bond (d_C–C = 1.494(7) Å). DFT calculations indicate an absence of π-interactions between the carbon atoms, supporting single bond character for the ligands at the nickel center. NMR spectroscopy provided further evidence of the pronounced sp³ character on the C₂H₄ motif in solution. Altogether, the authors concluded that 705 is best described as a nickelaspiropentane species, instead of a π-bound ethylene complex. The coordination sphere of the complex is completed by σ Ni–C bonds and by weak Mg–C contacts, proposed to be agostic interactions, between the magnesium metalloligands and the ethylene units. Additionally, DFT and NBO analyses indicated that the (nacnac)Mg unit acts as a Z-type ligand to the Ni complex. The nickel center in 705 was proposed to be in the oxidation state +II. This assignment was supported by X-ray photoelectron spectroscopy (XPS) performed on the complex. Furthermore, the authors studied the activation of dihydrogen and the hydrogenation of unsaturated substrates by 705 (see section 3.1).⁴⁷⁷

In 2019, Crimmin reported the complex [PdH₃{Mg(^Arnacnac)}₃] (706, ^Arnacnac = [CH(ArNCMe)₂]⁻; Ar = Dipp; Scheme 84a), which exhibits a remarkable hexagonal planar arrangement around the central zerovalent transition metal.⁴²² The species 706 can be seen as a palladate(0) fragment [PdH₃]^3– and three magnesium-nacnac cations [Mg(^Arnacnac)]⁺. DFT calculations suggest that the Pd–Mg interactions in 706 are significantly ionic, with the magnesium atoms bearing a substantial positive charge, while the negative charge is located mainly on the hydrides, but also to some extent on the palladium atom. In the presence of selected alkyl phosphines PR₃ (R = tBu, Cy), 706 was found to be in a ligand exchange equilibrium with a T-shaped palladium phosphine dihydride [Pd(PR₃)(H)₂{Mg(^Arnacnac)})₂] (PR₃ = tBu (707), Cy (708); Scheme 84a). The platinum complex [Pt(PR₃)(H)₂{Mg(^Arnacnac)})₂] (709) shares similar features in terms of structure and reactivity. By contrast, the lighter analogue [NiH₃{Mg(^Arnacnac)}₃] (710) adds PR₃ (R = tBu, Cy) without dissociation of Mg(^Arnacnac) ligands and preserving the original hexagonal arrangement of the hydrides and magnesium donors, thereby forming a compound with nearly ideal hexagonal pyramidal geometry (710·PR₃, Scheme 84a). Calculated Wiberg bond indexes (WBIs) suggest more covalent Pd–Mg and Mg–H interactions in 708 than in 706. Moreover, QTAIM calculations on 708 revealed bond-critical points between the magnesium and hydrogen atoms, but not between the palladium and magnesium centers, as opposed to what is observed for the hexagonal planar precursor 706.

Scheme 84. Hexagonal Planar Complexes of Low-Valent Group 10 Metals-M(nacnac) Complexes (M = Mg, Zn)^422,423.

Shortly thereafter, the Crimmin group expanded on their study of the gap between the hexagonal planar MH₃E₃ (D_3h) or trigonal planar M(E–H)₃ (C_3h) bond situations for a low-valent group 10 transition metal atom interacting with three E–H donors (M = Pd or Pt, E = Zn{^Arnacnac} or Mg{^Arnacnac}).⁴²³ X-ray and neutron diffraction studies in combination with computational studies revealed that the zinc-based donors tend to form C_3h geometries, while the magnesium analogues prefer to adopt a D_3h symmetry (Scheme 84b). The evidence from this work indicates that the hexagonal planar bonding extreme is favored in systems with large ionic contributions to the M–Mg interaction, implying a formal description of the system as a metalate(0) fragment [MH₃]^3– with three [Mg(^Arnacnac)]⁺ cations. The deformation from the trigonal planar to the hexagonal planar configuration was found to have a low energy penalty at the expense of the M–E bond, as evidenced by the corresponding Walsh diagram. Furthermore, it is noteworthy that 706 reversibly reacts with molecular hydrogen to form the tetrahydride species [PdH₄{Mg(^Arnacnac)}₂] (711, ^Arnacnac = [CH(ArNCMe)₂]⁻; Ar = Dipp) and the dimer [(^Arnacnac)Mg(μ-H)]₂ (712).⁴²⁴ In contrast, the heavier analogues [PtH₄{Mg(^Arnacnac)}₂] (713) and related [PtH₄{Zn(^Arnacnac)}₂] (714) feature [M^IIH₄]^4– fragments interacting with two cationic Mg or Zn moieties and, therefore, cannot be described as low-valent metalates.

Magnesium metalates based on metal aggregates have also been described. By treating the nickel cluster [(iPr₃P)Ni]₅H₆ (715) with MgCp₂, the nickel–magnesium compound [(η⁵-Cp)Ni[(iPr₃P)Ni]₄H₆(μ₄-MgCp)] (716, Scheme 85) was obtained.⁴⁷⁹ This compound formed by formal Cp^– transfer with release of a phosphine ligand from 715, as observed upon reaction with either MgCp₂ or TlCp. However, the anionic cluster fragment obtained was found to interact differently with both metal cations: while the thallium cation is covalently bonded to the nickel cluster, forming [(η⁵-Cp)Ni[(iPr₃P)Ni]₄H₄(μ₄-Tl)] (717) with release of H₂, the magnesium analogue stabilizes via ionic interactions. This disparate behavior was attributed to the presence of the harder Mg²⁺ cation in 716, instead of the soft Tl⁺ cation in 717.

Scheme 85. Synthesis of the Nickel–Magnesium Cluster [(η⁵-Cp)Ni[(iPr₃P)Ni]₄H₆(μ₄-MgCp)] (716)⁴⁷⁹.

2.3.3. Metalates Stabilized by Lewis-Acidic Group 13 Metallatranes and Related Complexes

A different tuning effect can be created by rational ligand design, using frameworks that accommodate the stereoelectronic requirements of the metal to support complexes which can react with a variety of substrates. The groups of Peters^{139,141,426−428} and Lu^78,431−435 have extensively investigated the use of Group 13 metallatrane ligands, their coordination to base metals, and the reactivity of the resulting molecular species toward small molecules.²⁵

2.3.3.1. Tris(phosphino)borane Complexes

In 2011, Peters and co-workers reported that the ligand tris[2-(diisopropylphosphino)phenyl]borane (P₃^B)^480−482 stabilizes low-valent iron centers to afford anionic dinitrogen iron metallaboratrane complexes.^139,483 A comproportionation reaction between FeBr₂ and iron powder in the presence of the P₃^B ligand affords the complex [(P₃^B)FeBr] (718). Reduction of 718 with sodium naphthalenide yields different, but related, products depending on the stoichiometry of reductant used. Treatment with a slight excess of NaC₁₀H₈ yields the neutral dinitrogen compound [(P₃^B)Fe(N₂)] (719, Scheme 86), whereas reaction with 2.5 equiv of the reductant leads to the anionic complex [(P₃^B)Fe(N₂)]⁻ (720; isolated as the Na⁺ and [Na([12]crown-4)₂]⁺ salts, Scheme 86 center right). 720 can also be generated in situ by reaction of the Fe(I) complex 718 with an excess of sodium amalgam under a nitrogen atmosphere.⁴²⁶

Scheme 86. Synthesis and Reactivity of Low-Valent Iron Compounds Derived from Tris(phosphino)borane (P₃^B) Ligand^139,426,427.

Exposure of 719 to vacuum resulted in spectroscopic changes indicative of a labile N₂-ligand. This was corroborated by reaction of 719 with carbon monoxide, which afforded the carbonyl compound [(P₃^B)Fe(CO)] (721; ν_CO in Table 1).¹³⁹ Furthermore, treatment of 719 with p-methoxyphenyl azide yielded the diamagnetic imido complex [(P₃^B)Fe≡N–Ar] (722, Ar = p-methoxyphenyl; Scheme 86, top right). Complex [Na([12]crown-4)₂][720] can be further reduced with KC₈ in DME to afford the dianionic complex [Na([12]crown-4)₂][K(DME)_x][(P₃^B)Fe(N₂)] (723).⁴⁸⁴ Compound 723 is diamagnetic in the ground state according to ⁵⁷Fe Mössbauer spectroscopy. Peters and co-workers generated this species in situ from 720 for use in nitrogen fixation studies (see section 3.2.6).

Crystallographic studies revealed that the complexes feature a trigonal bipyramidal structure, where the Lewis acidic borane unit of the ligand P₃^B occupies the apical position and may interact strongly with the Lewis basic Fe atom. A comparison of the crystallographic data for these complexes revealed that changes in the basicity of the metal center cause an adjustment in the Fe–B distance. Therefore, the P₃^B ligand acts as a flexible scaffold that stabilizes different electronic environments. While in the highly reduced complex 720 a short Fe–B distance (d_FeB = 2.311(2) Å) indicates a strong iron–boron contact, a longer Fe–B distance (d_FeB = 2.608 Å) in the Fe(II) compound 722 (Fe≡N–Ar) suggests only a weak Fe–B interaction. The elongation of the Fe–B distance also has consequences on the geometry around the metal center, which shifts from a trigonal-bipyramidal in 720 to a pseudotetrahedral configuration in 722.¹³⁹ Furthermore, 718 reacted with methyl lithium to form the Fe(I) complex [(P₃^B)FeMe] (724). This can be protonated with HBAr^F₄ to generate the coordinatively unsaturated cationic complex [(P₃^B)Fe]BAr^F₄ (725, Scheme 86, bottom). The Fe–B distance in 725 is rather short (d_FeB = 2.217(2) Å), with the iron center residing in a distorted trigonal pyramidal geometry.⁴²⁷ The overall conclusion from this work was that the ability to shuttle between different electronic configurations and molecular geometries makes these compounds particularly suitable for the activation and functionalization of small molecules, and the various contributions of the Peters group in this area are discussed in section 3.2 (vide infra).²⁵

The neutral complex [(P₃^B)Fe(N₂)] (719) served as a platform to synthesize a triad of highly reduced iron-nitrosyl complexes and study their electronic nature.^485,486 The series of complexes [(P₃^B)Fe(NO)]ⁿ [n = +1 (726), 0 (727), – 1 (728); see Scheme 87] are rare examples of isolable reduced nitrosyl compounds. The neutral dinitrogen complex 719 reacts with nitrosonium hexafluorophosphate ([NO]PF₆), which acts both as an oxidant and a nitrosyl source. Salt anion metathesis afforded compound [(P₃^B)Fe(NO)]BAr^F₄ (726). Reduction of 726 with cobaltocene yielded the neutral [(P₃^B)Fe(NO)] (727), which was further reduced to yield the anionic [Na([12]crown-4)₂][(P₃^B)Fe(NO)] (728). In the Feltham–Enemark notation,⁴⁸⁷ these species are described as {FeNO}^8–10. The crystal structures of 726–728 ({FeNO}^8–10) revealed a high linearity in the Fe–N–O angle for all the complexes (726: 175.8(3)°, 727: 176.18(6)° 728: 179.05(12)°).

Scheme 87. Synthesis of the Triad of Iron-Nitrosyl Complexes [(P₃^B)Fe(NO)]ⁿ (n = +1, 0, −1)^485,486.

The structure and electronic nature of the series of iron-nitrosyl complexes was interrogated by experimental and computational methods. The complexes feature an NO⁺ ligand and strong Fe–NO π-backbonding interactions. Hence, the compounds are significantly covalent, which could explain the remarkable stability observed across the three redox states, in contrast with observations for other nitrosyl complexes. The Fe-NO motifs exhibit closed-shell electronic structures thanks to the Fe–B interactions in the complexes. Compound 726 was described as an Fe(0)-NO⁺ species, which features a low spin {FeNO}⁸ configuration. Two very strong metal-to-ligand π-backbonding interactions are established with the NO⁺ ligand.

The crystal structure of this complex revealed a unique metal-olefin π-bond between the iron center and a C=C bond in one of the aromatic benzene rings in the P₃^B ligand. The reduction leading to 727 was found to be metal-based according to EPR, which would indicate the presence of an Fe(−I)-NO⁺ compound with a low spin {FeNO}⁹ configuration in the ground state. Upon reduction of 726 to 727, the Fe–NO bond becomes weaker due to reduction in the covalency of the two Fe–NO π-bonds. The weakening of this interaction was predicted by DFT calculations and is partly caused by the existence of spin polarization effects in the {FeNO}⁹ configuration. In turn, compound 728 is diamagnetic in a closed-shell ground state. Analysis of the molecular orbitals of the complex revealed a completed¹⁰ shell for the iron center, indicating that the second reduction is also centered on the metal and leads to a low spin {FeNO}¹⁰ configuration with a formally Fe(−II) center. The electron density of the highly reduced metal center, however, is accepted by the Fe–B interaction, which strengthens significantly due to the formation of an Fe–B σ single bond. Therefore, the iron center acts as a Lewis base and donates to the boron center. This interaction was described as a true reverse dative Fe → B bond, which resulted in the crucial stabilization of the highly reduced species. Consequently, compound 728 features an Fe(−II) ion ligated to a NO⁺ motif through Fe–NO π-bonds and significantly stabilized by an Fe–B bond. In summary, despite the low formal oxidation states achieved at the iron centers, the nitrosyl ligands maintain the NO⁺ redox state. These results highlight the role of the flexible P₃^B ligand scaffold in the stabilization of highly reduced metal centers, which in this case acts as an electron reservoir by storing two electrons on site. A similar theoretical analysis comparing [(P₃^B)Fe(N₂)]⁻ (720) to the isoelectronic compound 727 revealed that the dinitrogen ligand in the former is less effective at stabilizing the Fe(−I) center than the NO⁺ ligand in 727. These observations agree well with the relative π-accepting abilities of both ligands and explain the formation of an Fe–B σ bond already in the d⁹ state in 720. The authors suggested that these observations additionally support their hypothesis that Fe–B bonding is critical to achieve effective small molecule activation, including N₂ fixation.^485,486

2.3.3.2. Tris(phosphino)silyl Complexes

Peters and co-workers also developed the related tetradentate tris(phosphino)silyl ligands [Si(o-C₆H₄PR₂)₃]⁻ [P^R₃^Si, R = Ph (729), iPr (730)] and demonstrated their versatility in the stabilization of dinitrogen and monocarbonyl iron complexes in different formal oxidation states.^{140,141,429,430} Reduction of the chlorido-complexes [(P^R₃^Si)FeCl] [R = Ph (731), iPr (732)] with either sodium amalgam or sodium naphthalenide under a nitrogen atmosphere yielded the dinitrogen adducts [(P^R₃^Si)Fe(N₂)] [R = Ph (733), iPr (734); Scheme 88]. In both cases, measurements of the magnetic moment indicated the presence of an Fe(I) center [S = 1/2; μ_eff(733) = 1.8 μB, μ_eff(734) = 2.2 μB)]. The presence of terminally coordinated N₂ molecules was revealed via X-ray diffraction analysis. In 733, the short N≡N bond distance [d_N2(733) = 1.106(3) Å] is consistent with the high-energy N≡N stretching vibrations [ν_N≡N(733) = 2041 cm^–1], which indicate that the dinitrogen ligand is only weakly activated.¹⁴⁰ As a result, the dinitrogen ligand in these complexes is labile, being easily removed under vacuum or displaced by carbon monoxide to generate the carbonyl compound [(P^R₃^Si)Fe(CO)] ([R = Ph (735), iPr (736)]; see Scheme 88, Table 1 shows the CO stretching frequencies).¹⁴¹

Scheme 88. Synthesis of Dinitrogen- and Carbonyl-Iron Complexes Featuring Tetradentate Silylphosphino Ligands^140,141.

Further reduction of 733 affords the diamagnetic anionic complex [Na([12]crown-4)₂][(P^Ph₃^Si)Fe(N₂)] (737), formally an Fe(0) species. The analogous [(P^iPr₃^Si)Fe(N₂)]⁻ (738) was similarly obtained.⁴³⁰ By contrast to the neutral 733, the dinitrogen ligand is less labile in this highly reduced compound due to the stronger backbonding of the more electron-rich metal center, evidenced by the stretching vibration observed by FT-IR spectroscopy (ν_N≡N = 1967 cm^–1, see Table 2). Similar to the synthesis of precursor 731, divalent or trivalent cobalt, nickel and iridium compounds bearing the tetradentate ligands [Si(o-C₆H₄PR₂)₃]⁻ were obtained by the same group.⁴²⁹ Furthermore, analogous reduction of the cobalt precursors by sodium amalgam afforded the dinitrogen adducts [(P^R₃^Si)Co(N₂)] [R = Ph (739), iPr (740)].⁴²⁹ However, no further reduction to afford anionic complexes, similar to 737, was reported.

The redox chemistry of the carbonyl complex [(P^iPr₃^Si)Fe(CO)] (736) was investigated by cyclic voltammetry, which revealed two reversible one-electron redox couples.¹⁴¹ These data are consistent with the electrochemical generation of the [Fe-CO]⁺ (−0.68 V) and the [Fe-CO]⁻ (−1.9 V vs Fc/Fc⁺) species. Consequently, the chemical reduction of 736 was carried out, affording the diamagnetic ion-paired species [(P^iPr₃^Si)Fe{CONa(thf)₃}] (741, Scheme 89). The lower energy CO stretching vibrations in 741 (ν_CO = 1717 cm^–1, Table 1) with respect to the neutral 736 (ν_CO = 1850 cm^–1, Table 1) reflect the higher electron density and therefore increased backdonation in the former, more reduced species. Oxidation of 736 with H(OEt₂)₂[B(3,5-(CF₃)₂-C₆H₃)₄] yielded the cationic Fe(II) species [(P^iPr₃^Si)Fe(CO)]BAr^F₄ (742). Moreover, addition of [12]crown-4 (2 equiv) to 741 afforded the ion-separated complex [Na([12]crown-4)₂][(P^iPr₃^Si)Fe(CO)] (743, Scheme 89; ν_CO = 1757 cm^–1, Table 1).

Scheme 89. Synthesis and Reactivity of [(P^iPr₃^Si)Fe{CONa(thf)₃}] (741)¹⁴¹.

X-ray diffraction analysis on the related neutral, reduced, and oxidized species revealed little geometric change about the iron center corresponding to changes in its formal oxidation state. By treating the reduced species 741 with trimethylsilyl trifluoromethanesulfonate (Me₃SiOTf), functionalization of the terminal CO ligand occurs, affording the well-defined iron carbyne species [(P^iPr₃^Si)Fe≡C-OSiMe₃] (744, Scheme 89). The crystallographic data show that the Fe–C bond in 744 is significantly shorter than the one in 736 or 741 [d_FeC(736) = 1.769(2) Å or d_FeC(741) = 1.732(3) Å vs d_FeC(744) = 1.671(2) Å], and that the C–O bond is consequently elongated [d_CO(736) = 1.169(2) Å or d_CO(741) = 1.188(3) Å vs d_CO(744) = 1.278(3) Å], suggesting carbynic character in 744. DFT calculations along with comparative Mössbauer data for all the complexes enabled further insights into their electronic structure, providing evidence for the multiple bond character of the Fe–C bond. An initial analysis of the oxidation state of the metal center in 744 indicated the presence of an Fe(IV) (d⁴) center, ligated by an anionic silylphosphino ligand and a [COSiMe₃]^3– carbyne. However, in analogy to complex 741, the authors considered a scenario in which the carbyne ligand is regarded as a closed-shell cation [COSiMe₃]⁺, suggesting a “Fischer-type” carbyne character with inverted polarization, where the metal center back-donates electron density to the carbon atom. In this case, the iron center would formally correspond to Fe(0) (d⁸). Altogether, the theoretical and spectroscopic data available for 744 indicated the presence of a strongly π-accepting Fischer-type carbyne ligand which stabilizes a low-valent Fe(0) center, instead of a high-valent Fe(IV) center. The authors pointed out that the DFT analysis suggested the further possibility of the complex featuring an Fe(−II) (d¹⁰) center, which cannot be completely excluded when considering the metal and carbyne character of the occupied orbitals.¹⁴¹ Carbyne complex 744 is reminiscent of [{K(Et₂O)}₂{Cp*Co≡CNAr^Tripp2}] (117), which was obtained by Figueroa and co-workers by reduction of an isocyanide complex (see Scheme 12, section 2.2.1).¹³⁵ Further examples of related chemistry are discussed in section 3.3.

Peters and co-workers also investigated the behavior of anionic silatrane complexes of the 4d and 5d metals from groups 8 and 9.^488−490 Reduction of the ruthenium or osmium M(I) metalloradicals [(P^iPr₃^Si)M(N₂)] [M = Ru (745), Os (746); P^R₃^Si = [Si(o-C₆H₄PiPr₂)₃]⁻ (730)] was shown to be feasible by cyclic voltammetry (Ru^I/0 redox pair at −2.14 V and Os^I/0 –1.94 V, vs Fc/Fc⁺) and achieved on a preparative scale by treatment with KC₈ (Scheme 90).⁴⁸⁸ The resulting complexes [K(thf)_x][(P^iPr₃^Si)M(N₂)] [M = Ru (747), Os (748)] are competent catalysts for the N₂-to-NH₃ conversion reaction (see section 3.2).⁴⁸⁹

Scheme 90. Synthesis of Silatrane Complexes of the Group 8 and 9 Second- and Third-Row Transition Metals^488−490.

In a similar vein, the group 9 complexes [K(thf)₃][(P^iPr₃^Si)M] [M = Rh (749), Ir (750); Scheme 90] were isolated by reduction of the dinitrogen complexes [(P^iPr₃^Si)M(N₂)] [M = Rh (751), Ir (752)].^429,490,491 The structural characterization, pulse EPR studies, and theoretical calculations indicate that in 749 and 750 the metal centers bind the silicon atom of the silatrane ligand via 2c-3e σ bonds (“half-bonds”). The electronic structure of these zerovalent complexes is defined by primarily metal-based SOMOs, stabilized by strong TM 4d_z2 or 5d_z2–3p_z orbital mixing, which places substantial spin density on the silicon atom and destabilizes the antibonding transition-metal-silyl σ*-orbitals. Furthermore, 749 and 750 exhibit significant metalloradical character, as shown by EPR spectroscopy and supported by DFT calculations. This was additionally corroborated by H atom abstraction reactions, which generated the M(I)-hydride complexes [K(thf)₃][(P^iPr₃^Si)MH] [M = Rh (753), Ir (754); Scheme 90].

On related chemistry, Filippou and co-workers reported in 2018 that two-electron reduction of [Tp′Mo(CO)₂(PMe₃)Cl] (755, Tp′ = κ³-N,N′,N″-hydridotris(3,5-dimethylpyrazolyl)borate) with 2 equiv of sodium naphthalenide affords the metalate complex Na[Tp′Mo(CO)₂(PMe₃)] (756, Scheme 91).⁴⁹²756 reacted with the NHC-stabilized Si(II) reagent SiBr₂(SIDipp) (SIDipp = C[N(C₆H₃-2,6-iPr₂)CH₂]₂) to yield the metallasilylidyne complex [Tp′(CO)₂Mo≡Si–Mo(CO)₂(PMe₃)Tp′] (757). Cyclic voltammetry studies on 757 revealed an irreversible reduction event (−1.907 V vs Fc/Fc⁺), prompting the authors to investigate its chemical reduction. Two-electron reduction of 757 by KC₈ (2.1 equiv) with concomitant PMe₃ elimination led to the 1,3-dimetalla-2-silaallene dianionic complex [Tp′(CO)₂Mo=Si=Mo(CO)₂Tp′]^2– (758, Scheme 91), isolated as the [K(diglyme)]⁺ salt. Crystallographic characterization showed that dianion 758 features two cumulated double bonds to a linearly coordinated Si atom. The Si atom sits on a crystallographic inversion center, with the two 15 VE Tp′(CO)₂Mo fragments in antiperiplanar conformation. IR spectroscopic analysis revealed intense absorption bands at 1765 and 1696 cm^–1, considerably low wavenumbers, suggesting strong π-backbonding from the electron rich metal centers to the terminal carbonyl ligands. Quantum chemical calculations support the 1,3-dimetalla-2-silaallene description given for 758.

Scheme 91. Reactivity of Metalate Na[Tp′Mo(CO)₂(PMe₃)] (756) with SiBr₂(SIDipp), and Subsequent Reduction of the Obtained Product to Yield a Dianionic 1,3-Dimetalla-2-silaallene⁴⁹².

2.3.3.3. Metallalumatranes and Related Analogues

Metallalumatranes featuring a dative M → Al bond were unreported until 2011, when Lu and co-workers described the first examples with zerovalent iron, cobalt, and nickel.^431,432 This type of molecule is structurally similar to the aforementioned metallaboratranes by Peters.^139,426 In these compounds, the transition metal atom acts as a Lewis base, donating electron density to a Lewis acidic aluminum atom. Deprotonation of the heptadentate ligand N[o-(HNCH₂PiPr₂)C₆H₄]₃ (759, see Scheme 92) followed by addition of AlCl₃ yields the metalloligand AltraPhos (alumatrane-phosphine), which has three pendant phosphine groups.⁴⁹³

Scheme 92. Synthesis of (Hetero)metallic Complexes Featuring the Heptadentate Ligand N[o-(HNCH₂PiPr₂)C₆H₄]3 or Its Related Metalloligands M{N[o-(NCH₂PiPr₂)C₆H₄]₃} (M = Al, Ga, In)^431,433,434.

Reactions of AltraPhos (Scheme 92 top) with FeBr₂ or CoBr₂ and KC₈, or with the readily available Ni(0) precursor [Ni(cod)₂] (240, Scheme 92 middle, M = Al), afforded [(N₂)MAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] {M = Fe [760 (N₂)], Co [761 (N₂)} or [NiAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (762), respectively. End-on coordination of the dinitrogen ligand in 760 and 761 was determined by FT-IR spectroscopy (ν_N≡N = 2010 and 2081 cm^–1, respectively; see Table 2).

Crystallographic analysis of 760–762 confirmed the existence of the M → Al bond and, for 760 and 761, the coordination of the dinitrogen ligand. However, for the iron complex, both the end-on coordinated compound, 760 (N₂), and a N₂-bridged diiron complex, [{760}₂(μ-N₂)], were identified. The observation of both species suggests the existence of an equilibrium between them and shows that the N₂ ligand is labile. Furthermore, a slightly elongated N–N bond distance [1.107(4) Å vs d_NN = 1.0975 Å in free N₂(g)] implies the bridging N₂ ligand is partially activated. The measured M–Al bond lengths indicate that the strength of the metal–metal interaction decreases in the order Ni–Al (2.450(1) Å) > Co–Al (2.6202(9) Å) > Fe–Al (2.809(2) Å). Cyclic voltammetry studies (measured vs [FeCp₂]^0/+) showed single reversible reductions at −0.95 and −2.08 V for 760 (N₂) and 761 (N₂), respectively, assigned to the formal M^–I/M⁰ couples, with the cobalt center being easier to reduce than the iron one, according to the difference in their redox potentials (1.12 V). The nickel analogue 762 showed a reversible oxidation event at −0.74 V, assigned to the Ni⁰/Ni^I redox couple.

Lu and co-workers also reported the synthesis of the heterodinuclear complexes 763 and 764 (Scheme 92 middle, M = Ga, In)⁴³⁴ and of the mononuclear nickel complex [Ni{N[o-(HNCH₂PiPr₂)C₆H₄]₃}] (765, Scheme 92 bottom).⁴³⁵ The metalloligands M{N[o-(NCH₂PiPr₂)C₆H₄]₃} (M = Ga, In) and the complexes [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [M = Ga (763), In (764)] were synthesized by extension of the protocol used for the preparation of both AltraPhos and 762.⁴³¹ Complexes 762–764 constitute a family of compounds with systematic tuning of the electronic environment of the Ni(0) center.⁴³⁴ The different electronic environments created by the group 13 M(III) centers are evidenced by the downfield shift in the ³¹P NMR resonance corresponding to the Ni-coordinated phosphine moieties [δ(C₆D₆) = 30.8 (765), 31.3 (762), 38.3 (763), 45 (764) ppm]. For the indium(III) derivative, 764, the synthesis must be carried out under an argon atmosphere since, under N₂, the dinitrogen coordinated species 764 (N₂) is obtained (Scheme 92 middle). The dinitrogen ligand remains invariably coordinated even under prolonged vacuum but suffers ligand exchange after treatment with H₂ to generate the stable dihydrogen compound 764 (η²-H₂) (Scheme 92 middle), which does not lose dihydrogen under vacuum. An interesting behavior was observed for the rest of the Ni(0) complexes upon exposure to either N₂ or H₂. While neither 765 nor 762 form N₂ or H₂-adducts, complex 763 is able to form the dihydrogen complex 763 (η²-H₂). The analysis of HD coupling constants and T₁ relaxation time measurements performed by ¹H NMR spectroscopy suggest that 763 (η²-H₂) and 764 (η²-H₂) are genuine H₂ complexes with intact H–H bonds. H–D distances of 0.87 and 0.91 Å, respectively, were deduced from the H–D couplings, compared to 0.74 Å in free H₂. Unlike 764 (η²-H₂), 763 (η²-H₂) loses H₂ under vacuum, reverting to the parent compound 763.

Crystallographic characterization and NMR spectroscopic data of 762, 763, and 764 indicate stronger Ni → M(III) dative bonds moving down the series M = Al, Ga, In, inversely related to their ionic radii. The Ni center thus becomes more electron-deficient for the larger M(III) ions in the metalloligand. Structural data also revealed that, as more electron density is withdrawn from the Ni center, the metal is forced higher above the plane of the three P atoms, becoming more accessible to bind small molecules such as N₂ or H₂.⁴³⁴ The use of complexes 762–765 as catalyst precursors in the hydrogenation of olefins is discussed in section 3.1.4 (vide infra).

The adducts [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [M = Al (762), Ga (763)], bearing zerovalent nickel centers,^431,434 each feature a reverse dative Ni(0) → M bond. Cyclic voltammetry (CV) of these compounds returned strongly negative potentials, with a reversible one-electron transfer process observed at −2.82 and −2.48 V vs [FeCp₂]^0/+ for 762 and 763, respectively. Neither the mononuclear nickel complex 765 nor the free metalloligands M{N[o-(NCH₂PiPr₂)C₆H₄]₃} (M = Al, Ga) exhibited redox processes near those potentials, indicating that the electron transfer observed by CV was a function of the Ni–M moiety.

Consequently, the isolation of heterobimetallic metalates [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [M = Al (766), Ga (767), Scheme 93] resulting from the reduction of the neutral species 762 and 763 with KC₈ was reported in 2018.⁷⁸ The molecular and electronic structures of these anions were elucidated by X-ray crystallography, EPR spectroscopy, and quantum chemical calculations. The crystallographic investigations revealed that the Ni–Al bond in 766 contracts by 0.06 Å in comparison with 762, indicating a strengthened interaction between the metal atoms.

Scheme 93. Synthesis of Formal Nickelate(−I) Complexes Supported by Group 13 Metalloligands, and Their Proposed Resonance Structures⁷⁸.

The EPR spectroscopic studies revealed substantial hyperfine couplings to the ²⁷Al, ⁶⁹Ga, and ⁷¹Ga nuclei of 7.6, 37.0, and 47 mT, respectively. Based on the experimental and theoretical data, three limiting resonance structures were conceived: two featuring metal-centered metalloradicals [Ni(0)-M(II) or Ni(−I)-M(III)] and one intermediate structure where the additional electron injected into the complex upon reduction is shared between the Ni and M atoms. However, considering the changes observed in the bonds around the Ni center, the strength of the Ni–M dative bond, and the spin density distribution (as predicted via ³¹P-hyperfine coupling and theoretical calculations), the contribution of a Ni(0)-M(II) resonance structure was ruled out, thus leaving the structures shown in Scheme 93 as the most probable. Taken collectively, the analysis (X-ray diffraction, EPR, and theoretical calculations) supports the proposal of a strongly reduced formal Ni(−I) character on the transition metal center of complexes 766 and 767, with an unusual three-electron σ-bonding interaction.⁷⁸

2.3.3.4. Metalates with Bonds between First-Row Transition Metals

Heterobimetallic complexes with bonds between first-row transition metals (TM) were also synthesized by the group of C. Lu, who has obtained different TM-based metalloligands from the heptadentate ligand 759.^{433,435,494−496} For instance, the complex [Cr{N[o-(NCH₂PiPr₂)C₆H₄]₃} (768) served as platform for the synthesis of a series of first-row M–Cr heterobimetallic compounds [MCr{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [M = Mn (769), Fe (770), Co (771), Ni (772); Figure 15] for which a systematic tuning of the metal–chromium bonds was observed.^433,435

Figure 15.

First-row heterobimetallic complexes featuring the metalloligand Cr{N[o-(NCH₂PiPr₂)C₆H₄]₃} (768). Metal–metal and Cr–N distances to the apical N atom (N_ap) are shown for comparison. Average values for two crystallographically independent molecules are given for compounds Fe (770) and Co (771).⁴³³

The syntheses of 769–772 are largely analogous to those used for compounds 760–762 (Scheme 92). Crystallographic characterization showed increasing M–Cr bond distances and with decreasing formal bond order from 5 to 1 in the series from 769 (M = Mn) to 772 (M = Ni). Compound 769 has an ultrashort metal–metal bond distance of 1.8192(9) Å, associated with the presence of a quintuple bond. An inverse correlation was observed for the Cr–N distance, which decreased progressively from 769 to 772. Multiple one-electron transfer processes were identified by cyclic voltammetry for these complexes. Theoretical calculations indicated an increasing polarization of the M–Cr bonds from 769 to 772.⁴³³

Similar to the reactions that gave 766 and 767,⁷⁸ chemical reduction of 770 with KC₈ (1 equiv, 94a) proceeds with formation of the anionic complex [FeCr{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (773), isolated as the [K([18]crown-6)]⁺ or the [K([2.2.2]cryptand)]⁺ salt.⁴³⁵ Crystallographic analysis of 773 revealed a slight elongation of the Fe–Cr bond length [d_FeCr (770) = 1.944 vs d_FeCr (773) = 1.974 Å] after reduction. Cr–N_axial distance is also elongated (Δd = 0.06 Å), while the Cr–P bonds are shortened (Δd = 0.02 Å) with respect to the neutral complex. Although such observations are consistent with the more reduced state of 773, they indicate that the one-electron reduction is localized primarily at the chromium center, unlike in the cases of 766 and 767. Hence, the authors assigned the tentative oxidation states Fe°Cr^II for the metal centers in 773, and corroborated their assignment via DFT calculations and ⁵⁷Fe Mössbauer spectroscopy.⁴³⁵

Scheme 94. Chemical Reduction of Heterobimetallic Iron Complexes Featuring Metalloligands Based on First-Row Transition Metals^435,495.

Similarly, the titanium metalloligand [Ti{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (774) and the iron–titanium compound [FeTi{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (775) were described by the same group, and their redox behavior studied to yield the series [FeTi{N[o-(NCH₂PiPr₂)C₆H₄]₃}]ⁿ [n = 1+ (776), 0 (775), 1– (777)].^495,496 Reduction of complex 775 with KC₈ afforded the diamagnetic compound [K(thf)₃][FeTi{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (777, Scheme 94b). Analysis of the molecular orbital situation indicated an increased contribution of the titanium center in the Fe–Ti π-bonding MOs of 777, while in the rest of the series, this contribution is rather limited. In addition, with the successive reduction events, the Fe–Ti bond length decreases substantially (d_FeTi(776) = 2.1613(10) Å, d_FeTi(775) = 2.0545(7) Å, d_FeTi(777) = 1.9494(6) Å). This contraction was attributed to a better energy match between the Fe and Ti 3d orbitals, which increases the Ti character in the σ- and π-symmetric MOs and results from the accumulation of negative charge at the Fe center. The metal centers in 775 and 777 are stabilized by both increased π-backbonding interaction to the phosphine groups and increased covalent π-bonding between the Fe centers and the Ti support, which delocalizes some of the excessive electron density from Fe to Ti.

Considering this, the authors proposed that the electronic structures of 776 and 775 are best described as containing a Ti(IV) support, and Fe(0) or Fe(−I) centers, respectively. In turn, the polarized Fe≡Ti bond in 777 presents spectroscopic features which would suggest the presence of an Fe(−II) center. Collectively, the theoretical and spectroscopic evidence indicates that the reductions leading to each member of the series [FeTi{N[o-(NCH₂PiPr₂)C₆H₄]₃}]^+/0/–, from 776 to 775 to 777, are iron-based. This contrasts with the analogous FeCr complex 773, for which the chromium support metal was assigned as the center of reduction.⁴³⁵

3. Metalates in Small-Molecule Activation and Catalysis

The activation of small molecules is of great interest to contemporary inorganic chemistry due to their enduring relevance in important industrial processes such as catalysis and energy conversion. While the term “small molecules” might apply to a broad spectrum of chemical compounds, we wish to focus on the contributions studying the activation of H₂, N₂, CO₂, P₄, and some related model species.

As discussed in section 2.2.4, numerous alkene or arene metalates have been reported.^{246,376,386,441,497,498} The reactivity studies performed on these complexes have proven their remarkable potential toward the formation of other coordination compounds and the activation of inert chemical bonds. Therefore, low-valent/highly reduced species seem particularly suited as effective tools to perform the catalytic transformation of small molecules.⁴⁹⁹ Indeed, there has been steadily growing interest around small molecule activation promoted by transition metal anions in recent years. In this work, we will only highlight the contributions in which low-valent metalates mediate the transformation of small molecules or models for small molecules. However, in recent years, their participation as catalysts or catalytic intermediates in many other interesting reactions, such as cross-couplings (refs (417, 447−449, 455−458, 500−503)) olefin isomerization,^314,319 functionalization of dienes,¹⁹ carbonylation of heterocycles,⁵⁰⁴ oligomerization of isocyanides,⁵⁰⁵ or C–H bond activation and functionalization^22,506 has been widely documented.

3.1. Hydrogen Activation, Catalytic Hydrofunctionalization, and Dehydrogenation

Hydrogenation catalysis has been of keen interest to organometallic chemists for decades, and the attention this type of reaction receives is ever-growing as it is a versatile tool in, for example, the synthesis of numerous reduced organic compounds⁵⁰⁷ through either classic or transfer hydrogenation reactions.⁵⁰⁸ Likewise, other hydrofunctionalization reactions and catalytic dehydrogenations⁵⁰⁹ can be useful to produce not only organic compounds and polymers, among others, but also hydrogen. Accounts on the advances in catalytic hydrogenation promoted by low-valent transition metals are available elsewhere. For instance, in 2015, Chirik reviewed the use of iron and cobalt complexes bearing redox-active ligands as catalysts in the hydrogenation of alkenes.³⁹⁴ However, to the best of our knowledge, the chemistry of metalates in hydrogenation reactions has yet to be comprehensively reviewed.

3.1.1. Arene/Alkene Metalate Catalyzed Hydrogenation

Iron,^376,497 cobalt,^{246,386,441,497} and nickel⁴⁹⁸ arene- and alkene-metalates have been used as precatalysts for hydrogenation,^{246,441,497,498} transfer hydrogenation,³⁸⁶ and hydrofunctionalization³⁷⁶ reactions of a variety of unsaturated substrates. From these studies, it has become evident that metalates hold great potential to perform this type of reductive reaction. The reducing power of anionic complexes makes them particularly suitable for the activation of molecular hydrogen and, consequently, for hydrogenation reactions.

The use of bis(anthracene)metalates(−I) [M = Fe (52); Co (14)] as catalysts for hydrogenation reactions was reported for the first time in 2014.⁴⁹⁷ These quasi-“naked” anionic metal species are stabilized by labile anthracene ligands which allow for a rapid exchange in the presence of a large excess of other π-acid ligands, i.e. unsaturated substrates such as olefins, ketones, etc. The hydrogenation of α-, β-, and ring-substituted styrenes (see Scheme 95a) was achieved under mild conditions (1 mol % catalyst, 20 °C, 2 bar H₂, toluene), while other linear- and cyclic alkenes (terminal, internal, and di- and trisubstituted) required a higher catalyst loading, as well as higher pressure and temperature (5 mol %, 10 bar H₂, 60 °C, toluene). In general, cobaltate 14 resulted in a significantly more active catalyst than the more oxidation-prone ferrate 52, which efficiently hydrogenated only unbiased styrenes and 1-alkenes. This behavior might be due to the tendency of 52 to form the anthracene radical anion in the presence of strong donors in ligand exchange reactions, as previously demonstrated (see Scheme 21b, above).¹⁷⁹ In any case, both anthracenemetalates work essentially under ligand-free conditions and require no activation procedures or treatments, returning only minimal amounts of polymeric material as byproducts in reactions with styrenes and alkenes as substrates. Cobaltate complex 14 is also active in the hydrogenation of other unsaturated substrates, such as alkynes, aromatic and aliphatic ketones, and imines (10 bar H₂, 60 °C, Scheme 95a).

Scheme 95. Arene/Alkene Metalate-Catalyzed Hydrogenation of C=X Moieties (X = C, N, O) by Wolf, Jacobi von Wangelin, and Co-workers^246,497,498.

Based on monitoring studies and poisoning experiments (Hg; dibenzo[a,e]cyclooctatetraene, dct), it was proposed that the hydrogenation of olefins is homogeneous and proceeds by initial substitution of the anthracene ligands by the π-acceptor substrates. Rapid exchange of anthracene by styrene was identified by ¹H NMR spectroscopy. However, no reaction was observed between cobaltate precursor 14 and H₂ in absence of the substrate. Thus, it was proposed that H₂ activation occurred via oxidative addition at the cobalt-alkene species [Co(−I) → Co(I)], followed by insertion of the coordinated alkene into a Co–H bond to generate an alkylcobalt(I) hydride. The presence of π-acceptors (olefins, arenes) in excess in the reaction medium appear to stabilize the catalytically active species in the cycle. For the hydrogenation of carbonyl moieties, it was also postulated that the initiation of the cycle occurred at cobaltate 14, with a cobalt(I) active species or metallic cobalt particles operating after the first turnover, at elevated temperature and H₂ pressure.⁴⁹⁷

Other alkene metalates, the heteroleptic cobaltate [Co(η⁴-C₁₀H₈)(η⁴-cod)]⁻ (245), the classic [Co(η⁴-cod)₂]⁻ (239), and cobaltates [Co(η⁴-cod)(η²-styrene)₂]⁻ (265), [Co(η⁴-dct)(η⁴-cod)]⁻ (266), and [Co(η⁴-dct)₂]⁻ (267) (see Scheme 32a, above), were evaluated as precatalysts for hydrogenation of alkenes (see Scheme 95b).²⁴⁶ The bis(anthracene)metalates(−I) [M = Fe (52); Co (14)] were re-evaluated alongside these, due to their greater solubility in THF rather than toluene, which had been used in the previous catalytic experiments.⁴⁹⁷ For the sake of comparison, ferrates [Cp^RFe(η⁴-naphthalene)]⁻ [Cp^R = Cp* (201), Cp (202)] were also tested, in parallel reactions for all of the precatalysts, under identical reaction conditions (5 mol % catalyst, 20 °C, 2 bar H₂, THF, 24 h), using styrene and 1-dodecene as model substrates. In general, and consistent with previous observations for catalyst 52,⁴⁹⁷ the ferrates showed lower activity in the hydrogenation reactions than the cobaltates. The homoleptic bis(η⁴-dct)cobaltate 267 was completely inactive under the studied conditions, confirming the role of dct as a catalyst poison for homogeneous low-valent mononuclear species, as previously postulated in the case of the bis(anthracene)metalate(−I)-catalyzed reactions.⁴⁹⁷

It is believed that the strong coordination of the dct ligand inhibits the ligand exchange process with the substrates. The polyarene cobaltates proved the most active precursors for both model substrates, with the heteroleptic complex 245 being the most efficient (Scheme 95b), in terms of both activity and selectivity (lower isomerization to internal alkenes for 1-dodecene). An extension of the substrate scope to include polar substrates (ketones and imines) revealed that harsher reaction conditions (10 bar H₂, 60 °C) are required for precatalyst 14 to perform efficiently in the hydrogenation of ketones and imines. Precatalysts 245 and 239 resulted in moderately efficient catalysis for the hydrogenation of dibenzylketone, but for N-benzylideneaniline were essentially inactive.²⁴⁶

Extensive mechanistic studies (NMR spectroscopy, ESI–mass spectrometry, and poisoning experiments) of the alkene hydrogenation reactions support a homogeneous pathway initiated by coordination of the substrate to the metal center, in a facile redox-neutral π-ligand exchange process, prior to H₂-activation.⁴⁹⁷ For the reactions catalyzed by either 245 or 239, mechanistic studies at the NMR scale revealed that the addition of styrene (20 equiv) to solutions of each complex resulted in the clean formation of complex 265 in both cases. Application of a H₂ atmosphere to the reaction mixtures rapidly led to complete hydrogenation of the coordinated styrene to ethylbenzene. However, hydrogenation of 265 in the absence of additional substrate led to immediate color-change to form a black solution, due to full consumption of the stabilizing π-ligand. Analogous hydrogenation of 265 in the presence of an excess of naphthalene resulted in the formation of its precursor, 245, the most efficient among the studied complexes (see Scheme 32a). Such reconstitution of the catalyst was referred to by the authors as a release–catch mechanism (Scheme 96), which occurs after the complete hydrogenation of the reactive alkenes. Hence, 265 can be effectively assigned as a potential intermediate of styrene hydrogenations by cobaltate precatalysts. A species analogous to 265, namely [Co(η⁴-C₁₄H₁₀)(η²-styrene)₂]⁻ (778), was also observed through an ESI-MS analysis of a solution of 14 treated with styrene (20 equiv; see Figure 16, precatalyst formation). An equivalent experiment using 1-dodecene instead similarly afforded an analogue of 778.²⁴⁶ The observation of the 18-valence-electron bis(alkene) complexes, analogous to 265, in the different reaction mixtures obtained from the cobaltates used as precatalysts strongly supports the ligand exchange mechanism originally postulated by the authors (Figure 16). These are, presumably, the resting states that serve as a reservoir for the catalytically active cobalt species in each case. It was thus speculated that the H₂ activation occurs after loss of an alkene ligand from such resting states, with concomitant formation of an unsaturated and reactive 16 e^– monoalkene-cobalt species.²⁴⁶

Scheme 96. Release–Catch Mechanism Proposed as Part of the Reaction Pathway in the Hydrogenation of Alkenes by Labile Cobaltates²⁴⁶.

Figure 16.

Activation of arene metalate precatalysts for hydrogenation reactions by π-ligand exchange with alkenes.^246,497

The mechanism for the bis(anthracene)metalate-based hydrogenation catalysts was also examined by DFT calculations.⁵¹⁰ For both catalysts, 14 and 52, the general steps of the mechanistic proposal were calculated to be similar. Initially, one substrate molecule displaces one anthracene ligand, and one hydrogen molecule coordinates. Then, oxidative addition of the hydrogen molecule yields a M(I)-dihydride species, followed by migratory insertion to give the M(I)-alkyl (or M(I)-alkoxide) intermediate. In the alkene hydrogenation cycle, an anthracene molecule recoordinates the metal center, promoting the reductive elimination of the alkane product (C–H bond formation) and regenerating the catalyst. The results of the calculations suggest that, before the oxidative addition step, there is an equilibrium between the closed-shell singlet state of 14 and the open-shell diradical Co(0) intermediates, whereas oxidative addition of the hydrogen molecule generates a more stable Co(I)-dihydride species. Hence, this system involves a Co(−I)/Co(0)/Co(I) catalytic cycle, and species both in singlet and triplet states (nonadiabatic reaction with spin crossing). Similarly, it was proposed that two different spin states (doublet and quartet) and three different oxidation states [Fe(−I)/Fe(0)/Fe(I)] are involved in the Fe(−I)-catalyzed alkene hydrogenation reaction. The rate-determining step is the insertion of the alkene substrate into the Co(I)–H bond, overcoming an energy barrier of ∼22.0 kcalmol^–1. For the Fe(−I)-catalyzed reaction, the migratory insertion of the substrate into a M–H bond is also the rate-determining step, but the process is associated with a higher energy barrier of 25.4 kcalmol^–1.

The higher barrier was attributed to the lower stability of its reactive low-spin metal-dihydride intermediate [M(anthracene)(H)₂(alkene)]⁻ [ΔG = 19.0 (M = Fe) vs 13.6 (M = Co) kcalmol^–1]. Moreover, a comparison of the Mulliken charges of the M(−I) catalysts and the M(H₂) intermediates indicated that the formation of the key low-spin oxidative-addition Fe(I)-dihydride intermediate is disfavored, which might explain its lower reactivity. Flexible coordination modes of the acidic ligands (e.g., anthracene decoordination/recoordination) involved in the catalysis favor the key elementary steps in the cycle. Similar theoretical observations were made for the ketone hydrogenation system, although it should be noted that the hydrogenation reactions are likely by a heterotopic catalyst (i.e., metallic nanoparticles of Fe and Co, respectively, vide supra).

Highly reduced nickel complexes are rare, and not many examples exist regarding their use in activation of small molecules. Wolf and Jacobi von Wangelin described that the long-known nickelate(−II) [Li₂(thf)₄{Ni(η²-cod)(η⁴-cod)}] (9) enables the heterogeneous hydrogenation of C=X moieties (X = C, N, O) in olefins, imines, and enoates under mild conditions (Scheme 95c).⁴⁹⁸ Precatalyst 9 was previously synthesized by Jonas and co-workers,⁶² and only recently characterized by crystallographic analysis.⁵⁰² Unlike the bis(anthracene)metalates(−I) [M = Fe (52); Co (14)], which performed poorly for sterically hindered tri/tetra-substituted olefins,^246,4979 is an efficient precatalyst in the hydrogenation of hindered olefins. Using the same catalyst loading [1 mol % catalyst, 25 °C, 5 bar H₂, 15 h, 1,2-dimethoxyethane (DME)], the performance of the nickelate complex 9 in the hydrogenation of either triphenylethylene or 1-phenyl-1-cyclohexene was also significantly better than that of the related oxidized compound [Ni(η⁴-cod)₂] (240). Nonetheless, complex 9 does not catalyze the hydrogenation of polyaromatic compounds such as anthracene or naphthalene, or heterocycles such as quinolines. Compared to precatalysts 14 or 52, the active nickel species generated from 9 can hydrogenate sterically hindered tri/tetra-substituted olefins at lower H₂ pressures and temperature, while being compatible with several functional groups (OH, esters, halides). Different behavior was also observed in the monitoring of the reaction progress for selected substrates (1-octene, 2-octene, and α-methylstyrene) at lower pressures (1.9 bar H₂), for which considerable induction periods and sigmoidal behaviors were observed. These might indicate slow catalyst formation and nucleation processes to form heterogeneous species during the reaction. This proposal was strongly supported by poisoning experiments with Hg, dct, benzonitrile and naphthalene, which indicated that the operating active species was heterotopic. Analysis of the formed species through transmission electron microscopy (TEM) revealed the presence of particles 10–15 nm and larger in diameter. For reducible olefins (e.g., triphenylethylene), spectroscopic analysis and cyclic voltammetry suggested the formation of the triphenylethylene-dianion, indicating a possible electron-transfer initiated mechanism. The mechanistic experiments performed support the postulate that nickel nanoparticles are responsible for the catalytic activity observed, while the formation of the catalyst might be substrate dependent.⁴⁹⁸

3.1.2. Hydrogenation Catalysts Containing Redox-Active Ligands

Redox-active (noninnocent) ligands have attracted considerable interest as stabilizing scaffolds for cobalt hydrogenation catalysts, and have been the focus point of recent studies.^377,441 Fe(II)-BIAN precatalysts were evaluated for the hydrogenation of olefins (including tri- and tetrasubstituted susbtrates).³⁷⁷ The tetrahedral complex [(^DippBIAN)FeCl₂] (779), analogous to [(^DippBIAN)FeBr₂] (536),³⁷⁶ was the precatalyst of choice for the hydrogenation of the alkene substrates, and was found to catalyze the reaction only after activation with a strong reductant. Carrying out catalytic tests with a system consisting of in situ treatment of 779 with nBuLi (3 equiv), the selected olefinic substrates studied in this preliminary evaluation were hydrogenated in excellent yields (Scheme 97a).

Scheme 97. Hydrogenation of C=X Bonds Mediated by Reduced “(BIAN)M” (M = Fe, Co) Catalytic Systems^377,441.

Nonetheless, the authors tried to identify the formed species by reproducing the activation reaction on a preparative scale. This led to the isolation of two main complexes after extraction, in a 70:30 ratio: the previously reported complex [(^DippBIAN)Fe(η⁶-C₇H₈)] (780)³⁷⁵ and a species whose structure could not be fully identified (781, see Scheme 97a for a representation of the products). Complex 780 is the product of formal two-electron reduction of 779, whereas the minor species 781 was proposed to be a low-valent compound formed by three-electron reduction. It was suggested that the low-valent complex 781 might feature either a radical anion or a dianionic BIAN ligand. 781 reproduced the catalytic activity of the in situ generated system, while 780 proved inactive in the hydrogenation of α-methylstyrene (1.9 bar H₂, 20 °C, 3 h).

The use of BIAN-stabilized cobaltate complexes³⁷² as catalysts allowed for the hydrogenation of sterically hindered trisubstituted alkenes, imines, and quinolines under mild conditions (2–10 bar H₂, 20–80 °C).⁴⁴¹ The simple catalytic system is formed from the stable precursor [(^DippBIAN)CoBr₂] (782) and LiEt₃BH as cocatalyst (Scheme 97b). The reduction of the Co^II catalyst precursor in the presence of olefins proved beneficial for the overall catalytic activity, probably due to their coordination to, and consequent stabilization of, the low-valent species. Under standard conditions, turnover frequencies (TOF) of 780 h^–1 were achieved for the hydrogenation of α-methylstyrene. Synthetic, kinetic, and spectroscopic experiments provided further details on the reaction mechanism and support a homogeneous pathway involving homotopic cobaltate catalysts, probably stabilized by alkene/hydrido ligands. For instance, Hg poisoning experiments had minimal effect on the reaction rate, while addition of P(OMe)₃ resulted in the partial or complete inhibition of the catalytic activity even when substoichiometric amounts (0.3 mol %) of the poisoning agent were used.

Furthermore, when complex 782 was treated with LiEt₃BH (3 equiv), and the crude catalyst mixture was analyzed using LIFDI–MS (liquid injection field desorption–mass spectrometry), the low-valent dimeric species [(^DippBIAN)Co]₂ was identified. The authors surmised that such a dimeric species might be indicative of the participation of a monomeric “(^DippBIAN)Co” fragment in the catalysis. Carrying out the reductive dehalogenation of 782 in the presence and absence of arenes/olefins yielded three different reduced species (Scheme 98) providing further evidence of the redox noninnocence of the BIAN ligand. Reactions of 782 with LiEt₃BH (3 equiv) in THF with 1,5-cod present, in benzene, or in Et₂O, afforded the cobaltate [Li(thf)_3.5{(^DippBIAN)Co(cod)}] (783), the neutral complex [(^DippBIAN)Co(η⁶-C₆H₆)] (784), or the dinuclear hydridocobaltate [Li(thf)₃(L){(^DippBIAN)Co}₂(μ-H)₃] (785, L = Et₂O, thf), respectively. According to the bond distances determined by X-ray diffraction analysis, complex 784 features the radical anion state of the BIAN ligand [C–C bond length = 1.433(2) Å; average C–N bond length = 1.3235 Å]. Additional characterization (EPR spectroscopy, magnetic moment, LIFDI–MS, and cyclic voltammetry) revealed that 784 contains a high-spin Co(I) center and thus is accurately described as [(BIAN^–)Co^I(η⁶-C₆H₆)]. Compound 785 also bears the radical anion state of the BIAN ligand according to the structural parameters [C–N bond lengths = 1.333(3)–1.349(3) Å; C–C bond lengths = 1.412(3)–1.419(3)Å], which are in good agreement with the theoretical predictions.⁴⁴¹

Scheme 98. Cobalt Complexes Isolated from the Reactivity Studies of [(^DippBIAN)CoBr₂] (782) with LiEt₃BH in the Presence/Absence of Olefins/Arenes⁴⁴¹.

It was revealed that the counterion plays a significant role on the catalytic activity of the precatalysts. While the potassium salt 540 resulted in complete inactivity, the lithium analogue 783 catalyzed the reaction, although less effectively than the in situ combination of 782 and LiEt₃BH. Addition of Lewis acids improved the activity of 783. The η⁶-benzene complex 784 was also completely inactive, while the dinuclear hydridocobaltate anion 785 had a high activity and was considered a catalytically competent off-cycle intermediate.⁴⁴¹

Roşca and co-workers found that phosphine α-iminopyridine (PNN)-based Fe–N₂ complexes activate molecular hydrogen via metal–ligand cooperation (MLC) to afford an anionic Fe(0)-hydride complex.⁵¹¹ The group reported that the highly reactive anionic dinitrogen compound K[(PNN)Fe(N₂)] (786, Scheme 99) shows divergent behavior in the presence and absence of molecular hydrogen. Compound 786 is obtained by reduction of the dinuclear iron dinitrogen complex [{(PNN)Fe(N₂)}₂(μ-N₂)] (787) with KN(SiMe₃)₂ under N₂ in the presence of a crown ether or cryptand at low temperatures (<−30 °C, Scheme 99), and features a dearomatized pyridine core with a deprotonated benzylic position (radical anionic ligand). Above this temperature and in the absence of H₂, 786 undergoes ligand-assisted disproportionation to afford two new paramagnetic species, [K(chelate)][788] and [K(chelate)][789] (chelate = [2.2.2]cryptand or [18]crown-6; see Scheme 99, top). The formation of these two species was explained by an intermolecular hydrogen atom transfer from the methyl substituent on the α-carbon atom of the imino group to the vinylic carbon atom adjacent to the PtBu₂ moiety of another complex. Compound 788 corresponds to the one-electron reduction product of the dinuclear complex 787 which can be independently synthesized by reduction of 787 with KC₈ in the presence of a chelating agent. FT-IR stretching bands for the dinitrogen ligand in 788 show that this complex is a highly reduced product with increased metal-to-ligand backbonding [ν_NN(788) = 1927 or 1912 cm^–1 for the [K([2.2.2]cryptand)]⁺ or [K([18]crown-6)]⁺ adducts, respectively, vs ν_NN(787) = 2039 and 2059 cm^–1]. Compound 789 is a conjugation-stabilized complex. The relative ease of C–H bond cleavage in the α-imine methyl group (calculated BDFE = 48.7 kcalmol^–1) and subsequent proton coupled electron transfer (PCET) to stabilize the negative charge through ligand-based reduction was proposed as the driving force for the formation of 788 and 789.

Scheme 99. Divergent Behavior of K[(PNN)Fe(N₂)] (786) in the Presence and in Absence of Molecular Hydrogen, and Formation of an Anionic Fe(0)-Hydride Complex⁵¹¹.

Upon treatment of the anionic complex 786 with dihydrogen (7 bar, 16 h), a color change from brown to green was observed, and a diamagnetic species M[(PNN)FeH(N₂)] [M = K([18]crown-6) (790)] with a phosphorus-coupled hydride resonance in the ¹H NMR (δ_H = −22 ppm, J_HP = 63 Hz) formed. In the hydride complex, the pyridine ring of the PNN ligand is rearomatized, with the ligand oxidized to its neutral form. The formation of 790 was attributed to a ligand-assisted dihydrogen splitting reaction, where the driving force is the rearomatization of the chelating ligand. The MLC was confirmed by the reaction of 786 with D₂, upon which not only was a deuteride complex observed, but also deuteration in the benzylic position of the phosphino arm of the ligand. The mixture of disproportionation products 788 and 789 react with dihydrogen under the same conditions to afford hydride 790. However, when 788 was exposed to D₂, no incorporation of deuterium on the α-positions of the ligand was observed, suggesting a different mechanism of activation for the dihydrogen molecule. In this case, a bimolecular (homolytic) metal-based dihydrogen splitting occurred.

The anionic hydride complex can be more conveniently accessed (as the sodium salt 791) upon reaction of the dinuclear complex 787 with an excess of NaBMe₃H (4 equiv per Fe center). Reaction of either the dinuclear complex 787 or the anionic species 786 with carbon monoxide afforded the dicarbonyl compounds [(PNN)Fe(CO)₂]^0/– [792 (neutral), 793 (anionic)]. The anionic complex features a dearomatized pyridine core and exhibits contrasting behavior compared with the dinitrogen analogue 786. 793 is stable at room temperature and in solution (1 week) without undergoing PCET reactivity, probably because of the nonradical nature of the chelate, and does not react with H₂, up to pressures of 10 bar. The contrasting behavior of the described complexes highlights the regulating role that the PNN ligand plays on the reactivity of its compounds.⁵¹¹

3.1.3. Pincer-Supported Manganates in Hydrogenation Catalysis

Kempe and co-workers described an efficient and chemoselective Mn(I)-mediated system for the hydrogenation of imines to amines and sought insight into its mechanistic and kinetic minutiae. Potassium manganate species were found to play mechanistically crucial roles in imine hydrogenation reactions using complex [BrMn(CO)₂(PN^triazineP)] (794, PN^triazineP = 2,6-bis(PiPr₂NH)-4-Ph-triazine) as the precatalyst.⁴⁴⁰ The reduction of aldimines and ketimines (over 30 examples; conditions: 20 bar H₂, 50 °C, 4 h, see Scheme 100) was achieved using a low catalyst loading (0.4 mol %) for most substrates. The system tolerates imines with challenging and sensitive functional groups (olefins, ketones, nitriles, nitro, aryl iodo substituents, benzyl ether). For the model substrate N-benzylideneaniline (Ph-N=CH-Ph) and other selected examples, the reaction could be conducted on a multigram scale, with yields of at least 90%.

Scheme 100. Hydrogenation of Imines Catalyzed by [BrMn(CO)₂(PN^triazineP)] (794)⁴⁴⁰.

Mechanistic investigations included studies on the role of the base in the activation of the catalyst, determination of the order of the reaction components, stoichiometric tests to gain insights into the reaction pathway, and a Hammett study. Stoichiometric tests on the NMR scale were carried out using the previously described hydride complex [HMn(CO)₂(PN^triazineP)] (795) since this complex was known to form from the reaction between 794, KOtBu (1 equiv), and H₂.⁵¹² The reaction of 795 with 1 or 2 equiv of KOtBu gave the products of single and double deprotonation of the NH motifs in the ligand backbone, [Mn–H]HK (796, [Mn–H] = [HMn(CO)₂(PN^triazineP)]) and [Mn–H]K₂ (797), were identified. No further deprotonation was observed upon addition of excess base (10 equiv). Reproducing these reactions in the presence of substrate (equimolar ratio) revealed that for substoichiometric amounts of base, no hydride transfer occurred. By contrast, the reaction is significantly accelerated when 2 equiv of KOtBu are used. Therefore, the doubly deprotonated potassium manganate species 797 was identified as the catalytically active species in this system (Scheme 101).

Scheme 101. Mechanistic Cycle Proposed for the Hydrogenation of Imines, Involving Potassium Manganate Intermediates⁴⁴⁰.

In the bimetallic complex 797, the potassium cations are coordinated to the anionic nitrogen atoms in the ligand backbone, and to the triazine cavity. The inactivity of the neutral and singly deprotonated complexes (795 and 796, respectively) contrasts with the efficiency of the doubly deprotonated compound 797, suggesting that the hydride transfer might be favored by the anionic nature of the manganese species, resulting from the two negatively charged nitrogen atoms. Because of these charged atoms, conjugated with the triazine motif, the hydride ligand is more electron-rich. Further investigations into the catalytic cycle indicated that catalyst precursor 794 is activated by an excess KOtBu in the presence of H₂, to produce the doubly deprotonated species 797 (Scheme 101). After hydride transfer to the imine substrate, the amine product and species [Mn]K (798, Scheme 101) are formed. The latter compound reacts with H₂ and KOtBu to regenerate the catalytically active species, 797.

X-ray crystallographic studies showed that compound 798 crystallizes as a coordination polymer, in which the potassium cation binds both the triazine cavity and a negatively charged nitrogen atom, additionally linking another unit of the complex. The authors attributed the exceptional stability of the bimetallic catalyst to the cavity provided by the triazine motif. The influence of the reaction components on the rate of reaction was also investigated. The reaction rate has a first order dependence on the concentration of imine substrate and on the bimetallic hydride complex [Mn–H]K₂ (797) but is independent of the concentration of base. Therefore, the process corresponds to a global second order reaction. The hydride transfer is a well-defined reaction, probably via an outer-sphere mechanism since no observable manganese amide complex was identified. The dihydrogen molecule is activated by the bimetallic dicarbonylmanganate 798 to regenerate 797. In conclusion, both manganate species are essential in the performance of the described catalytic imine hydrogenation system.⁴⁴⁰

Liu and co-workers reported that a related polymeric manganate hydride complex bearing a PNP pincer ligand, [Li{HMn(CO)₂(PNP)}]_n [799, PNP = N(CH₂CH₂PPh₂)₂] also served as an active catalyst in hydrogenation reactions.⁵¹³ Synthesized by deprotonation of the neutral precursor [HMn(CO)₂(PN^HP)}] [800, PN^HP = NH(CH₂CH₂PPh₂)₂] by LiCH₂SiMe₃ under 20 bar of H₂, 10 mol % of the lithium manganate hydrogenated a series of aldimines and ketimines; conditions: 40 bar H₂, 140 °C, 16 h. Notably, 800 was much less effective under the same conditions.

It is noteworthy that carbonyl manganate anions related to 797 and 799 have been proposed as steady state species and catalyst intermediates in chemoselective electrochemical hydrogenations of aldehydes and ketones.⁵¹⁴

In related work, Shaouma and co-workers presented the catalytic hydrogenation of CO₂ to formate using the PNP-pincer complex 801 (see Scheme 102).⁵¹⁵ Deprotonation of 801 gives a reactive hydride species [(*PNP)MnH]⁻, which binds CO₂ with its ligand backbone to generate the species [(CO₂–PNP)MnH] (802). This complex is capable of eliminating formate with binding of further CO₂ to give 803. Following a second ligand deprotonation, yielding 804, addition of H₂ regenerates the hydride complex 802. This work shows how metal–ligand cooperativity in anionic manganese pincer complexes enables an effective mechanism for CO₂ hydrogenation, which circumvents the classic insertion step into the metal hydride bond. Ligand deprotonation is crucial to boost the hydridcity of the complex.

Scheme 102. Catalytic Hydrogenation of CO₂ to Formate Enabled by Low-Valent Manganate Species⁵¹⁵.

These studies illustrate the significant potential of bifunctional alkali metal manganate catalysts for hydrogenation reactions.

3.1.4. Bimetallic Transition-Metal Magnesium Catalysts

In 2022, Maron and Xu described the activation of the dihydrogen molecule at the heterotrimetallic complex [{(^Arnacnac)Mg}₂Ni(C₂H₄)₂] (705, vide supra, section 2.3.2).⁴⁷⁷ Reaction of 705 with H₂ (r.t. for 24 h or 60 °C for 2 h) allowed the isolation of the diamagnetic compound [{(^Arnacnac)Mg}₂Ni(μ-H)₄] (805, Scheme 103), along with free ethane (detected by ¹H NMR spectroscopy). X-ray diffraction analysis of 805 showed that the core of the trimetallic (Mg–Ni–Mg) complex is slightly bent, with the four hydride ligands approximately in the same plane.

Scheme 103. Activation of Dihydrogen at the Heterotrimetallic Complex [{(^Arnacnac)Mg}₂Ni(C₂H₄)₂] (705)⁴⁷⁷.

The hydrides present weak interactions with the magnesium centers (on average, d_MgH = 2.03 Å). This evidence, along with characterization by ¹H NMR spectroscopy indicating the presence of hydride ligands (δ = −13.47 ppm) and XPS measurements consistent with a Ni(II) center in 805 as in 705, suggest that the activation of H₂ likely involves a cooperative heterometallic Ni/Mg mechanism, instead of oxidative addition to the Ni center. By exposing 805 to an ethylene atmosphere, the nickelaspiropentane complex 705 was reobtained. Additionally, evidence of the stepwise conversion of 705 into 805 was obtained from the reaction of 805 with a stoichiometric amount of ethylene (2 equiv), yielding the mixed-ligand complex [{(^Arnacnac)Mg}₂Ni(C₂H₄)(μ-H)₂] (806, Scheme 103). Compound 806 is an isolable intermediate in the transformation of 705 and H₂ (r.t.) to 805. Altogether, the experimental evidence and theoretical calculations supported a cooperative Mg/Ni dihydrogen activation mechanism in which the transition metal center does not suffer oxidation.⁴⁷⁷

Using 1 to 5 mol % of 705 as a precatalyst, under 1 bar of H₂ and at room temperature, 705 efficiently catalyzed the hydrogenation of a variety of unsaturated substrates (aliphatic and aromatic olefins, silyl enol ethers, enamines, imines, quinolines, arenes and alkynes; 16 examples in total, see Scheme 104). Most olefinic substrates were hydrogenated in quantitative yield using a catalyst loading of 1 mol %, except for p-fluoro styrene, the silyl enol ether 1-phenyl-1-trimethylsiloxyethylene, and the enamine 1-pyrrolidinocyclohexene, which required a higher catalyst loading (5 mol %). The hydrogenation of the silyl enol ether was also evaluated using the nickel(0) and nickel(II) complexes [Ni(cod)₂] (240) and (Ph₃P)₂NiCl₂ (807) under the same reaction conditions. Whereas the trimetallic nickelaspiropentane complex 705 gave quantitative conversion of the substrate to its hydrogenated product, the Ni(0) species 240 returned a maximum of 16% yield after a prolonged reaction time (24 h). No product was detected using the Ni(II)-phosphine complex 807. These results stress the importance of the presence of the Z-type Mg-metalloligands at the transition metal and support their involvement in the cooperative activation of the dihydrogen molecule as key in the efficiency of 705 as a precatalyst in hydrogenation reactions.⁴⁷⁷ Additionally, the system constitutes a rare example of nickel-catalyzed hydrogenation of imines, yielding over 90% of the corresponding amine products using a catalyst loading of 5 mol %. Notably, the complex is capable of hydrogenating more challenging substrates in high yields under mild conditions (room temperature and 1 bar H₂) using a higher catalyst loading. In the case of arene substrates, higher temperatures (80 °C) were necessary to achieve comparable yields of hydrogenated products.⁴⁷⁷

Scheme 104. Hydrogenation of Unsaturated Substrates Catalyzed by [{(^Arnacnac)Mg}₂Ni(C₂H₄)₂] (705)⁴⁷⁷.

In 2019, Chen and Zeng described a simple system based on chromium or cobalt salts, diimine or carbene ligands, and MeMgBr, which mediates the hydrogenation of polycyclic aromatic hydrocarbons (PAHs).^17,516 While the precatalysts are salts in common oxidation states [M(II), M(III); M = Cr, Co], the theoretical mechanistic modeling (DFT) revealed that low-valent anionic intermediates participate during the reaction given the presence of MeMgBr. Using the continuous variations method (Job’s plots), the authors determined that the reactive species in the hydrogenation, for either metal, has a 1:1 metal-to-ligand stoichiometry. For the Cr-mediated reaction, according to DFT studies, oxidative addition of H₂ to diimine Cr(0) or Cr(I) intermediates would be thermodynamically unfavorable. Accessible pathways were calculated for a transmetalation process from MeMgBr to the Cr center, in which hydrogenolysis of the resulting methyl-Cr complex produces an active monohydride metalate with loss of methane (Scheme 105). Theoretical analysis of the cobalt complex revealed a similar activation process in the NHC-Co system (Scheme 105). Moreover, the calculations indicated that for both Cr and Co, energetically low-lying intermediates are found only in the case of the M(0) species for the second hydrogenolysis, occurring upon coordination of the PAHs substrate (specifically anthracene). Consequently, in both cases, zerovalent anionic monohydrides, formed via hydrogenolysis, were invoked as the active species for the hydrogenation of PAHs.⁵¹⁶

Scheme 105. Formation via Hydrogenolysis of Active Monohydride M(0) (M = Cr or Co) Intermediates Involved in the Hydrogenation of PAHs⁵¹⁶.

3.1.5. Hydrogen Activation and Hydrogenation Catalysis with Metallatranes and Related Complexes

Dihydrogen complexes have been postulated as intermediates preceding H₂ activation at transition metals, initiating the process through side-on coordination of the H₂ molecule.^517−519 Peters and Lu have contributed to this field by using 3d metal complexes featuring chelating metalloligands, reporting examples of hydride transfer reactions as well as catalytic hydrogenations.^{434,519−525} For instance, these two groups independently reported hydride transfer reactions from cobalt- and nickel-dihydrogen complexes, respectively, supported by boratrane ligand scaffolds.^520,521 The anionic dihydrogen complex [(P₃^B)Co(η²-H₂)]⁻ (808, P₃^B = tris[2-(diisopropylphosphino)phenyl]borane; as [M(thf)_n]⁺ salt, M = Na, K)^526,527 was obtained by reduction of the Co(I) compound [(P₃^B)CoBr] (809) with strong reductants (NaC₁₀H₈ or K). 808 acts as a strong hydride donor, slowly reacting with the poor hydride acceptor BEt₃ under a hydrogen atmosphere to generate the corresponding Et₃BH^– salt in high yields (86% in 20 h) and the mixed-hydride complex [(P₃^B)CoH(η²-H₂)] (810, Scheme 106a).⁵²⁰ This behavior shows that the thermodynamic hydricity of 808 almost equates that of main group hydrogen donors or other strong hydridic transition metal complexes. Although the behavior was kinetically reproducible, the experimental evidence could not unequivocally support either of the possible reaction pathways for the hydride transfer, i.e., direct hydride transfer or formation of a dihydride intermediate. Theoretical calculations pointed toward the dihydride intermediate, [(P₃^B)Co(H)₂]⁻, as the lower energy transition state, favorable by 4.8 kcal mol^–1.⁵²⁰

Scheme 106. Cobalt- and Nickel-Dihydrogen Complexes Supported by Boratrane Ligands, Reported by the Peters and Lu Groups^520,521.

Later, the Lu group described a similar behavior at an anionic Ni(0) hydride complex, featuring a related double-decker boron metalloligand.⁵²¹ Complex [HNiB{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (811), obtained from the in situ generated dihydrogen adduct [(η²-H₂)NiB{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [812(η²-H₂)] in the presence of [2.2.2]cryptand and base (Scheme 106b), acts as an extraordinarily strong hydride donor, also capable of reacting with BEt₃ via complete hydride transfer within minutes. Contrasting with the behavior observed for the cobalt-dihydrogen complex 808,⁵²⁰ the experimental and theoretical evidence suggests that hydride transfer from 811 is considerably more favorable than H atom transfer. Furthermore, this study suggested that the ease of deprotonating dihydrogen ligands is more dependent on the stability of the resulting hydride complex, than on the binding energy or degree of activation of the coordinated dihydrogen molecule.

Considering the propensity of [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (M = Ga (763), In (764), Scheme 92, section 2.3.3.3, vide supra) to bind dihydrogen,⁵¹⁹ these were used as precatalysts in olefin hydrogenation (Scheme 107a).⁴³⁴ The hydrogenation of styrene was efficiently catalyzed by complex 763 under mild conditions (>99% yield, TOF = 2.4 h^–1, 5 mol % precatalyst, 1 atm H₂, r.t.), while 764 performed poorly. Neither the monometallic compound 765 nor the metallalumatrane 762 exhibited activity. Precatalyst 763 performed better for terminal unhindered olefins (e.g., 1-octene, 1-hexene) than for internal olefins, and it was inactive for the hydrogenation of alkynes or aldehydes. 764 mediated hydrogenation much slower for all the olefins–for styrene, hydrogenation was 24-times slower using 764 instead of 763. In the presence of H₂, 764 primarily promoted their isomerization, while in its absence only traces of isomerized substrates were detected. Although no catalytic cycle was explicitly proposed, key mechanistic insights were obtained. Mercury poisoning tests suggested that a homogeneous species is responsible for the catalysis. Furthermore, since 764 promotes olefin isomerization only in the presence of H₂, it was proposed that the same mechanistic pathway might be operating for both precatalysts, albeit much slower for 764. Labeling studies carried out under catalytic conditions with complex 763 and D₂ showed incorporation of deuterium atoms in every position of the double bond, which indicates a reversibility on the insertion of the olefin substrate, made possible via β-H elimination. The differing reactivity between [NiGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (763) and [NiIn{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (764) was preliminarily attributed to a lower relative rate of reductive elimination of the hydrogenated product for 764 with respect to 763, rendering β-H elimination a competitive pathway and leading to the isomerization products. It was proposed that isolable dihydrogen complex 764(η²-H₂) might be the resting state, with further activation of H₂ acting as the rate-determining step in the catalytic reaction mediated by 764. At low temperatures, 763(η²-H₂) was also proposed as the resting state, while at room temperature, 763(η²-H₂) reversibly coordinates the substrate in a fast equilibrium. Furthermore, in the hydrogenation of styrene, a normal primary kinetic isotope effect of 1.7 was measured, suggesting that the cleavage of the H–H bond might be involved in the rate-determining step.⁴³⁴

Scheme 107. Hydrogenation of (a) Olefins and (b) CO₂ Catalyzed by [NiGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (763)^434,522.

Compound 763 also showed remarkable activity in the direct hydrogenation of CO₂ to formate at ambient temperature, in the presence of a Verkade proazaphosphatrane base (Scheme 107b).⁵²² High turnover numbers (TON = 3150) and frequencies (initial TOF = 9700 h^–1), as well as excellent yields were obtained. Base-assisted heterolytic cleavage of H₂ occurred at the Ni(0) complex 763(η²-H₂) thanks to the cooperative interaction between the transition metal and the Lewis acidic Ga(III) center from the stabilizing metalloligand, which acts as a σ-acceptor. The significant effect exerted by the metalloligand was ascribed to a strong inverse trans influence of the Ga Lewis acid. This becomes evident by comparison of the catalytic tests using complexes 765 or 763. The former complex is practically inactive for the hydrogenation of CO₂ under the studied conditions. Unlike 763, 765 does not bind H₂,⁴³⁴ indicating this coordination is a crucial step in the catalysis. Likewise, the in situ combination of 765 with GaCl₃ did not afford the product, proving that the metalloligand framework in the precatalyst, with the intact Ni–Ga interaction, is necessary for the performance of the catalyst. The necessity of a strong base was confirmed, by contrasting the reactivity of the Verkade base with weaker bases such as NEt₃, for which no appreciable activity was observed.

Two key catalytic intermediates were identified for the hydrogenation reaction promoted by 763: the anionic Ni(0) hydride [HNiGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (813) and the anionic formate adduct [(HCO₂)NiGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (814). Both intermediates were observed by NMR spectroscopy, independently prepared, and structurally characterized as bis(triphenylphosphine)iminium (PPN) salts. 813 was obtained by deprotonation of 763 with n-BuLi, while 814 is formed after exposure of 813 to an atmosphere of CO₂. By contrast, the Ni(0) complex 763 does not bind CO₂ even at higher pressures (34 atm). ³¹P NMR spectroscopic screening of the reaction indicated that the predominant species after the catalysis is the formate adduct 814, suggesting no appreciable decomposition of the catalyst. In addition to characterization by physical and spectroscopic methods, DFT calculations support the formation of a terminal hydride complex, which is more stable than a bridged Ni(μ-H)Ga intermediate. The thermodynamic hydricity of 813 indicates that it is a strong hydride donor (ΔG°_H– ≈ 31 kcalmol^–1), the strongest reported for nickel at this time, and among the strongest of any 3d metal complex (note that the aforementioned Ni(0) hydride 811 has an even lower thermodynamic hydricity, ΔG°_H– ≈ 21,4 ± 1.0 kcal mol^–1).⁵²¹ A mechanism for the hydrogenation of CO₂ to formate was postulated using this heterobimetallic system (see Figure 17), in which complex 763 initially coordinates H₂, to form the nonclassical dihydrogen species 763(η²-H₂). Subsequent deprotonation generates the anionic terminal hydride 813. Hydride transfer to CO₂ produces complex 814, which, after release of the formate product, regenerates catalyst 763 (Figure 17).⁵²²

Figure 17.

General catalytic cycle for the hydrogenation of CO₂ to formate mediated by [NiGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (763).⁵²²

Lu and Gagliardi gathered further mechanistic details on the hydrogenation of CO₂ catalyzed by 763 and similar complexes, compiled in a follow-up study.⁵²³ A key insight from this work is that the deprotonation of the coordinated dihydrogen ligand in 763(η²-H₂) not only depends on the basicity but, more strongly, on the steric hindrance of the base, while the hydride transfer from 813 to the CO₂ molecules occurs via an outer-sphere mechanism. The dependence on the base strength is reduced once enough formate has been formed, according to energy barrier calculations. Formate then has a cocatalytic role in the reaction and acts as a proton shuttle between 763(η²-H₂) and the Verkade base (see Figure 17). These observations might indicate that the overall rate-determining step is the formate-assisted hydride transfer to CO₂, rather than the deprotonation of the coordinated η²-H₂ ligand as originally proposed.

The study was extended to similar heterobimetallic complexes, in two different series: M₁–M₂ complexes (M₁ = Fe, Co, Pd, or Pt and M₂ = Al or Ga) and Ni–M₂ (M₂ = Fe, Co, Al, Ga, or In). For these, it was found that the free energy of activation for the hydride transfer to CO₂ is linearly dependent on the thermodynamic hydricity of the metal hydride. The calculations also indicate that the formate release step should be favored for various combinations of metals in the heterobimetallic complexes: CoAl, CoGa, NiAl, and NiGa. The authors found that, theoretically, the best catalysts for the hydrogenation of CO₂ in these series should be the NiGa (as in 763) and CoGa combinations. Moreover, it was proposed that further improvement of the performance of catalyst 763 might be achieved by reducing steric hindrance on the phosphine motifs of the heptadentate ligand and on the base. This might not only facilitate the deprotonation on the dihydrogen complex but could also allow the use of weaker or cheaper bases.⁵²³

In further studies, Lu and co-workers observed moderate hydrogen activation at Co^–I(η²-H₂) scaffolds featuring the Lewis acidic group 13 metalloligands.⁵²⁴ The highly reduced cobaltates [Co(η²-H₂)M{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ {M = Al [815(η²-H₂)], Ga [816(η²-H₂)], In [817(η²-H₂)]} were obtained by treating the corresponding metalloligands with CoBr₂ and LiEt₃BH at low temperatures under argon (see Scheme 108).

Scheme 108. Synthesis of [Li(thf)_n][Co(η²-H₂)M{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [815(η²-H₂)–817(η²-H₂)] Cobaltates by Lu and Co-workers⁵²⁴.

The supporting group 13 ions exert an inverse trans-influence, which induces side-on H₂ coordination and fine-tunes its degree of activation in these rare d¹⁰ dihydrogen adducts. Consistent with the low-valent, formally negative oxidation state, the compounds are diamagnetic. T_1 min relaxation times and J_HD measurements performed on the broad hydridic signals observed by ¹H NMR spectroscopy (δ = −7.0, −7.7, and −7.7 ppm for M = Al, Ga, In, respectively) correlate well with the assignment of the H₂ motif as a nonclassical dihydrogen ligand (T_1 min = 26, 27, and 29 ms, respectively) and with the fast H/D scrambling observed at ambient temperature. The d_HH bond distances obtained from these NMR experiments [0.96(1) (Al), 0.98(3) (Ga), 1.00(1) (In) Å] are longer than those previously obtained for the Ni analogues 763(η²-H₂) and 764(η²-H₂) [0.87 (Ga), 0.91 (In) Å)], for which short T_1 min relaxation times of ≤16 and 23 ms were measured.⁴³⁴ The isoelectronic Ni(0) complexes, however, do not scramble H₂/D₂. Altogether, these observations indicate greater degree of activation for the dihydrogen ligand in the cobaltates compared to the Ni compounds, possibly due to the greater π-basicity of Co relative to Ni or the formally negative oxidation state of the Co center.

To gain insight into the nature of the H₂ ligand, DFT calculations were performed considering the possible hydridic species, namely the nonclassical hydride [Co(η²-H₂)ML]⁻, the terminal dihydride [Co(H)₂ML]⁻, or the bridging/terminal dihydride [H–Co(μ-H)ML]⁻ (L = N[o-(NCH₂PiPr₂)C₆H₄]₃), as in those reported for similar Ni-boratrane species.^528,529 The dihydrogen-Co complexes were energetically more favorable than the species featuring terminal dihydrides by only a small energy difference. The authors suggested that such a small energy difference might indicate facile H₂ cleavage at the Co(−I) center, in agreement with the fast H/D scrambling observed in the NMR experiments.⁵²⁴

The influence of the group 13 center on the reactivity of the heterobimetallic cobaltates is clearly illustrated by their performance as precatalysts for CO₂ hydrogenation.⁵²⁵ The group of Lu observed preliminarily that, upon exposure to a mixture of H₂/CO₂ (1:1; 1.8 atm), complexes 815(η²-H₂) and 816(η²-H₂) produced formate and another cobalt-containing species. For comparison, the related compounds [(N₂)CoM{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ {M = Al [815(N₂)], Ga [816(N₂)], In [817(N₂)]} were prepared. Considering the ease of preparation of the end-on dinitrogen adducts and the good yields of the synthetic protocol used, further in-depth catalytic and mechanistic studies were performed with complexes 815(N₂)–817(N₂). The synthesis of 815(N₂)–817(N₂) is conducted as a one-pot reaction, by stirring the corresponding metalloligand (M = Al, Ga, In) with CoCl₂ (THF as solvent), followed by Na/Hg (3.1 equiv) and [PPN][tetrakis(3,5-bis(trifluoromethyl)phenyl)borate]. Upon exposure to a hydrogen atmosphere (1.8 atm), 815(N₂)–817(N₂) transform rapidly into the dihydrogen adducts 815(η²-H₂)–817(η²-H₂), confirming that 815(N₂)–817(N₂) could be suitable precursors in the hydrogenation reactions. Consequently, 815(N₂)–817(N₂) were tested as precatalysts for the hydrogenation of CO₂ (Scheme 109, top) under the optimized conditions established for the Ni(0)Ga system [Catalyst loading: 0.031 mol %, Base: Verkade proazaphosphatrane (stoichiometric), CO₂/H₂ = 1:1, 34 atm, room temperature; Scheme 107b, vide supra].⁵²² In these tests, 817(N₂) was completely inactive, while 815(N₂) and 816(N₂) catalyzed the hydrogenation of CO₂, affording moderate to high yields of formate (59% and 97%, respectively) and initial turnover frequencies of 1500 and 11000 h^–1 (see Scheme 109, bottom). For the most active complex 816(N₂), the catalyst loading could be lowered as far as 0.004 mol %, achieving a remarkable 6-fold increase in turnover number from 3100 (for a catalyst loading of 0.031 mol %) to 19000.

Scheme 109. Hydrogenation of CO₂ Catalyzed by Group 13 Metalloligand-Stabilized Cobaltates⁵²⁵.

Model reactions were performed to understand the behavior of the catalytic system. The reversible coordination of the dihydrogen ligand to the cobalt center was, for example, found to be mechanistically relevant. By replacing the atmosphere of the reaction vessel from H₂ to N₂, the dihydrogen adducts transform back into the dinitrogen complexes (Scheme 110, top).

Scheme 110. Synthesis of Mechanistically Relevant Intermediates in the Hydrogenation of CO₂ by Group 13 Metalloligand-Stabilized Cobaltates⁵²⁵.

As previously observed, exposure of 816(N₂) to a CO₂/H₂ (1:1, 1.8 atm) mixture afforded formate, along with a new cobalt species identified as the mixed-hydride Co(I) complex [(η²-H₂)HCoGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [818(η²-H₂)]. Under vacuum, 818(η²-H₂) releases H₂, forming [HCoGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (818). Similar behavior was observed for the aluminum derivative 815(N₂), yielding the analogous compounds 819(η²-H₂) and 819. Both Co(I)Ga hydrides, 818(η²-H₂) and 818, as well as the Co(I)Al analogue 819, were independently synthesized (Scheme 110). Spectroscopic characterization of 818(η²-H₂) and 819(η²-H₂) not only revealed the lability of the dihydrogen ligand but also the reversibility of its formation under hydrogen atmosphere/vacuum. The different oxidation states of the metal center in the dihydrogen complexes substantially affect the approximate d_HH bond distances of the coordinated η²-H₂ ligand inferred from NMR experiments (T_1 min relaxation times): 0.96 or 0.98 Å in the anionic Co(−I) complexes 815(η²-H₂) and 816(η²-H₂), respectively (vide supra) vs 0.85 or 0.86 Å in the dihydrogen-hydride species 819(η²-H₂) and 818(η²-H₂).⁵²⁵

NMR experiments provided further mechanistic details on the Co–Ga system. These indicated that different resting states operate, depending on the base used. Using the strong Verkade base, complex 816(η²-H₂) was identified as the resting state, while for weaker bases it was found to be the dihydrogen-hydride species 818(η²-H₂). These observations are reflected in the rate-determining step for the reaction. In the first case, hydride transfer from 816(η²-H₂) to CO₂ is rate-limiting, while for the weak base system this becomes the deprotonation of 818(η²-H₂).

DFT calculations gave additional insights and support an unusual Co(−I)/Co(I) redox reaction for the H₂-activation step (see Figure 18), in which a dihydride complex is formed. 816(η²-H₂) thus acts as a masked Co(I) dihydride, sufficiently hydridic (according to the thermodynamic considerations and DFT calculations) to transfer to CO₂. Another crucial conclusion from the theoretical studies is that the release of formate is assisted by the anionic charge on the formate adduct (Figure 18, species 820). The deprotonation of 818(η²-H₂) to regenerate the catalytically active species 816(η²-H₂) is initially promoted by the presence of the strong Verkade base, and then sustained by formate acting as a proton shuttle, as previously observed for the Ni–Ga analogous catalyst 763.⁵²³ DFT calculations also support mechanistic proposals different from that portrayed in Figure 18, albeit ones which would involve an overall four-electron redox cycle, involving Co(III) intermediates.⁵²⁵

Figure 18.

One of the mechanistic proposals for the hydrogenation of CO₂ to formate mediated by 816(N₂).⁵²⁵

3.1.6. Polysilane Hydrogenation Catalyzed by Ni(0) Hydride Complexes

Hydrogen can also be used to cleave chemical bonds via hydrogenolysis.⁵³⁰ Trincado and Grützmacher showed that the Ni(I) complexes [Ni(TFA)(trop₂NR)] [trop₂NR, R = H (821), Me (822); trop = 5Hdibenzo[a,d]cyclohepten-5-yl; TFA = trifluoroacetate] react with different hydrogen sources under basic conditions (Me₂NHBH₃ (DMAB)/KOtBu or LiAlH₄ or H₂/KOtBu) to afford the Ni(0) compounds M[NiH(trop₂NR)] [M = K, R = H (823); M = Li, R = Me (824)] or K[(trop₂NR)Ni(μ-H)Ni(trop₂NR)] [R = H (825), Me (826)] (Scheme 111a). Compound 826 reacted with Ph₂SiH₂ to give [Ni(trop₂NMe)(η²-HSiHPh₂)] (827). The latter species reacted with a base (K[Me₂NBH₃]) to afford the anionic silyl complex [Ni(SiHPh₂)(trop₂NMe)]⁻ (828).

Scheme 111. Anionic Ni(0)-Hydride Complexes Featuring the Amino-bis(olefin) Ligands trop₂NR and Their Use as Catalysts for the Hydrogenolysis of Polysilanes⁵³⁰.

Complexes 824–826 were evaluated as catalyst precursors for the hydrogenolysis of polysilanes. The catalytic hydrogenolysis of the oligosilanes H(SiPh₂)_nH using the anionic complexes 824–826 (0.03 equiv K[Me₂NBH₃], r.t., 1 bar H₂) quantitatively affords the monomer SiH₂Ph₂. The Ni(I) compound 822 was also a competent catalyst for this reaction, albeit affording lower yields of product (79%, see Scheme 111b). In addition, the authors demonstrated that the trifluoroacetate complex 822 transforms into the dinuclear compound 826 in the presence of molecular hydrogen and a base. In general, the hydrogenolysis of different polysilanes (5 examples, Scheme 111b) returned good-to-excellent yields of product and could be scaled up (for methylated silanes) affording over 80% isolated yield of Me₂SiH₂. Catalyst poisoning tests with Hg or PPh₃ indicated that a homogeneous system might be operative. DFT calculations suggested that the Ni(0)-hydride anion, in which one olefin of the trop₂NR ligand has decoordinated, serves as a catalytic intermediate.

The presence of an electron-rich metal center able to promote both σ-bond metathesis and oxidative addition reactions and the double hemilabile ligand creating a vacant site appear to be key features responsible for the high catalytic activity of the anionic d¹⁰-[Ni(SiHPh₂)(trop₂NR)]⁻ (R = H, Me; X = H, SiR₃) complexes.

3.1.7. Transfer Hydrogenation, Hydroboration, and Hydrosilylation

In addition to the activation of molecular hydrogen, transfer hydrogenation³⁸⁶ and other hydrofunctionalization^376,531 reactions mediated by metalate complexes have been studied. For instance, Xu, Cui, and co-workers reported a system for the efficient hydrosilylation of carbonyl compounds catalyzed by the ferrate complex [(η⁵-Cp)Fe(CO)₂]⁻, as the potassium or [NEt₄]⁺ salts (K[2] or [NEt₄][2]).^13,531 An initial screening of the reaction conditions revealed that both K[2] and the more soluble [NEt₄][2] facilitate the hydrosilylation of acetophenone (catalyst loading = 0.5 mol %) with PhSiH₃ as reductant (Scheme 112). No difference in activity was observed between both catalysts, suggesting that the cation does not play a significant role in this system, beyond improving the solubility of the complex.

Scheme 112. Hydrosilylation of Ketones and Aldehydes Catalyzed by [(η⁵-Cp)Fe(CO)₂]⁻⁵³¹.

In the case of the aromatic ketones/aldehydes (10 examples, Scheme 112), all three Si–H bonds of PhSiH₃ participate in the reaction, converting 3 equiv of substrate. For aliphatic substrates, the product corresponds to the doubly substituted bis(alkoxy)silane (9 examples, Scheme 112). The catalyst tolerates a wide range of functional groups, and the system is chemoselective. Quantitative conversions were achieved under the optimized reaction conditions [neat, 0.5 mol % catalyst, room temperature]. At catalyst loadings as low as 0.02 mol %, catalyst [NEt₄][2] converted 82% of acetophenone in 10 min, a TOF of 24540 h^–1. Furthermore, the system could be scaled up to 100 mmol of substrate without a loss in efficiency. In search of mechanistic information, [NEt₄][2] was individually treated with acetophenone or PhSiH₃. However, no reaction was observed under the reaction conditions used. Nonetheless, the authors suggested that a possible reaction pathway might involve the oxidative addition of the hydrosilane to ferrate 2, to yield a highly reactive iron hydride intermediate as a catalytically active species. No further mechanistic evidence was provided.⁵³¹

Jacobi von Wangelin and Wolf reported an efficient system for the dehydrogenation of a variety of amine-boranes and the transfer hydrogenation of challenging olefins, imines and N-heteroarenes.³⁸⁶ Previous studies had found that the precursors [M(η⁴-C₁₄H₁₀)₂]⁻ [M = Fe (52); Co (14)] and [Co(η⁴-cod)₂]⁻ (239) performed poorly for the hydrogenation of trisubstituted alkenes and for the dehydrogenation of amine-boranes.^246,497 In search for more efficient precatalysts, the group turned to the complexes [K(thf)_n{Co(^ArBIAN)(η⁴-cod)}] (540: n = 1.5, Ar = Dipp; 541: n = 1, Ar = Mes; see Scheme 65 for the synthetic procedure), which incorporate the redox-active ^ArBIAN ligand (section 2.2.10.1, vide supra). These highly reduced cobaltate anions served as precatalysts for the dehydrogenation of ammonia borane (NH₃BH₃, AB) and related amine-boranes under mild conditions. Complex 540 promoted the dehydrogenation of dimethylamine-borane (Me₂NHBH₃, DMAB), yielding a mixture of polyaminoborane, borazine and polyborazine. The mixture of products indicated that the system released more than one equivalent of H₂ from the amine-borane. The evidence gathered from reaction monitoring and poisoning experiments indicated that the catalytic species might be homotopic. Additionally, the authors found that 540 effectively catalyzes the transfer hydrogenation of olefins, imines, and quinolines using NH₃BH₃ as a dihydrogen surrogate (Scheme 113a). The mass balance indicated that up to 2 equiv of H₂ could be transferred from AB.³⁸⁶

Scheme 113. Transfer Hydrogenation, Hydrogenation, and Hydroboration Reactions Catalyzed by the Highly Reduced Complexes [K(thf)_n{Co(^ArBIAN)(η⁴-cod)}] (540/541)³⁸⁶ and [K([18]crown-6)(thf)_0.5][(^DippBIAN)Fe(η⁴-cod)] (537)³⁷⁶.

Mechanistic studies and poisoning tests (Hg test or addition of P(OMe)₃ or dct) suggest that the reaction occurs via a homogeneous catalyst. Moreover, the experiments revealed that in the transfer hydrogenation the rate-determining step probably involves a proton transfer from the amine-borane. Given the second-order rate law obtained with respect to cobalt, these experiments also suggest that a species composed of more than one cobalt atom is operative. Good conversions were observed for the hydrogenation of alkenes from AB using complex 541 as a precatalyst (Scheme 113b). In this case, challenging trisubstituted olefins were transformed via a different mixed protocol, involving activation of the catalyst 541 by AB (used as an additive), with subsequent hydrogenation using molecular hydrogen (10 bar).³⁸⁶

Wolf and co-workers, furthermore, reported an efficient iron-catalyzed hydroboration of ketones, using low-valent complexes featuring the redox-active ligand BIAN as precatalysts.³⁷⁶ Cyclic voltammetry showed high reduction potentials for complexes [(^DippBIAN)Fe(L)]⁻ (L = η⁴-cod, bis(norbornyl), ^DippBIAN, 537–539; see Scheme 64 above), which were therefore anticipated to be good candidates to perform reductive transformations. In fact, under mild conditions [0.1 mol % catalyst load, small excess HBpin (1.03 equiv), 25 °C, THF, 10 min], both 537 and 538 returned quantitative conversions (>99%) of the model substrate acetophenone, though the homoleptic complex 539 afforded only moderate yields of product (67%). This result was attributed to the lack of vacant coordination sites in the precatalyst 539.

Investigations of the scope of the system (19 examples, see Scheme 113) were carried out under the optimized conditions using complex [(^DippBIAN)Fe(η⁴-cod)]⁻ (537), finding that the system efficiently transforms substituted acetophenones, with either electron-donating or electron-withdrawing groups. Sterically demanding substrates were also reduced, requiring only longer reaction times. Furthermore, the system was selective for halogenated acetophenones (no observable dehalogenation), and for α,β-unsaturated substrates (no reduction of C=C bonds). Monitoring of the model reaction (hydroboration of acetophenone) using in situ IR spectroscopy revealed full consumption of the substrate in only 30 s with a catalyst loading of 0.05 mol %, corresponding to a turnover frequency of 220 000 h^–1. Further mechanistic experiments included a poisoning test using mercury, where no inhibition was observed, and individual reactions between complex 537 and the substrate or HBpin. While the reaction of 537 with HBpin did not lead to new signals in the ¹¹B NMR spectrum, addition of acetophenone to 537 gave an immediate color change, along with disappearance of the ν_C=O stretching band of acetophenone from the IR spectrum. Thus, it was concluded that 537 acts as a precatalyst in this system.³⁷⁶

The nickel(0) complexes [(κ²-C,N-^ArNHCPyO)Ni(cod)]⁻ by Kennedy and co-workers bearing bidentate NHC–pyridone ligands (Scheme 50, section 2.2.7, vide supra) catalyze the hydroboration of styrene by HBpin (Scheme 114).³²⁴ Complexes 412–414 proved to be competent precatalysts, affording quantitative conversion of the olefin with >70% selectivity toward the branched product with 5 mol % catalyst loading at slightly elevated temperatures (50 °C) in toluene. Under the same conditions, the tricoordinate complex [(κ²-C,N-^DippNHCPyO)Ni(η²-MeCN)]⁻ (417) was also an efficient catalyst, yielding 70% of the branched product. Only negligible amounts of the dehydroboration product were detected.

Scheme 114. Hydroboration of Styrene Catalyzed by Nickel(0) Complexes [(κ²-C,N-^ArNHCPyO)Ni(cod)]⁻ Bearing Bidentate NHC–Pyridone Ligands³²⁴.

Using catalyst [(κ²-C,N-^MesNHCPyO)Ni(cod)]⁻ (412), efficient hydroboration was reported for trans-β-methylstyrene (80% yield branched product), while for α-methylstyrene the reaction gave poor results in terms of yield and selectivity (16% yield, branched:linear = 58:42) probably due to increased steric hindrance at the benzylic position of the substrate.

For comparison, the hydroboration of styrene was carried out using the analogous neutral Ni(0) compound [(κ²-C,N-^MesNHCPy)Ni(cod)] (829), featuring an unsymmetrical NHC-pyridine ligand. Complex 829 converted styrene much less efficiently (60%, 3% yield of branched product) and selectively (branched:linear:dehydroboration product ratio = 31:46:23) than 412. The superior catalytic properties of the NHC-pyridone complexes likely result from Lewis acid–base interactions between HBpin and the pyridone O atom.³²⁴

3.1.8. Dehydrogenation/Dehydrogenative Coupling of (Alkyl)amine–Boranes

Grützmacher and Lichtenberg reported the use of the low-valent Fe(I)-trop₂dae complexes [M(solv)_n{Fe(trop₂dae)}] [M = Li, solv = Et₂O, n = 2 (586); M = Na, solv = thf, n = 3 (587), see section 2.2.10.4] as precatalysts in the dehydrogenation of dimethylamine-borane (Me₂NHBH₃, DMAB).³⁸¹ Compound 587 does not react with H₂ in nonpolar solvents (1.5 bar, T = 25 °C), but dehydrogenates DMAB efficiently at room temperature (>99% conversion, 5 mol % precat, 4 h, open system; see Scheme 115a). Under the same conditions, 586 showed only moderate activity (35%, 4 h). The marked difference in performance of both catalysts indicates a strong influence of the contact ion pair. This was evidenced by performing catalytic tests in the presence of additives such as [15]crown-5 or [(nBu)₄N]Br which decreased the reaction rate, suggesting an important role of ion-pairing in the mechanism. Other mechanistic experiments, such as selective poisoning of the catalyst (0.2 equiv of PPh₃ or 0.1 equiv of P(OMe)₃) or analysis of small aliquots of reaction solutions by scanning electron microscopy (SEM), support a homogeneous mechanism for the dehydrogenation reactions, although no mechanistic proposal was reported. Complex 587 also catalyzes the dehydrogenative alcoholysis of silanes to produce oligo/poly(silyl ethers) from polyols and silanes in an efficient manner, under mild conditions (3 mol % precatalyst, room temperature).³⁸¹

Scheme 115. Catalytic Dehydrogenation/Dehydrogenative Coupling of (Alkyl)amine–Boranes Promoted by Fe(I)-dad Complexes^380,381.

Similarly, an efficient dehydrogenative coupling (dehydrogenative polymerization) of (alkyl)amine–boranes was reported by the same group.³⁸⁰ The complexes [Fe(trop₂dad)(L)] [L = thf (582), MeCN (583)] were tested as precatalysts for the dehydrogenative polymerization of methylamine–borane (MAB) using a catalyst loading of 5 mol % (toluene, 23 °C, open system; Scheme 115b). Both catalysts show remarkable activity in the dehydrogenative coupling of MAB, giving full conversion after a reaction time of 8.5 min, evidenced by the release of 1 equiv of H₂. Analysis of the polymeric material obtained (poly-MAB, analyzed by mass spectrometry) indicated that it consisted of at least 22 repeating units. A turnover frequency (TOF) of 5.1 × 10^–2 s^–1 was reported, reflecting the high activity of the catalytic system. A short induction period was observed for the reactions catalyzed by either of the complexes, thus suggesting that these are acting as precatalysts. Mechanistic tests, such as varying the solvent and poisoning experiments, indicate that a homogeneous catalytic regime might operate. Good conversions (5 mol % of precatalyst 582, 23 °C, open system; see Scheme 115b) were also reported for the dehydrogenative polymerization of ammonia-borane H₃N·BH₃ (AB) or the cyclodimerization of dimethylamine–borane (DMAB), although the excellent activity observed in the conversion of MAB was not matched. The complexes [{Na(thf)₃[Fe(trop₂dad)]}] (579) and [Na(thf)₃{Fe(trop₂dae)}] (587) were also evaluated as potential catalysts for dehydrogenative polymerization reactions, with the former being only moderately active, and the latter species being inactive.³⁸⁰

3.2. Nitrogen Activation and Functionalization

Nitrogen fixation by transition metal complexes is currently.^{25−27,532−543} Coordination of N₂ to one or several low-oxidation state metal centers is typically required for subsequent NN bond cleavage and conversion of N₂ into ammonia (NH₃) or other nitrogen compounds. Due to their electron-rich nature, low-valent transition metalate complexes are capable of strongly activating N₂ and mediating subsequent transformations of the N₂ ligand.

3.2.1. General Comments on N₂ Activation by Metalate Complexes

The complexes M[(N₂)Co(PMe₃)₃], [M = Li (830), Na (831), K (832)], (Et₂O)RMg[(N₂)Co(PMe₃)₃] (833, R = CH₃, CH₂CH(CH₃)₂, C(CH₃)₃, C₆H₅) and Me₂Al[(N₂)Co(PMe₃)₃] (834) reported by Klein and Huttner represent early examples of metalates containing dinitrogen ligands.^543,544 In the solid state, the potassium salt K[Co(N₂)(PMe₃)₃] (832) features an end-on coordinated N₂ ligand and forms a hexamer through additional interactions of N₂ with the potassium cations, while the magnesium and aluminum compounds have dimeric N₂-bridged structures.⁵⁴³ Furthermore, the protonation of coordinated dinitrogen ligands at electron-rich ferrates and cobaltates was described as early as 1983. Reactions of compounds M[(N₂)Co(PPh₃)₃] (830–832), [Mg(thf)₄][(N₂)FeEt(PPh₃)₂]₂ (835), or [Mg(thf)₄][(N₂)Co(PPh₃)₃]₂ (836) with H₂SO₄ afforded small amounts of hydrazine and ammonia (0.11 mol −0.31 mol per mol of TM).⁵⁴⁵ A more recent example showed that coordinated dinitrogen ligands at the complexes [MgCl(thf)₂][{PhB(CH₂PiPr₂)₃}Fe(N₂)] (837) or [Mg(thf)₄][{PhB(CH₂PiPr₂)₃}Co(N₂)]₂ (838) react with electrophiles to produce diazenido complexes.⁵⁴⁶

The degree of N≡N bond activation in transition metal complexes can be readily identified through the ν_N≡N band observed by FT-IR spectroscopy. Selected ν_N≡N stretching frequencies can be found in Table 2. For example, it is possible to observe similarities between the stretching frequencies for the free N₂ (g) molecule,⁵⁴⁷ and that for Fe(N₂)(CO)₂(CNAr^Tripp2)₂ (69)¹⁴⁷ or (N₂)Co(SiMe₃)(CNAr^Mes2)₃ (98)¹²⁵ in which the dinitrogen ligands exhibit a low degree of activation (ν_N≡N approximates to the value of free N₂ (g); see Table 2). By contrast, changes in the degree of activation of a coordinated N₂ ligand become evident by comparing the stretching frequencies of the neutral complex [(^AdP^PymDI)Fe(N₂)] (551), and that of its reduced analogue [Na([18]crown-6)(thf)₂][(^AdP^PymDI)Fe(N₂)] (558),³⁹¹ or the series of neutral, anionic and dianionic complexes [(P₃^B)Fe(N₂)]^0/1–/2– (0: 719, 1–: 720, 2–: 723; see Table 2).^139,484 The increase in backbonding from the more electron-rich metal center causes a shift in the ν_N≡N band to lower values, and indicates a higher degree of activation.³⁹¹

3.2.2. N₂ Activation and Functionalization at Transition Metal-Phosphine Complexes

In 2019, Zhang and Xi reported that low valent anionic species with highly activated dinitrogen ligands were obtained upon reduction of neutral mononuclear chromium or dinuclear chromium-dinitrogen complexes featuring multisubstituted cyclopentadienyl-phosphine ligands.⁵⁵³ Reduction of either the mononuclear Cr(II) precursor [{o-(C₆H₄PCy₂)C₅Et₄}CrCl] (839) or the dinuclear Cr(I) complex [{o-(C₆H₄PCy₂)C₅Et₄}Cr(N₂)]₂(μ-N₂) (840) by alkali metals allowed isolation of the anionic Cr(0) compounds [M([2.2.2]cryptand)][{o-(C₆H₄PCy₂)C₅Et₄}Cr(N₂)₂] [M = K (841), Rb (842), Cs (843); see Scheme 116]. These diamagnetic complexes have similar features in the solid state, with the main difference being that the alkali metal cations coordinate one or four N atoms in 842 or 843, respectively, whereas in the potassium metalate 841 no interaction was observed. The molecular structures show Cr–N bond lengths in the range of single bonds (1.820 to 1.834 Å) and slightly activated N₂ ligands (d_NN = 1.132 to 1.151 Å; ν_N≡N = 1822 to 1907 cm^–1).

Scheme 116. Synthesis of Anionic Cr(0)-Dinitrogen Complexes and N₂-Functionalization to Generate a Chromium Hydrazido Compound⁵⁵³.

Preliminary N₂ functionalization tests were performed on 841 with acids [HBAr^F₄·2Et₂O, (LutH)X (X = Cl, OTf; Lut = 2,6-lutidine)] or with Me₃SiCl. In the protonation reactions, the authors detected small amounts of ammonia or hydrazine (∼5% based on N atoms). In turn, the silylation of 841 with Me₃SiCl (1 equiv) generated the bis(silyl)hydrazido complex [{o-(C₆H₄PCy₂)C₅Et₄}Cr=NN(SiMe₃)₂] (844) along with the Cr(II) species 839. The monosilylated product was not observed. The structural differences observed by crystallographic analysis of 844 compared to 841 evidence the formation of a hydrazido motif: the Cr–N distance is in the range of a M–N double bond (1.680 Å) while the N–N distance is much longer than in the anionic complex (1.372 Å). Since hydrazido complexes have been proposed as intermediates in the catalytic reduction of N₂, the authors evaluated the performance of the series of synthesized complexes in the silylation of N₂ to N(SiMe₃)₃. The most active precatalyst was the neutral complex [{o-(C₆H₄PiPr₂)C₅Et₄}CrCl] (845), an analogue of 839, which afforded 26 equiv of amine product per Cr (2000 equiv Me₃SiCl, 2000 equiv K as reductant).⁵⁵³

Nitrogen functionalization has been achieved at cobalt complexes with chelating phosphine ligands.^554,555 In 2019, Miller and Long described Co(−I)-dinitrogen complexes featuring two examples of the triphos family of tripodal ligands, specifically N-triphos and C-triphos.⁵⁵⁴ Two-electron reduction of the Co(I) precursors [(triphos)CoCl] [triphos = N(CH₂PPh₂)₃ (846), MeC(CH₂PPh₂)₃ (847)] with Mg powder in THF under a nitrogen atmosphere afforded the Co(−I) complexes [Mg(thf)₄][{E(CH₂PPh₂)₃}Co(N₂)]₂ [E = N (848), MeC (849); see Scheme 117, top], analogous to the aforementioned [Mg(thf)₄][(N₂)Co(PPh₃)₃]₂ (836) or [Mg(thf)₄][{PhB(CH₂PiPr₂)₃}Co(N₂)]₂ (838), among others.^545,546,556

Scheme 117. N₂ Activation and Functionalization at (Triphos)Co(−I)-Dinitrogen Complexes⁵⁵⁴.

848 and 849 coordinate dinitrogen between the cobalt centers and the stabilizing Mg²⁺ cation in an end-on bridging fashion. The ligands show a significant degree of activation, as evidenced by the FT-IR data [ν_N≡N(848) = 1878 cm^–1; ν_N≡N(849) = 1872 cm^–1] and the elongated N–N bond distances in the molecular structure of 848 (1.157(6) and 1.156(7) Å). Metathesis of the counterion with alkali metal cations in both cases yielded the corresponding ion-paired mononuclear complexes [{E(CH₂PPh₂)₃}Co(μ-N₂)M] [E = N (850), MeC (851); M = Li(thf)_x, Na(thf)_x, K(thf)_x, K([18]crown-6); Scheme 117]. Additionally, 848 and 849 reacted with Me₃SiCl (2.5 equiv) in THF, affording products identified as the silyldiazenido derivatives [{E(CH₂PPh₂)₃}Co(NNSiMe₃)] [E = N (852), MeC (853); Scheme 117], on the basis of the lower frequency NN stretching bands [ν_NN(852) = 1678 cm^–1; ν_NN(853) = 1672 cm^–1] and ²⁹Si NMR spectroscopic characterization.⁵⁵⁴

In 2021, another contribution reported that dinitrogen coordinates to Co(−I) centers supported by phosphine ligands, specifically the chelating diphosphine 1,2-bis(dicyclohexylphosphino)ethane (dcpe).⁵⁵⁵ Reduction of the Co(II) complex [(dcpe)CoCl(N–P)] [854, N–P = (3,5-Me₂-C₆H₃)NPiPr₂] by an excess of alkali metal (Na, K or Cs) under a nitrogen atmosphere gave the anionic compounds [M(solv)][(dcpe)Co(N₂)₂] [M = Na (855), K (856), Cs (857); Scheme 118]. Compounds 855–857 were also accessed via sequential reduction of the Co(II) precursor, first with KC₈ to afford a Co(I) intermediate 858, which was further reduced with alkali metals under a nitrogen atmosphere (3 equiv, Scheme 118).

Scheme 118. Synthesis of (Diphosphine)Co(−I)-Dinitrogen Complexes and Silylation of Coordinated N₂-Ligands Yielding Diazenido Motifs⁵⁵⁵.

Crystallographic characterization of 855–857 confirmed the coordination of two N₂ molecules to the cobalt center, in a rare example of cobalt complexes with more than one dinitrogen ligand. In all cases, the N–N bond lengths are longer and the IR stretching bands are shifted to lower frequencies (e.g., for 856, dNN = 1.128(3)–1.150(3) Å, ν_N≡N = 1837 and 1931 cm^–1) than those in the free N₂ molecule, indicating significant activation. Treating 855–857 with [2.2.2]cryptand afforded the [M([2.2.2]cryptand)]⁺ analogues, [M([2.2.2]cryptand)][(dcpe)Co(N₂)₂] [M = Na (859), K (860)] and [(dcpe)Co(N₂)₂ Cs([2.2.2]cryptand)] (861). In the Cs compound, the counterion is coordinated to both dinitrogen ligands. Compared to 855–857, the encapsulation of the cation by [2.2.2]cryptand causes a decrease in the degree of activation of the N₂ units. Silylation of 856 with iPr₃SiCl (1.5 equiv) resulted in the diazenido complex [{(dcpe)Co(NNSiiPr₃)}₂(μ-N₂)] (862), which was crystallographically characterized. In the diazenido complex, the N–N bond distances of the functionalized motifs are in the range of typical N=N double bonds [1.2080(18) Å - 1.2051(17) Å]. FT-IR spectroscopy further supported the reduction of the dinitrogen ligands to diazenido units (ν_NN = 1696 cm^–1). Quantum chemical calculations revealed that 862 has an open-shell singlet ground state, and that the silyldiazenido fragment and the cobalt center exhibit delocalization due to a combination of σ-donation from N to the Co center and π-backbonding back to the diazenido moiety.

The Schneider group has studied N₂ splitting at pincer (PNP) rhenium complexes (PNP^– = N(CH₂CH₂PtBu₂)₂^–).^26,557,558 In 2014, the group established that the reductive transformation of dinitrogen by the Re(III) precursor [ReCl₂(PNP)] (863) to the Re(V) nitride complex [Re(N)Cl(PNP)] (864) occurs through an intermediate N₂-bridged dinuclear complex, [{(PNP)ClRe}₂(N₂)] (865), as supported by DFT calculations.⁵⁵⁸ The group subsequently revisited and expanded this mechanistic proposal, including additional structural, spectroscopic, electrochemical, and kinetic data which pointed toward the participation of an anionic Re(I) species as an intermediate. Compound 865 was isolated and characterized, and was found to feature a moderately activated bridging N₂ ligand.⁵⁵⁷ Electrochemical and computational investigations indicated that 863 coordinates the N₂ ligand and undergoes two-electron reduction to generate an anionic Re(I) complex, [ReCl(N₂)(PNP)]⁻ (866). This undergoes comproportionation with further 863 to give the bridged N₂-complex 865, with concomitant chloride loss. Furthermore, computational calculations indicate that prior to the Re(I)/Re(III) comproportionation process, no significant N₂ activation occurs. In the electrochemically driven reaction, the formation of the bridged complex 865 is irreversible.

3.2.3. N₂ Activation and Functionalization by N-Heterocyclic Carbene Complexes

Very recently, Wei, Xi, and co-workers described the functionalization of N₂ at a low-valent anionic chromium complex featuring NHC and Cp* ligands.⁵⁵⁹ The bis(dinitrogen) compound [K([2.2.2]cryptand)][Cp*Cr(IiPr₂Me₂)(N₂)₂] (867, IiPr₂Me₂ = 1,3-diisopropyl-4,5-dimethylimidazol-2-yildene) formed either directly, by reaction between the Cr(II) precursor [Cp*CrCl(IiPr₂Me₂)] (868) with KC₈ (3 equiv) in the presence of [2.2.2]cryptand, or by stepwise reduction of 868 (1.2 equiv KC₈), forming the dinuclear compound [{Cp*Cr(IiPr₂Me₂)}(μ-N₂){Cp*Cr(IiPr₂Me₂)(N₂)}] (869), which is further reduced (3 equiv KC₈) to yield 867. Anionic 867 features significantly activated dinitrogen ligands (ν_NN = 1760 and 1846 cm^–1; d_NN = 1.153(2) Å and 1.1425(19) Å) according to spectroscopic and crystallographic characterization. To functionalize the N₂ ligand, the reactivity of in situ generated 867 was evaluated. Different products were obtained depending on the bulkiness of the silyl reagent used in the silylation of 867. Whereas reaction between 867 and iPr₃SiCl (1.1 equiv) led to the formation of the chromium diazenido complex [Cp*Cr(IiPr₂Me₂)(NNSiiPr₃)] (870; Scheme 119), the use of the less bulky reagent Me₃SiCl afforded the chromium side-on η²-hydrazido species [Cp*Cr(IiPr₂Me₂)(η²-Me₃SiNNSiMe₃)] (871) as the major product, along with the Cr(II) complex 868. Crystallographic characterization of the diazenido and hydrazido derivatives confirmed the significant reduction of the N–N bond order to values in the range of typical double (d_NN = 1.243(2) Å) or single (d_NN = 1.4671(15) Å) bonds, respectively.

Scheme 119. N₂ Functionalization at the Anionic Chromium Complex [Cp*Cr(IiPr₂Me₂)(N₂)₂]⁻ (867)⁵⁵⁹.

867–871 were evaluated as catalysts in the silylation of N₂ to produce N(SiMe₃)₃, finding that the complexes were active, producing 2.0 to 5.9 equiv of the silylated amine in 24 h (1000 equiv KC₈, 1000 equiv Me₃SiCl). Remarkably, the authors anticipated a nucleophilic character for complex 871, confirmed by reaction with unsaturated substrates such as CO₂ and tBuNCO. In the first case, two molecules of CO₂ inserted into the N–Si and Cr–N bonds of the coordinated hydrazido unit, yielding a new N,O-chelating hydrazidochromium complex, [Cp*Cr(IiPr₂Me₂){κ²-N,O-OC(Me₃Si)NNC(O)OSiMe₃}] (872). Similarly, reaction with tBuNCO yielded the analogous product of insertion [Cp*Cr(IiPr₂Me₂){κ²-N,O-OC(NtBu)N(SiMe₃)N}] (873). Both compounds were crystallographically characterized.⁵⁵⁹

The potential of metalates to mediate N₂ functionalization reactions is illustrated in recent work by Peters and co-workers, in which the syntheses of complexes [K([18]crown-6){(N₂)Fe(CAAC)₂}] (392)⁷⁶ and [(P₃^B)Fe(N₂)]⁻ (720)¹³⁹ were reported (see sections 2.2.7 and 2.3.3.1, respectively). According to the FT-IR data, both complexes feature significantly activated dinitrogen ligands (see Table 2). The functionalization of these species with silyl chlorides was then achieved.^76,426 By treating the carbene complex 392 with trimethylsilyl chloride (Me₃SiCl; see Scheme 120a) a rapid color change, accompanied by the appearance of a broad band in the FT-IR spectrum (ν_N≡N = 1675 cm^–1), were suggestive of functionalization of the N₂ molecule. However, the complex that presumably resulted, [(CAAC)₂Fe(N=NSiMe₃)] (874), decomposed rapidly to [Fe(CAAC)₂] (385). The use of a bulkier chloro-silane reagent, triethylsilyl chloride, gave access to the diazenido compound [(CAAC)₂Fe(N=NSiEt₃)] (875, ν_N≡N = 1690 cm^–1). Peters and co-workers furthermore showed that systems reactive toward silylation at the β-nitrogen of Fe–N₂ complexes might be susceptible to efficient reductive protonation.^139,428,560 Therefore, the anionic complex 392, the related neutral compound [Fe(CAAC)₂] (385) and its cationic precursor [Fe(CAAC)₂]BAr^F₄ (390, see Scheme 46a, above) were tested as catalysts for nitrogen fixation.

Scheme 120. (a) Silylation of the Coordinated N₂-Ligand in Complex [K([18]crown-6){(N₂)Fe(CAAC)₂}] (392) and (b) Low-Temperature Nitrogen Fixation Using (CAAC)₂Fe Complexes as Catalysts⁷⁶.

At room temperature, 385 is unable to coordinate N₂ and performed poorly in the presence of excess KC₈ and HBAr^F₄·2Et₂O in diethyl ether. Conducting the reaction at −95 °C, however, afforded 3.3 equiv of NH₃ per equivalent of iron (see Scheme 120b). A similar performance was reported for the cationic compound 390 and slightly lower yields were obtained with the anionic precursor 392, under the same conditions. Even at temperatures as low as −78 °C, the performance of the catalyst was rather poor. Compound 385 was also evaluated as a catalyst in N₂ silylation, in the presence of a large excess of both KC₈ and trimethylsilyl chloride (600 equiv of each), leading to the formation of 24.4 ± 2.7 equiv of N(SiMe₃)₃.⁷⁶

Deng and co-workers reported NHC-stabilized cobalt-dinitrogen complexes, and their functionalization to yield coordinated diazene compounds.⁵⁴⁸ Reduction of the cobalt(I) complex [(ICy)₃CoCl] (876, ICy = 1,3-dicyclohexylimidazol-2-ylidene) by KC₈ under a N₂ atmosphere afforded the Co(0)–N₂ complex [(ICy)₃Co(N₂)] (877), featuring an end-on dinitrogen ligand (Scheme 121a). The low energy N–N stretching frequencies observed (ν_N≡N = 1917 cm^–1, in KBr) indicate a moderate activation of the N₂ unit. Further reduction of 877 with alkali metals (AM = K, Rb, Cs; 1.2 equiv) yielded the series of bis(dinitrogen)cobalt(−I) complexes [AM(ICy)₂Co(N₂)₂] (AM = K[878], Rb[879], Cs[880]; Scheme 121a), obtained as 1D-coordination polymers. Treatment of K[878] with [18]crown-6 afforded [K([18]crown-6)(ICy)₂Co(N₂)₂] (K([18]crown-6)[878]), in which the [K([18]crown-6]⁺ cation coordinates one of the N₂ ligands. DFT calculations indicated that the dinitrogen ligands at the formally d¹⁰ Co(−I) center engage in substantial π-backbonding into the N₂ π* orbital, as reflected by the significantly red-shifted IR stretching vibration (ν_N≡N = 1807 cm^–1, in KBr). The Co(−I) complexes 878–880 reacted with anhydrous acids (triflic acid or HCl, –78 °C, Et₂O) to yield hydrazine (23–31% yields per cobalt atom) and trace amounts of ammonia. By contrast, analogous reaction with the neutral complex 877 gave only traces of hydrazine product. Attempts to isolate the intermediates in the protonation reactions were unsuccessful. Nonetheless, by treating K[878] with chlorosilanes (R₃SiCl), the cobalt diazene complexes [(ICy)₂Co(η²-R₃SiNNSiR₃)] [R = Me (881), Et (882), Scheme 121a] were formed.

Scheme 121. (a) NHC-Stabilized Cobalt-Dinitrogen Complexes or Their Functionalized Products and (b) Catalytic Silylation of N₂ by Low-Valent Cobalt⁵⁴⁸.

The solution magnetic moments of species 881 or 882 (approximately 3.0 μB) agree well with reported examples of low-spin cobalt(II)–NHC compounds. These observations are also consistent with the EPR characterization, which exhibits signals typical of low-spin square planar Co(II) complexes. The N–N distances of the dianionic diazene η²-R₃SiNNSiR₃ ligands indicate the presence of single bonds (881: 1.450(4) Å; 882: 1.457(3) Å). Compounds 876–882 afforded comparable yields of the silylated product N(SiMe₃)₃ (15–19%, relative to Me₃SiCl) in the silylation of dinitrogen under the following conditions (Scheme 121b): N₂ (1 atm), KC₈ (2000 equiv relative to catalyst) and Me₃SiCl (2000 equiv relative to catalyst), at room temperature. These yields correspond to TONs between 107 and 125. The catalytic activity of complexes 881 and 882 is particularly noteworthy, since diazene species have frequently been proposed as key intermediates in the alternating N₂ reduction pathway.^561−563 Independent reactivity tests with complex 881 provided further insights on this matter: While 881 failed to react with either further reductant (KC₈) or Me₃SiCl at room temperature, the presence of both reagents in excess amounts produces N(SiMe₃)₃ in 85% yield. Therefore, the diazene complex 881 was proposed to be a possible intermediate in the catalytic silylation of N₂ by the NHC-stabilized cobalt complexes.

3.2.4. N₂ Activation and Functionalization by Amido and β-Diketiminate Metalates

Schley, Tonks, and co-workers reported that the reduction of [Ti{N(SiMe₃)₂}₃] (883) by KC₈ in toluene affords the mixed-valent inverse sandwich complex [K(toluene)][{N(SiMe₃)₂}₂Ti(μ-C₇H₈)Ti{N(SiMe₃)₂}₂] (884, Scheme 122).⁵⁶⁴ This complex had previously been incorrectly identified as [Li(tmeda)₂][Ti{N(SiMe₃)₂}₂(μ,η²:η²-N₂)]₂ (885) featuring two bridging side-on N₂ ligands.⁵⁶⁵ The significant elongation of the C–C bonds (1.436(4)–1.449(4) Å) in the bridging toluene unit in 884 indicates that it corresponds to a C₇H₈^2–. The metric parameters and the solution magnetic moment (1.67 μB, overall S = 1/2) suggested that 884 is a mixed-valent Ti(II)/Ti(III) complex.

Scheme 122. Reassessment of the N₂ Activation by Low-Valent Ti-Amide Complexes: Evidence of the Formation of an Inverse Sandwich Complex Instead of a Bridged Side-on Dinitrogen Adduct⁵⁶⁴.

Theoretical modeling (DFT) revealed that a structure with an inverse sandwich arrangement is in better agreement with the data of 885 than a dinitrogen bridged compound. Attempts to synthesize the latter complex again from (tmeda)₂TiCl₂ (886) under the originally reported conditions led only to the formation of 884 and the end-on dinitrogen bridging compound [(tmeda)Ti{N(SiMe₃)₂}₂Cl]₂(μ-N₂) (887, Scheme 122), also described in the previous work.^564,565 In addition, Schley and Tonks examined the reduction of Ti(III) complexes under nitrogen, and could identify different Ti(IV)-nitride products.

Holland and co-workers have made remarkable contributions to nitrogen fixation chemistry through investigations into the use of low-coordinate iron and cobalt compounds.^{549,550,566−572} Among these are the dinuclear complexes AM₂[(nacnac)M(NN)M(nacnac)] [M = Fe, nacnac = CH[C(Me)N(2,6-iPr₂C₆H₃)]₂, AM = Na (888), K (889) Rb (890), Cs (891) or nacnac = CH[C(tBu)N(2,6-iPr₂C₆H₃)]₂, AM = K (892); M = Co, nacnac = CH[C(tBu)N(2,6-iPr₂C₆H₃)]₂, AM = Na (893), K (894)], featuring dinitrogen ligands with NN bond orders of two as evidenced by the N–N bond lengths and stretching frequencies (Scheme 123 shows the potassium ferrates and the sodium and potassium cobaltates).^{550,567,568,572}

Scheme 123. Dinuclear (a) Iron, (b) Cobalt, or Nickel Anionic Complexes Featuring Significantly Activated Bridging Dinitrogen Ligands^{550,567,568,573}.

The greatest weakening of the dinitrogen ligands was observed for the iron complex 892 (ν_N≡N = 1589 cm^–1) and for the cobalt compound 894 (ν_N≡N = 1599 cm^–1).^550,567 The solid-state geometry of the iron complexes 888–892 was affected by the nature of the alkali metal cation. Rotation about the FeNNFe core was observed, with greater rotation of the β-diketiminate ligand backbones with respect to one another accommodating increasing size of the counterion. The extent of N₂ activation was largely unaffected by these structural changes, and computational studies showed this was due to orbital mixing enabling similar backbonding into the π* orbitals of N₂ regardless of the orientation of the ligands.⁵⁷²

Remarkably, the reduction of the Fe(II) complex [(^Me3nacnac)Fe(μ-Cl)]₂] (895) by potassium under a nitrogen atmosphere afforded a tetrairon bis(nitride) complex [(^Me3nacnac)₃Fe₃(N)₂K₂(Cl)₂Fe(nacnac^Me3)] (896, ^Me3nacnac = MeC[C(tBu)N(2,6-iPr₂C₆H₃)]₂). 896 features an Fe₃N₂ core, in which three iron atoms cooperate with the potassium cations to fully cleave the N–N triple bond, promoting a six-electron reduction of dinitrogen. Based on the combined characterization data, the tetrairon complex was reported to be composed of two Fe²⁺ and two Fe³⁺ centers.⁵⁴⁹ The reactivity of 896 is significantly influenced by pairing interactions with the alkali metal counterions, as demonstrated, upon addition of [18]crown-6, by the formation of a mixed-valent Fe(II)/Fe(III) anionic triiron complex derived from the aforementioned Fe₃N₂ core.⁵⁶⁶ In this species, new N–H and Fe–C bonds are formed after intramolecular C–H bond activation.

The authors proposed that the ability to form and break bonds in such a compound can be regarded as an “alkali control” strategy, in which reactions of N₂-derived nitrides (e.g., N–C bond formation) can be promoted. This very interesting reactivity, however, involves compounds of iron in its most common oxidation states, and therefore, will not be further discussed herein. For details, we refer the reader to the original literature.^549,566

Limberg and co-workers described the dinuclear Ni(I) compound [(nacnac)Ni(NN)Ni(nacnac)] (897, nacnac = CH[C(tBu)N(2,6-iPr₂C₆H₃)]₂) which similarly coordinates dinitrogen in a bridging end-on fashion.⁵⁷³ The dinitrogen ligand is weakly activated, evidenced by the stretching vibration in the FT-IR spectrum (ν_N≡N = 2164 cm^–1). Compound 897 can be generated in situ by reduction of [(nacnac)NiBr] (898) with KC₈, and further reduced to yield the singly and doubly reduced complexes K[(nacnac)Ni(NN)Ni(nacnac)] (899) and K₂[(nacnac)Ni(NN)Ni(nacnac)] (900, see Scheme 123c). The two single-electron reduction events further activated the dinitrogen ligands, as indicated by their elongated N–N bond distances (Scheme 123, bottom). DFT calculations predicted that, for both 899 and 900, these reduction events affected the bridging N₂ ligand instead of the β-diketiminato units or the metal centers. Given the significantly elongated NN bond, the N₂ unit in 900 has a diazene character.

In another example of the role of the cation in the nature of the isolated metalate complex, Holland and co-workers described additional iron and cobalt complexes featuring linear M–NN–Mg–NN–M cores (M = Fe, Co).⁵⁷⁴ Reduction of precursors [(^tBunacnac)FeCl] (901), [(^Menacnac)Fe(μ-Br)]₂ (902), or [(^tBunacnac)Co(thf)] (903) by activated Rieke magnesium in THF afforded complexes [Mg(thf)₄][(^Rnacnac)M(μ-N₂)]₂ [M = Fe, R = tBu (904), M = Fe, R = Me (905), M = Co, R = tBu (906)]. These compounds share similar structural features in the central core with the (triphos)cobalt complexes [Mg(thf)₄][{E(CH₂PPh₂)₃}Co(N₂)]₂ [E = N (848), MeC (849)], described above.⁵⁵⁴ The iron complexes 904 and 905 are metastable, and could only be characterized in solution. A comparison of the stretching frequencies of 904–906 [ν_N≡N(904) = 1808 cm^–1, ν_N≡N(905) = 1818 cm^–1, ν_N≡N(906) = 1868 cm^–1) with those of the dinuclear complexes 892 (ν_N≡N = 1589 cm^–1) and 894 (ν_N≡N = 1599 cm^–1) evidence the striking difference between both types of compounds: While in all the complexes the dinitrogen ligands are significantly activated, N≡N weakening in the potassium metalates 892 and 894 is substantially more acute than in the trinuclear M–NN–Mg–NN–M compounds 904–906, which contain formally zerovalent transition metal centers. This effect was attributed to the different interaction established between the dinitrogen ligands and the cations, since in 892 and 894 the counterions bind the N₂ motifs in a side-on fashion, whereas in 904–906 these units are coordinated between the transition metal and magnesium cation in a bridging end-on mode.⁵⁷⁴

Later, Holland reported that an iron-based system composed of the diketiminato-iron(II) bromide complex, [(^Dippnacnac)Fe(μ-Br)]₂ (907, ^Dippnacnac = CH[C(Me)N(2,6-iPr₂C₆H₃)]₂), sodium, a crown ether, and trimethylsilyl bromide promotes the coupling of N₂ and unactivated arenes at low temperature, to afford aniline derivatives.⁵⁷⁰ The efficiency of the system relies on the ability of the catalyst to both coordinate the arene and dinitrogen and on the coordinated N₂ motif to undergo partial silylation. The proposed mechanism consists of a sequential activation of the arene substrate, functionalization of N₂ and migration of the aryl ligand to the functionalized N₂-ligand.

In initial studies, 907 was converted to the Fe(I) species [(^Dippnacnac)Fe·benzene] (908), which was reduced further by either KC₈ or Na in the presence of a crown ether ([18]crown-6 or [15]crown-5) to afford the Fe(0) complex [K([18]crown-6][(^Dippnacnac)Fe(η⁴-C₆H₆)] (K[909]) and the benzene C–H activation product [Na([15]crown-5][(^Dippnacnac)Fe(Ph)(H)] (Na[910]), respectively. Further studies revealed that the Fe(0) and Fe(II) species 909 and 910 are in an equilibrium in THF. In turn, Na[910] formally loses the hydride ligand as H·, affording the three-coordinate complex [Na([15]crown-5][(^Dippnacnac)Fe(Ph)] (Na[911]), which served as platform to explore N₂ binding. Under a nitrogen atmosphere at low temperature, 911 coordinates N₂ to afford [(^Dippnacnac)Fe(N₂)(Ph)]⁻ (912, Scheme 124, top). Silylation of the coordinated N₂ ligand by Me₃SiBr (2 equiv) at low temperature produced the arylhydrazido complex [(^Dippnacnac)FeN(Ph)N(SiMe₃)₂]⁻ (913), in which migration of the phenyl ligand to the functionalized dinitrogen unit has occurred. Performing the silylation of 912 with the bulkier iPr₃SiOTf, afforded the singly silylated product [(^Dippnacnac)Fe(Ph){NN(SiiPr₃)}]⁻ (914) and the Fe(II) complex [(^Dippnacnac)Fe(Ph)] (915). 914 appears to be an initial stage in the silylation of N₂ before migration of the phenyl ligand. Addition of an excess of the silylation reagent and reductant to 913 or 914 afforded the aniline derivative (Scheme 124, top). The presence of a large excess of bromide from Me₃SiBr then promotes the regeneration of 907. Based on these insights, a synthetic cycle for the coupling of N₂ and arenes at iron was developed.⁵⁷⁰

Scheme 124. Binding of N₂ to the Fe(I) Species [Na([15]crown-5][(^Dippnacnac)Fe(Ph)] (Na[911]) and Functionalization of Coordinated N₂ (Top) And Product Distribution in the Iron-Mediated Coupling of Arenes with N₂ (Bottom)⁵⁷⁰.

Yields shown are relative to iron atom.

To ensure that all reactions are compatible, the iron(II) complex 907 was treated with Na (30 equiv/Fe), benzene (20 equiv), and [15]crown-5 (5 equiv) at room temperature, followed by cooling the solution to −108 °C, and adding Me₃SiBr (6 equiv). This protocol produced C₆H₅N(SiMe₃)₂ in 24% yield per iron atom. Slight modification enabled the development of a cyclic procedure in which the addition of the reagents is performed at the proper temperature for either C–H activation or nitrogen binding and silylation. After addition of Me₃SiBr, the solution is allowed to warm to room temperature to regenerate 911, evidenced by a color change. Further cooling–silylation–warming cycles, with addition of 2 equiv of Me₃SiBr in each cycle, provided the formation of the aniline derivatives in higher cumulative yields, although the yield of N(SiMe₃)₃ increases with each cycle, and the yield of the aniline derivative attenuates. Scheme 124 (bottom) shows the product distribution in the coupling of three arene substrates with N₂ after 5 cycles of Me₃SiBr addition (2 equiv/cycle).

The proposed reaction mechanism for the transformation of arenes and dinitrogen to anilines can be represented by the simplified cycle depicted in Figure 19. Reduction of the Fe(II) complex 907 in the presence of the arene substrate produces the (arene)Fe(0) species 909, which undergoes C–H activation to form the hydride species 910. Hydride loss under reducing conditions, and N₂ binding at low temperature, affords the Fe(I) complex 912, in which the N₂ unit is silylated to afford the (arylhydrazido)Fe(II) complex 913. At this point, the mixture is allowed to warm to room temperature, after which further silylation under reducing conditions yields the amine products and an Fe(II) species which, upon warming to room temperature, reduces and coordinates the arene to form 909, thereby closing the cycle. This remarkable strategy could pave the way for the development of future catalytic systems for the coupling of hydrocarbons and N₂.

Figure 19.

Cyclic pathway for the coupling of of N₂ and unactivated arenes, including key isolated intermediates.⁵⁷⁰

Further mechanistic details on this transformation have recently been reported.⁵⁷¹ In the previous study,⁵⁷⁰ the exact pathway through which reduction, silylation, and migration occurred could not be completely described. With the aid of synthetic, structural, magnetic, spectroscopic, kinetic, and computational studies, the authors have now elucidated additional key steps, reporting a high-valent iron(IV) disilylhydrazido(2−) complex, an intermediate in the silylation of 912 to afford the (arylhydrazido)Fe(II) complex 913 (Figure 19).

The transient Fe(IV) species [(^Dippnacnac)Fe(Ph){NN(SiMe₃)₂}] (916, Scheme 125) was isolated from the reaction of the Fe(I) complex 912 with Me₃SiOTf (2 equiv) at −116 °C, extracting the product at −78 °C and crystallizing it at −35 °C. The low temperature handling of this compound was necessary, due to its thermal instability. Upon warming to room temperature, the product of aryl migration, 913, was obtained. In an attempt to further stabilize the product, the group turned to the bulkier xylyl substituent, synthesizing the analogue to the Fe(I) complex 911, [(^Dippnacnac)Fe(Xyl)]⁻ (917, Xyl = 2,6-Me₂C₆H₃), which coordinates N₂ to give [(^Dippnacnac)Fe(N₂)(Xyl)]⁻ (918). Treatment of the latter compound with Me₃SiOTf (2 equiv) and K([18]crown-6)(C₁₀H₈) afforded the product of aryl migration [(^Dippnacnac)FeN(Xyl)N(SiMe₃)₂]⁻ (919), analogous to 913, which was crystallographically characterized. The corresponding intermediate of this transformation, the iron(IV) disilylhydrazido(2−) with a xylyl substituent, was not isolated but could be spectroscopically identified at low temperature. Kinetics of the aryl migration in these complexes and additional computational investigations of the mechanism of N–C bond formation from the high-valent Fe(IV) intermediate were also reported. For further details, we refer the reader to the original literature.⁵⁷¹

Scheme 125. Synthesis of an Iron(IV) Disilylhydrazido(2−) Intermediate from the Anionic Fe(I) Complex 912(571).

3.2.5. N₂ Binding and Activation at Biomimetic Metalate Species

Holland and co-workers have reported N₂ binding by sulfur-supported iron(0) metalates, acting as facsimiles of the FeMo cofactor (FeMoco), a sulfur-rich iron–molybdenum site for N₂ reduction in nitrogenase enzymes.^575,576 Initial reduction of the high-spin iron(II) precursor [LFe(thf)₂] [920, L = {1,3-(2-S-3-Ar-6–F-C₆H₂)₂(C₆H₄)}; Ar = 2,4,6-triisopropylphenyl] by stoichiometric KC₈ yielded the corresponding iron(I) complex K[LFe] (921, Scheme 126).⁵⁷⁵ EPR spectroscopy and a solution magnetic moment of 2.1 μ_B indicated a low = spin (S = 1/2) iron(I) center in 921, while the crystal structure showed η⁶-bonding to the central arene ring of the ligand.

Scheme 126. Nitrogen Activation by Sulfur-Supported Iron Complexes by Holland and Co-workers^575,576.

Further reduction by an additional equivalent of KC₈ under a nitrogen atmosphere, followed by addition of [18]crown-6, resulted in the dark red iron(0) complex [K{[18]crown-6}]₂[LFeN₂] (922). XRD analysis confirmed that N₂ was bound as a terminal ligand and featured only η²-coordination to the arene. IR spectroscopy revealed an N–N stretching frequency of 1880 cm^–1. This strong activation was attributed to the powerful electron-donating abilities of the thiolate groups, enabling significant backbonding from the iron center into the N₂ π* orbitals. Despite this strong Fe–N₂ interaction, the N₂ ligand remained labile, undergoing exchange with ¹⁵N₂ at −70 °C in the solid state, and samples of 922 kept at ambient temperature for a few hours no longer exhibited N₂ vibrations. SQUID magnetometry suggested the complex featured an unusual high-spin (S = 1) iron(0) center, observations corroborated by DFT calculations performed on a truncated model of 922.

To examine whether Fe–S bond dissociation could provide a coordination site for N₂ binding, as hypothesized for FeMoco, the tris(thiolate) complex K[LFe(SAr′)] (923) was synthesized by reaction of precursor 920 with KSAr′ [Ar′ = −2-(Ar)-C₆H₄] (Scheme 126). The unsaturated, three-coordinate 923 was supported by reversible binding of a molecule of THF at low temperature. Reduction by 2.4 equiv of KC₈ under nitrogen yielded 922, with production of free thiolate, confirming that Fe–S cleavage could facilitate N₂ binding to the low-valent iron sulfide complex.

The mechanistic details of these processes were expanded on in a subsequent report.⁵⁷⁶ Reduction of 923 with stoichiometric KC₈ yields the corresponding iron(I) complex K₂[LFe(SAr′)] (924, Scheme 126), demonstrating that the Fe–S linkage remained intact at this level of reduction. Though crystallization attempts were unsuccessful, a combination of experimental evidence (Mössbauer, EPR, and EXAFS spectroscopy) and calculations indicated a high-spin (S = 3/2) iron(I) tris(thiolate) complex, with η²-coordination to the arene. The complex exhibited limited thermal stability. While maintaining 924 at −70 °C, no loss of thiolate was observed by Mössbauer spectroscopy within 40 min. Upon warming to room temperature, however, features consistent with the presence of 921 were observed within 20 min.

In contrast to the intact Fe–S linkage retained in 924, reduction of 923 by 2.4 equiv of KC₈ under argon resulted in the iron(0) complex K₂[LFe] (925), with loss of KSAr′. Thus, loss of the thiolate is triggered by reduction and does not occur only in the presence of N₂. Complex 925 could also be prepared directly from precursor 920 using excess KC₈.

In the solid-state, 925 was found to be structurally similar to 921, with η⁶-bonding to the central arene, albeit with longer Fe–S/Fe–C bond lengths consistent with a more reduced iron center. Thermal instability limited characterization of 925 beyond Mössbauer spectroscopy, which alongside DFT calculations was taken to indicate a high-spin (S = 1) iron(0) center. Exposure of 925 to N₂ resulted in the formation of 922, as observed by Mössbauer spectroscopy. The change from η⁶- to η²-bonding to the central arene led the authors to suggest that N₂ binding competes with binding to the arene, a hypothesis supported by DFT calculations that indicated π-backbonding between the iron center and both ligands.

Mössbauer spectra of samples of 923 treated by 3.6 equiv of KC₈ under N₂ indicated the formation of a new, highly unstable species which was speculated to be a formal iron(−I) species, K₃[LFeN₂] (926). Though not substantially characterized, treatment of 926 with BHT or H₂O produced substoichiometric amounts of ammonia and hydrazine. While attempts to catalytically perform N₂ reduction were unsuccessful, treatment of 923 with KC₈ and acid (50 equiv each) afforded substoichiometric amounts of NH₃, up to 0.41 equiv of NH₃ per Fe center using BHT as the proton source.

Hydride species have been invoked as possible intermediates in the intimate nitrogen fixation mechanism at FeMoco. A bridging hydride form of this cofactor was proposed, based on spectroscopic data.⁵⁷⁷ Considering that transient hydride ligands bridging two or more iron centers might facilitate N₂ binding at an iron site of FeMoco, Peters and co-workers reported a series of diiron complexes with bridging hydride ligands. These included an anionic mixed valent Fe(II)/Fe(I) species which exhibited an impressive affinity toward N₂ binding.⁵⁷⁸ The synthesis of the dinuclear complexes was enabled by the use of a bulky hexadentate ligand, [SiP₂O]H₂, reminiscent of the previously described silatrane scaffolds reported by the same group (section 2.3.3.2, vide supra).^{140,141,429,430,488−490} The reaction of ligand [SiP₂O]H₂ (Scheme 127) with FeBr₂ (2 equiv) and sodium amalgam under a N₂ atmosphere afforded the bridging hydride complex [{SiP₂O}Fe₂(μ-H)₂(N₂)] (927). Further reaction of 927 with sodium amalgam under a nitrogen atmosphere in the presence of a sequestering agent afforded the anionic complex [Na([12]crown-4)₂][{SiP₂O}Fe₂(μ-H)₂(N₂)₂] (928, see Scheme 127).

Scheme 127. Formation of a Mixed-Valent Fe₂(μ-H)₂ Anionic Species⁵⁷⁸.

Solution magnetometry measurements indicated that 928 is an S = 1/2 species (μ_eff = 1.7 μB), whereas two coupled ν_N≡N vibrations (2023 and 1979 cm^–1) were observed by infrared spectroscopy. At room temperature, the N₂ ligands in 927 and 928 are not labile, as demonstrated by spectroscopic analysis after multiple freeze–pump–thaw cycles. The authors also proved that 927 coordinates another molecule of N₂ at −80 °C to afford the bis(dinitrogen) species [{SiP₂O}Fe₂(μ-H)₂(N₂)₂] (929), according to IR spectroscopy. However, the second N₂ ligand is easily lost in solution upon warming to room temperature, affording 927 again. Cyclic voltammetry and UV–vis measurements confirmed that the neutral bis(N₂) species is predominant in solution at low temperature, and exhibits a reversible reduction event (−2.04 V), assigned to the formation of the anionic complex 928. Therefore, the chemical reduction of 927 is coupled to the coordination of the second N₂ molecule to form 928. The authors attributed this observation to a much stronger binding affinity of N₂ (10⁶-fold enhancement) to the reduced Fe₂(μ-H)₂ core, according to a large equilibrium binding constant and thermodynamic data obtained through electrochemical analysis. Moreover, the electrochemical data suggests that 929 is an intermediate in the reduction of 927 to 928, and that N₂ coordination precedes the electron transfer process. Both 927 and 928 reacted with acid and a reductant (HBAr^F₄/KC₈, 48 equiv each, −78 °C Et₂O), to give N₂ to NH₃ production in comparable amounts (1.4 ± 0.5 equiv NH₃ for 927, 1.1 ± 0.2 equiv NH₃ for 928), demonstrating the potential of these complexes as nitrogenase models.⁵⁷⁸

3.2.6. N₂ Activation and Functionalization by Metallatrane Metalates

As described in section 2.3.3, Peters and co-workers developed metalates featuring a family of Sacconi-type P₃^E tetradentate ligands (Figure 20) (refs (139, 140, 426−428, 430, 551, 560, 579, and 580)). Using this platform, the Peters group has extensively contributed to the field of nitrogen activation and functionalization.^25,581

Figure 20.

Anionic iron and cobalt complexes featuring P₃^E (E = B, C, Al, Si, Ga) tetradentate ligand scaffolds, of relevance in nitrogen activation, developed by Peters and co-workers.

For instance, the reactivity of iron-coordinated dinitrogen ligands toward silyl halide reagents was evaluated.⁴²⁶ By treating the in situ generated species [(P₃^B)Fe(N₂)]⁻ (720) with trimethylsilyl chloride, the silyldiazenido complex [(P₃^B)Fe(N=NSiMe₃)] (930) was obtained (Scheme 128), comparable to compound 874 (see Scheme 120a).⁷⁶ Further reduction of 930 with sodium amalgam afforded [Na(thf)][(P₃^B)Fe(NNSiMe₃)] (931). Compound 931 can also be accessed by direct reduction of the Fe(I) precursor [(P₃^B)FeBr] (718) in the presence of an excess of Me₃SiCl (Scheme 128, upper left).

Scheme 128. Reactivity of N₂ Bound Iron Metallaboratranes toward Silyl Chlorides by Peters and Co-workers⁴²⁶.

As with complexes 719–722 (vide supra), the P₃^B ligand acts as a flexible scaffold to stabilize different electronic environments,¹³⁹ adjusting Fe–B distance [d_FeB(720) = 2.293 Å vs d_FeB(930) = 2.435 Å vs d_FeB(931) = 2.319 Å] or the geometry at the metal center. This was further demonstrated by the reaction of 718 with sodium amalgam (excess) under nitrogen in the presence of 1,2-bis(chlorodimethylsilyl)ethane, which led to the disilylation of the bound N₂ ligand (Scheme 128, down left). The disilylhydrazido(2−) motif in the product 932 is coordinated to the metal center through an Fe≡N triple bond. A comparison of the Fe–N and N–N bond distances for complexes 720, 930–932’ (optimized structure of 932 described by the simplified model, 932′, in which the iPr groups were substituted by Me; see Figure 21) clearly illustrates the changes that occur at the dinitrogen ligand: upon functionalization, the Fe–N bond distances shorten as the N–N lengths elongate.

Figure 21.

Fe–N and N–N bond distances in complexes 720, 930–932′ (* From geometry optimization at the B3LYP/6-31G(d) level).⁴²⁶

Further treatment of 932 with donor ligands such as CO or tBuNC led to substitution of one of the phosphine arms of the P₃^B framework, yielding the Fe≡N compounds 933 (L = CO) and 934 (L = tBuNC) (Scheme 128). In solution, the isocyanide derivative 934 slowly decomposed to generate a phosphoraniminato/disilylamido iron(II) complex, 935, resulting from N–N bond cleavage (Scheme 128, bottom).⁴²⁶ The related anionic compound [(P₃^C)Fe(N₂)]⁻ (936) also reacted with trimethylsilyl chloride to generate the diamagnetic diazenido complex [(P₃^C)Fe(N₂SiMe₃)] (937), similar to 930.⁵⁶⁰

The reactivity exhibited by the (P₃^B)Fe systems served as a basis to investigate the catalytic conversion of N₂ into NH₃.⁴²⁸ At–78 °C, the addition of excess acid (HBAr^F₄) to the highly reduced compound [Na([12]crown-4)₂][(P₃^B)Fe(N₂)] (720) afforded a mixture of compounds in which the main species identified were the ammonia complex [(P₃^B)Fe(NH₃)]⁺ (938) (30–35% of the total Fe) and the unsaturated compound [(P₃^B)Fe]⁺ (725, 40–45% of the total Fe). Reduction of such a mixture regenerated 720. This behavior indicated that a cyclic protonation, followed by reduction, was possible in the frame of a nitrogen fixation system. After evaluating a variety of reductants and acids (e.g., HBAr^F₄/KC₈ or [Ph₂NH₂]OTf/Cp*₂Co), the authors found that, under a nitrogen atmosphere, the sequential addition of an excess of HBAr^F₄ (48 equiv) to 720 followed by reductant (KC₈, 58 equiv, −78 °C, Et₂O) affords 7.0 equiv of NH₃ per Fe center (on average; see Scheme 129). A similar catalytic run using the cationic complex [(P₃^B)Fe]BAr^F₄ (725) yielded 6.2 equiv of NH₃ per Fe center. Other iron salts and molecular species were also evaluated, at best yielding only substoichiometric conversion of N₂ to NH₃. The terminally bonded species 938, [(P₃^B)Fe(N₂H₄)]⁺ (939), and [(P₃^B)Fe(NH₂)] (940) were identified as potential intermediates.⁴²⁷ Nevertheless, the authors did not exclude the possible involvement of species of higher nuclearity or, for example, iron nitride species in the catalytic pathway.⁴²⁸

Scheme 129. Catalytic Conversion of N₂ into NH₃ at Low-Valent (P₃^B)Fe Complexes^427,428.

Protonation of the compounds [(P₃^B)Fe(N₂)]⁻ (720) and [(P^iPr₃^Si)Fe(N₂)]⁻ (738) afforded the doubly protonated products [(P₃^B)Fe≡N–NH₂]⁺ (941)⁵⁷⁹ and [(P^iPr₃^Si)Fe=NNH₂]⁺ (942),⁵⁸⁰ respectively, which were studied as potential intermediates in the iron-catalyzed reduction of dinitrogen (Scheme 130). The unstable hydrazido(2−) product [(P₃^B)Fe≡N–NH₂]⁺ (941)⁵⁷⁹ is reminiscent of compound 932, featuring a disilylhydrazido(2−) motif (Scheme 128, vide supra).⁴²⁶ Species 941 was detected after addition of an excess of acid (HBAr^F₄·2Et₂O, 10 equiv) to 720 at −136 °C in the absence of an exogenous reductant, and characterized by a combined experimental (EPR, ENDOR, EXAFS, ⁵⁷Fe Mössbauer spectroscopy) and theoretical analysis. The characterization strongly supports the presence of the Fe–N triple bond and the double protonation at the distal nitrogen atom. This indicates that the protonation occurs via a “Chatt-type” pathway. Compound 941, however, was characterized in conditions (10 equiv HBAr^F₄, −136 °C, 2-MeTHF) that differ considerably from the catalytic nitrogen fixation system described in Scheme 129 (HBAr^F₄, KC₈, −78 °C, Et₂O).

Scheme 130. Protonation of the Dinitrogen Ligand in the Anionic Compounds [(P₃^B)Fe(N₂)]⁻ (720) and [(P^iPr₃^Si)Fe(N₂)]⁻ (738)^579,580.

To determine whether 941 could be detected in mixtures emulating the studied catalytic system (without exogenous reductant), EPR spectra of two solutions of 720, at −136 °C in 2-MeTHF and at −78 °C in Et₂O, were recorded after addition of acid. Signals attributed to 941 were observed in both mixtures. Upon warming (−40 °C), 941 transforms into different species, including the ammonia-coordinated complex [(P₃^B)Fe(NH₃)]⁺ (938), even in the absence of a reductant.⁴²⁷ This suggests that a double protonation of the distal nitrogen atom, to produce 941 or a related species, is viable and likely involved in the catalytic nitrogen fixation promoted by [(P₃^B)Fe(N₂)]⁻ (720).⁵⁷⁹ Theoretical calculations (DFT) on the mechanism of protonation of 720 indicated that, although the protonation of the anionic complex can occur via many possible reaction pathways, the favored path seems to involve a triple proton transfer to the distal nitrogen atom, forming one ammonia molecule and an iron nitrido complex.⁵⁸²

Despite the evidence of the possible role of 941, due to its instability, it was not possible to perform crystallographic analysis or further reactivity studies on this complex. Thus, the isostructural compound [(P^iPr₃^Si)Fe=NNH₂]⁺ (942), formed by protonation of [(P^iPr₃^Si)Fe(N₂)]⁻ (738) at −135 °C, was crystallographically characterized.⁵⁸⁰ The more stable methylated complex [(P^iPr₃^Si)Fe=NNMe₂]⁺ (943) was formed by the reaction of 738 with MeOTf. Moreover, the transformation of 942 to NH₃ via an Fe-NH₂NH₂⁺ intermediate was demonstrated. One-electron reduction of the cationic complexes afforded the neutral compounds [(P^iPr₃^Si)Fe=NNR₂] [R = H (944), Me (945)]. Thawing of solutions containing both 942 and 944 resulted in disproportionation to generate [(P^iPr₃^Si)Fe-NH₂NH₂]⁺ (946), along with the neutral [(P^iPr₃^Si)Fe(N₂)] (947) and the ammonia complex [(P^iPr₃^Si)Fe(NH₃)]⁺ (948), consistent with previous observations of liberation of ammonia from 942.^140,430 These mechanistically relevant observations indicate that the protonation of a coordinated N₂ ligand in these systems can proceed via a distal intermediate that undergoes reduction/disproportionation to afford intermediates such as 946, which eventually serves as a source of NH₃ by late-stage N–N cleavage.⁵⁸⁰

Increasing the amount of acid and reductant in the reaction catalyzed by [(P₃^B)Fe(N₂)]⁻ (720; Scheme 129) led to a substantial increase in the quantity of ammonia produced.^583,584 The formation of up to 64 equiv of ammonia per Fe center was observed by increasing the amount of HBAr^F₄ and KC₈ to 1500 and 1800 equiv, respectively. Under the same conditions, the analogous complexes [K(Et₂O)_0.5][(P₃^C)Fe(N₂)] (936) and [Na([12]crown-4)₂][(P^iPr₃^Si)Fe(N₂)] (738) afforded 47 and 4.4 equiv of ammonia, respectively (per Fe equiv).⁵⁸⁴ Moreover, using the increased amounts of acid and reductant, and by subjecting the reaction to mercury lamp irradiation, the efficiency of the catalysis by 720 was further increased, generating up to 94 equiv of ammonia (88.1 ± 8.0 equiv NH₃/Fe equiv).⁵⁸³ The related complex [(C^SiP^Ph₃)Fe(N₂)]⁻ (949, C^SiP^Ph₃ = (Ph₂PCH₂SiMe₂)₃CH] yielded only insignificant amounts of NH₃ upon treatment with acid and reductant ([H(Et₂O)₂][BAr^F₄] and KC₈).^560,585

The role of the hydride/borohydride complex [(P₃^B)(μ-H)Fe(N₂)(H)] (950), in the catalytic N₂-to-NH₃ system by [(P₃^B)Fe(N₂)]⁻ (720),^428,586 was also investigated in detail.⁵⁸⁴950 can be synthetically converted to the anionic complex 720 upon reaction with HBAr^F₄ and KC₈, and is a competent catalyst in the reduction of N₂. Mechanistic evidence suggests that compound 950 is an off-cycle resting state of the overall catalysis, and that it can be converted to the catalytically active 720, reentering the catalytic cycle.⁵⁸⁴ Moreover, NMR monitoring of a reaction performed under photolytic conditions with the hydride/borohydride complex [(P₃^B)(μ-H)Fe(N₂)(H)] (950, –78 °C) led to the identification of the neutral [(P₃^B)Fe(N₂)] (719) and the hydrido complex [(P₃^B)(μ-H)Fe(H₂)(H)] (951). This observation was attributed to a transformation involving reductive elimination of H₂ from 950 affording 719, which might regenerate 720 under irradiation.⁵⁸³

Peters and co-workers proved that protonation of a highly reduced Fe–N₂ complex leads to the formation of a neutral Fe(NNH₂) hydrazido(2−) intermediate, which can be protonated in turn to promote the heterolytic cleavage of the N–N bond, forming an iron nitride species and releasing ammonia.⁴⁸⁴ The protonation of the 18-valence-electron dianionic complex [(P₃^B)Fe(N₂)]^2– (723), freshly prepared in situ from the anionic 720, was studied (Scheme 131). Compound 723 reacted with an excess of either triflic acid or HBAr^F₄ in supercooled 2-MeTHF (91–137 K) within 15 min to afford the neutral hydrazido compound [(P₃^B)Fe=N=NH₂] (952, S = 0), according to ⁵⁷Fe Mössbauer spectroscopy. Reactions of 723 with either acid for prolonged periods of time afforded another product, in approximately 50% yield (HOTf), displaying the characteristic parameters of Fe(IV) nitrides in C₃ symmetry in the ⁵⁷Fe Mössbauer spectrum. The latter product was assigned to be complex [(P₃^B)Fe≡N]⁺ (953, Scheme 131). It was proposed that the formation of the Fe(IV)-nitride from 723 proceeds by rapid protonation at low temperature, to yield the hydrazido complex 952 via a diazenido [(P₃^B)Fe(N₂H)]⁻ species. Further protonation of 952, via an unobserved transient hydrazidium cationic species [(P₃^B)Fe(NNH₃)]⁺, resulted in heterolytic N–N bond rupture leading to the nitride complex 953 with release of ammonia. In a larger-scale experiment, protonation of complex 723 with HOTf in supercooled 2-MeTHF afforded NH₃ in 36.0(5)% isolated yield.

Scheme 131. Sequential Protonation of the Dianionic Complex [(P₃^B)Fe(N₂)]^2– (723) Leading to a Terminal Fe(IV) Nitride and Ammonia⁴⁸⁴.

Moreover, the authors proved that it is possible to double the isolated yield to 73(17)% by the sequential reaction of 723 with HOTf and Cp*₂Co in supercooled 2-MeTHF. Therefore, the simultaneous use of the two reductants, KC₈ and Cp*₂Co, can drive the catalytic N₂ fixation in this system. However, the dianionic complex 723 is only accessible by treatment with the stronger reductant KC₈, since Cp*₂Co alone is only sufficiently strong to produce [(P₃^B)Fe(N₂)]⁻ (720).⁵⁸⁷

From complex 723 to the nitride compound 953, a range of six formal oxidation states of iron are observed, from Fe(−II) (d¹⁰) to Fe(IV) (d⁴). It should be noted that the assignment of these formal oxidation states may not account for the effect of the Z-type ligand (Fe-to-B σ-backbonding) or the additional metal-to-phosphine π-backbonding.⁴⁸⁴ In fact, the experimental and calculated pre-edge transitions in the XANES spectrum of the dianionic complex 723 indicate that its physical oxidation state must involve a dⁿ configuration of n < 10. Therefore, the (P₃^B)Fe unit donates electron density into the covalent Fe–B/P backbonding interactions, modulating the oxidation state of the metal center until the electrons are transferred to the N–N moiety upon protonation.

The electronic structure of the series of complexes [(P₃^B)Fe(N₂)]ⁿ [n = 0 (719), 1– (720), 2– (723)] was investigated using DFT calculations by Vyas, Grover and co-workers.⁵⁸⁸ The theoretical study examined model systems of the complexes, featuring a simplified ligand (tris[(dimethylphosphino)phenyl]borane, P₃^B′). The variation in the oxidation states of the iron center was attributed to the flexible coordination environment provided by the P₃^B ligand. The Fe–B interaction can, therefore, function as a reservoir of electron density to support and modulate the various oxidation states. These changes were found to be crucial in the system of nitrogen activation. Moreover, calculated optimized structures of the complexes indicated that complex [(P₃^B)Fe(N₂)]^2– (723) corresponds to the species of lowest energy. As reduction of the iron center takes place, the calculated Fe–B bond distances elongate in the series of complexes [(P₃^B)Fe(N₂)]ⁿ (n = 0: 2.280 Å, n = 1–: 2.345 Å, and n = 2–: 2.369 Å), and the Fe–N bond distances shorten (n = 0: 1.973 Å, n = 1–: 1.807 Å, and n = 2–: 1.779 Å), due to increased π-backbonding from the metal center to the dinitrogen ligand. The HOMO–LUMO gap was found to be of 4.43, 3.59 and 2.76 eV for the series [(P₃^B)Fe(N₂)]ⁿ (n = 0, 1–, 2−), respectively, indicating that the dianionic complex [(P₃^B)Fe(N₂)]^2– (723) will be the more reactive species. Consequently, the theoretical analysis of the bonding situation corroborates the experimental evidence, suggesting that the nitrogen activation process is more viable at the more reduced iron center.

A comparison of (P₃^E)Fe (E = B, C, Si) revealed that the (P₃^B)-supported system is the most catalytically active for the N₂-to-NH₃ reaction in the presence of external proton and electron sources under all the conditions evaluated.^{428,430,560,583,584} To further investigate this matter, a structure-to-function study was carried out aiming to determine the effect of the apical Lewis acidic atom, including additional dinitrogen–iron compounds featuring the tetradentate P₃^E ligands (E = B, Al, Ga).⁵⁸⁹ Similar to the synthesis of [(P₃^B)Fe(N₂)]ⁿ [n = 0 (719), 1– (720)] (Scheme 86, vide supra), the series of complexes [(P₃^E)Fe(N₂)]ⁿ [P₃^E = P₃^Al: n = 0 (954), 1– (955) or P₃^E = P₃^Ga: n = 0 (956), 1– (957)] was obtained. Their bonding is best described as featuring a dative Fe→E(III) interaction, with a comparable degree of activation of the coordinated N₂ ligands in each set of complexes, [(P₃^E)Fe(N₂)] or [(P₃^E)Fe(N₂)]⁻. Their electronic structures and the flexibility of the Fe–E bonds are also similar, and the anionic complexes feature Fe(−I) centers as a result of iron-based reduction events. Therefore, the anionic complexes 955 and 957 were also tested for the catalytic N₂-to-NH₃ reaction under previously optimized conditions, using excess HBAr^F₄/KC₈ or [Ph₂NH₂]OTf/Cp*₂Co (−78 °C, Et₂O). Using the combination HBAr^F₄/KC₈, the catalysts 955 and 957 afforded only 2.5 ± 0.1 (17 ± 1% yield NH₃/H⁺) and 2.7 ± 0.2 (17 ± 1% yield NH₃/H⁺) equivalents of NH₃ per Fe center, respectively. With the combination [Ph₂NH₂]OTf/Cp*₂Co, the efficiency of the reaction improved (955: 4.1 ± 0.9 equiv of NH₃ per Fe; 957: 3.6 ± 0.3 equiv of NH₃ per Fe). Nonetheless, none of these systems was as efficient as the one catalyzed by [(P₃^B)Fe(N₂)]⁻ (720). Individual monitoring of the reactions between the anionic complexes 955 and 957 and lower stoichiometries of acid and reductant (10 and 12 equiv, respectively) indicated that the catalysts are rather robust under these conditions and, consequently, their significantly lower activity in the reduction of N₂ cannot be attributed to decomposition. Complexes 955 and 957 were found to be more selective toward the hydrogen evolution reaction, and might be susceptible to the formation of hydride species similar to [(P₃^B)(μ-H)Fe(N₂)(H)] (950), explaining the lower efficiency of the catalysis.

Related (P₃^B)Co complexes are also active in nitrogen fixation.⁵⁵¹ Reduction of the cobalt compound [(P₃^B)Co(N₂)] (958)⁵²⁶ with sodium naphthalenide (NaC₁₀H₈), followed by addition of [12]crown-4, afforded the diamagnetic anionic complex [Na([12]crown-4)₂][(P₃^B)Co(N₂)] (959, Scheme 132a). The dinitrogen ligand becomes significantly more activated upon reduction, as evidenced by the red-shift in the stretching vibrations in the FT-IR spectrum [ν_N≡N(958) = 2089 cm^–1 vs ν_N≡N(959) = 1978 cm^–1, see Table 2].^526,551 Treatment of 959 with an excess of HBAr^F₄/KC₈ (48 equiv HBAr^F₄, 58 equiv KC₈, −78 °C, Et₂O) generates 2.4 ± 0.3 equiv of NH₃/Co (Scheme 132b). The analogous iron complex [(P₃^B)Fe(N₂)]⁻ (720) is more active than the cobalt analogue by a 3-fold ratio (Scheme 132b).⁴²⁸ In the absence of either acid or reductant, no formation of NH₃ was observed. Various neutral cobalt complexes were evaluated for this reaction under the same conditions, yielding substoichiometric amounts of ammonia in the best case. Apart from complex 959, only the oxidized product [(P₃^B)Co]BAr^F₄ (960) mediated the N₂-to-NH₃ conversion in superstoichiometric yields (1.6 ± 0.2 equiv of NH₃/Co), albeit less effectively than 959.

Scheme 132. (a) Synthesis of the Anionic Complex [(P₃^B)Co(N₂)]⁻ (959) and (b) N₂-to-NH₃ Conversion Mediated by Anionic Iron or Cobalt Complexes^428,551.

The divergent activity of the (P₃^B)Fe and (P₃^B)Co complexes were rationalized by performing a structure-to-function analysis. While a priori the degree of activation of the dinitrogen ligand at the metal center seems to correlate well with the better performance of the iron complex in the reaction [ν_N≡N(720) = 1905 cm^–1 vs ν_N≡N(959) = 1978 cm^–1], other complexes with an apparent higher degree of activation (e.g., [Na([12]crown-4)₂][(P^iPr₃^Si)Fe(N₂)] (738), ν_N≡N = 1920 cm^–1)⁴³⁰ gave lower N₂-to-NH₃ conversion (0.8 ± 0.4 equiv/Fe). Instead, the flexibility of the M–E bond trans to the M–N₂ moiety seems to be a crucial factor in explaining why certain complexes appear to be particularly active N₂ reduction catalysts. Moreover, the results seem to indicate that the anionic charge in metalate complexes results in a higher basicity of the N₂ ligand, which, together with the flexibility of the M–E bond, favors the production of ammonia from N₂.⁵⁵¹

In related work, Lu and co-workers showed that reduction of the complexes [(N₂)MAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] {M = Fe [760(N₂)], Co [761(N₂)]} (see section 2.3.3.3) with KC₈ affords anionic complexes [K(L)][(N₂)MAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (M = Fe, L = [18]crown-6 [961(N₂)]; M = Co, L = [2.2.2]cryptand [962(N₂)], Scheme 133a).^431,432 In the solid state, the iron complex 760(N₂) is in equilibrium between the end-on species and a N₂-bridged dimer [{760}₂(μ-N₂)] (vide supra). For the latter, the bridging N₂ ligand was activated to some extent, as reflected by the slightly elongated N–N bond distance [1.107(4) Å].^431,432 Reduction of the zerovalent complexes 760(N₂) and 761(N₂) further activates the coordinated dinitrogen ligand in 961(N₂) and 962(N₂), formally in the negative oxidation state M(−I).⁴³²

Scheme 133. (a) Chemical Reduction of [(N₂)MAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] {M = Fe [760(N₂)], Co [761(N₂)]} and (b) Functionalization of Coordinated N₂ at the Fe(−I) Complex 961(N₂).

An alternative preparation of these complexes consists of treating FeBr₂ or CoBr₂, with metalloligand AltraPhos and 3 equiv of KC₈, as opposed to the 2 equiv used for the synthesis of the zerovalent compounds (see Scheme 92, above). FT-IR characterization revealed, after reduction, a decrease in the ν_N≡N stretching frequencies (on average 85 cm^–1, see Table 2). In accordance with this, the d_NN bond distances measured for both complexes [d_NN(961) = 1.135(4) Å vs d_NN(962) = 1.116(8) Å] are longer in the anionic complexes (cf. d[{760}₂(μ-N₂)] = 1.107(4) Å vs d_NN = 1.0975 Å in free N₂). Further evidence of the higher degree of activation of the N₂ ligand at Fe(−I) was obtained by reactions of 961(N₂) or 962(N₂) with 1,2-bis(chlorodimethylsilyl)ethane. At room temperature, 962(N₂) furnished the neutral 761(N₂), while the reaction at −78 °C resulted in an intractable mixture. By contrast, the reaction of 961(N₂) with 1,2-bis(chlorodimethylsilyl)ethane afforded a disilylhydrazido(2−) complex [{N₂(SiMe₂CH₂)₂}FeAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (963, Scheme 133b) (c.f. complex 932, Scheme 128b).⁴²⁶

The same group described a bimetallic iron–tin complex which catalyzes the conversion of N₂ into NH₃.⁵⁹⁰ One- and two-electron reductions of the previously reported Fe(II) complex [BrFeSn{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁵⁹¹ (964) by either 1.1 or 2.2 equiv of KC₈ under a nitrogen atmosphere afforded the neutral, [(N₂)FeSn{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (965) and doubly reduced complex [{(thf)₃K}(N₂)FeSn{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (966), respectively. The increased electron density in 966 with respect to 965 is reflected by the red-shift of the stretching bands of their dinitrogen ligands (1944 cm^–1 for [K([2.2.2]cryptand)][966] vs 2011 cm^–1 for 965).

The valency of complexes 964–966 was described using an adapted Feltham–Enemark notation,⁴⁸⁷ given that assigning individual oxidation states for both metals can be complicated by the presence of an intermetal covalent bond. Therefore, 964, 965, and 966 were assigned electronic configurations of {FeSn}⁸, {FeSn}⁹, and {FeSn}¹⁰, respectively. The reaction of 966 with a slight excess of Me₃SiCl allowed isolation and structural characterization of the diazenido complex [(Me₃SiN₂)FeSn{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (967, Scheme 134a). A comparison of the Fe–N and N–N bond lengths for the triad of complexes 965–967 shows progressive elongation of the N–N bonds (1.112(2) Å in 965, 1.143(6) Å in 966, 1.182(3) Å in 967) and contraction of the Fe–N bonds (1.7933(14) Å in 965, 1.762(5) Å in 966, 1.686(2) Å in 967). The structural analysis of 967 indicates that the Me₃SiN₂ unit is between the two possible resonance forms for a diazenido ligand: bent diazenido and diazenium ligand.

Scheme 134. N₂ Functionalization and Catalytic N₂ to NH₃ Reduction at Iron–Tin Complexes⁵⁹⁰.

It remained unclear, after analysis by ⁵⁷Fe Mössbauer spectroscopy, whether the electron configuration of 967 should be described as an {FeSn}¹⁰ unit engaged in π-backbonding with a silyldiazenium(1+) ligand or as an {FeSn}⁸ unit acting as π-donor to a diazenido(2−) ligand. Quantum chemical calculations (DFT) indicated a strong correlation between the charge on the Sn center (atom trans to the N₂ unit) and both the polarization of the N₂ unit and the charge on the distal N atom.

Complexes 965–967 were evaluated in the catalytic N₂ to NH₃ reduction reaction, using [Ph₂NH₂]OTf (108 equiv) as the acid and Cp*₂Co (54 equiv) as a reductant (1 atm N₂, Et₂O, −78 °C, 3 h). Under these conditions (see Scheme 134b), while the neutral complex 965 generated 5.9(5) turnovers of NH₃ (33% yield), its catalytic performance did not match that of Peters’ cationic complex [(P₃^B)Fe]BAr^F₄ (725),⁴²⁷ which afforded 12.8 turnovers of NH₃ (72% yield). The anionic complex 966 afforded only substoichiometric amounts of NH₃ (0.8(5) turnovers of NH₃, 5% yield) as a precatalyst, attributed to its low solubility in Et₂O. The diazenido complex 967 matched the catalytic performance of 965 (4.6(2) turnovers of NH₃, 26% yield), within the experimental error. 967 was also tested as catalyst in the silylation of N₂ with Me₃SiCl and KC₈, given that diazenido complexes have been proposed as intermediates in such reactions. However, only 1.2 equiv of N(SiMe₃)₃ were afforded under the studied conditions.

Similarly, a series of anionic cobalt-dinitrogen complexes [(N₂)CoV{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [968(N₂)], (N₂)CoCr{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [969(N₂)] and [(N₂)Co₂{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [970(N₂)], featuring V, Cr, or Co-based metalloligands were later reported.^{494,496,552,592}Figure 22 shows the structures of these heterobimetallic cobalt complexes, and the N–N bond distances obtained from crystallographic analyses. The dinitrogen molecule is slightly less activated in the dicobalt and CoAl complexes, 970(N₂) and 962(N₂), than in the analogues 968(N₂) and 969(N₂). These distances are consistent with the values observed for the N₂ stretching vibration by infrared spectroscopy (see Table 2) and, in general, indicate weakly activated N₂ units.^432,494,552 Furthermore, across this series of heterobimetallic cobalt complexes DFT calculations indicated that the N₂ π* molecular orbitals are energetically inaccessible, which is consistent both with the crystallographic parameters and the trends in the FT-IR spectra. In addition, the theoretical analysis indicated that the electronic structures of the anionic CoV, CoCr (and the hypothetical CoTi species, which was not isolated) are approximately similar to that of the CoAl complex [(N₂)CoAl{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [962(N₂), Scheme 133a, vide supra],⁴³² i.e., d¹⁰ Co(−I) centers bound to M(III) supporting metals.⁴⁹⁴ In contrast, complex 970(N₂), featuring a late transition metal support, consists of a Co(0) center binding to a Co(II) atom.⁵⁵²

Figure 22.

Heterobimetallic cobalt complexes featuring V, Cr, Co, and Al metalloligands, M{N[o-(NCH₂PiPr₂)C₆H₄]₃}. N–N bond distances are shown for comparison. *Average values for two crystallographically independent molecules.^432,494,552

The dicobalt complex [(N₂)Co₂{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [970(N₂)] is stable under vacuum.⁵⁵² Consequently, this complex and its neutral analogue [(N₂)Co₂{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (971) were evaluated in the catalytic silylation of N₂. Using 971 as a precatalyst (0.05 mol %) in the presence of a large excess of Me₃SiCl and KC₈ under an atmosphere of N₂ (r.t., 12 h, THF), N(SiMe₃)₃ was obtained in 30% yield, with a TON of 195 ± 25 (Scheme 135). The anionic complex 970(N₂) was equally efficient (27% yield, TON = 178 ± 37), on average. The performance of other cobalt species such as [761(N₂)]⁴³¹ was comparatively poor.

Scheme 135. Catalytic Silylation of N₂ Using Complexes [(N₂)Co₂{N[o-(NCH₂PiPr₂)C₆H₄]₃}]^0/–552.

Only the mixture of CoCl₂ (2 equiv) with the heptadentate ligand N[o-(HNCH₂PiPr₂)C₆H₄]₃ reproduced the catalytic activity of 971/970 (N₂) (25% yield, TON = 172 ± 16). Catalyst performance was practically unaffected by the presence of additives such as PMe₃ or tBuNC. This, among other mechanistic tests, indicated that a homogeneous species was responsible for the catalysis. According to the proposed mechanism, the reaction occurs only at the Co(0) atom. Nonetheless, DFT calculations revealed that the Co–Co interaction is weakened substantially upon coordination of the N₂ molecule and broken during the subsequent silylation step. Importantly, the Co(0)–Co(II) interaction reforms with the release of trisilylhydrazide. The hemilabile nature of the metal–metal interaction thus appears to be a critical factor influencing the catalysis.

Very recently, Fang, Cui, and co-workers reported a series of Co(−I)–N₂ complexes supported by rare-earth metals via Co→RE (RE = rare-earth metal) interactions, and their activity in catalytic silylation of N₂.⁵⁹³ Reduction of the Co(I) complexes [{N[CH₂–o-(NCH₂PPh₂)C₆H₄]₃}RECoI] [RE = Sc (972), Lu (973), Yb (974), Y (975), Gd (976)] with KHBEt₃ (3 equiv) in the presence of crown ether afforded the anionic dinitrogen adducts [K([18]crown-6)(THF)_x][{N[CH₂–o-(NCH₂PPh₂)C₆H₄]₃}RECo(N₂)] [RE = Sc (977), Lu (978), Yb (979), Y (980), Gd (981); see Scheme 136a]. The analogous Co(I) and Co(−I) complexes lacking a supporting rare-earth metal center, [{N[CH₂–o-(NHCH₂PPh₂)C₆H₄]₃}CoI] (982) and [Na([12]crown-4)₂][{N[CH₂–o-(NHCH₂PPh₂)C₆H₄]₃}Co(N₂)] (983), were also synthesized (Scheme 136a).

Scheme 136. (a) Synthesis of Co(−I)-RE(III) and Unsupported Co(−I) N₂-Adducts and (b) Catalytic Silylation of N₂ Using the Synthesized Complexes⁵⁹³.

The characterization of the C₃-symmetric complexes 977–981 and computational calculations support Co→RE dative bonds in all cases, as well as weakly activated dinitrogen ligands (ν_N≡N = 2044–2026 cm^–1). This might be the result of an interplay between a reduced Lewis acidity of the supporting RE(III) metal and the less electron-donating diphenyl phosphine groups, compared to the alkyl substituted phosphines in Lu’s heptadentate ligand.^432,494,552 Nonetheless, a correlation between the degree of activation of the dinitrogen ligand and the pKa values of the aqueous [RE(OH₂)₆]³⁺ ions showed that more Lewis acidic RE(III) ions in the ligand are associated with less activated N₂ ligands in the Co(−I)–N₂ complexes. Accordingly, 981, stabilized by the least Lewis acidic Gd(III) ion, induced the strongest activation of the coordinated N₂ unit. The marked effect exerted by the RE(III) supporting metal is also evident upon comparison of the ν_N≡N stretching bands of 977–981 with that of the unsupported complex 983, which features a more activated N₂ ligand (ν_N≡N = 1969 cm^–1).

The anionic complexes 977–981, as well as the mononuclear Co(I) compound 982, were tested in the catalytic silylation of N₂ in the presence of KC₈ and Me₃SiCl (THF, 25 °C, 4 days; Scheme 136b). All the complexes were active for this reaction, with the CoLu compound 978 exhibiting the best catalytic performance within the tested compounds (24% yield, TON = 16). Moreover, the supporting RE(III) ions were found to be crucial to promoting catalytic turnovers: whereas complexes 977–981 catalyzed the reaction with TONs in the range of 3 to 16, and yields of N(SiMe₃)₃ up to 24%, the unsupported Co(I) complex 982 performed poorly under the same conditions, affording only 2% yield of product, with a TON of 1.3. The anionic compound 983 was not tested due to its instability in solution. Therefore, the authors proposed that the supporting RE(III) ions play an additional role in the stabilization of the highly reduced Co(−I) centers, and possibly in other catalytically relevant intermediates, as well as in maintaining the geometry and rigidity of the species involved in the catalysis.⁵⁹³

Peters and co-workers demonstrated that the isostructural ruthenium and osmium silatrane complexes [K(thf)_x][(P^iPr₃^Si)M(N₂)] [M = Ru (747), Os (748); section 2.3.3.2, vide supra] are competent catalysts for the N₂-to-NH₃ conversion (Scheme 137).⁴⁸⁹ FT-IR characterization of 747 and 748 revealed that these compounds feature significantly activated N₂ ligands [ν_N≡N(747) = 1960 cm^–1, ν_N≡N(748) = 1931 cm^–1]. Consequently, N₂-to-NH₃ conversion was studied under conditions similar to those used with related iron systems such as [(P^iPr₃^Si)Fe(N₂)]⁻ (738) or [(P₃^B)Fe(N₂)]⁻ (720) (46 equiv HBAr^F₄, 50 equiv KC₈, 1 atm, −78 °C, Et₂O).^428,430 Using complexes 747 and 748 the reaction afforded 4.3 ± 0.3 equiv and 1.6 ± 0.3 equiv of NH₃ per metal complex, respectively.

Scheme 137. (a) N₂-to-NH₃ Conversion Reactions by Group 8 M-Silatrane Complexes and (b) Protonation Reactions of [(P^iPr₃^Si)Os(N₂)]⁻ (748)⁴⁸⁹.

With the milder reagents [H₂NPh₂]OTf (46 equiv) as the acid source and Cp*₂Co (50 equiv) as the reductant in Et₂O at −78 °C, the ruthenium complex 747 or the iron analogue 738 afforded only substoichiometric or stoichiometric amounts of ammonia, respectively, while the osmium complex 748 afforded 7.1 ± 0.6 equiv of NH₃ per osmium center (Scheme 137a). Increasing the amount of acid and reductant in the reaction catalyzed by [(P^iPr₃^Si)Os(N₂)]⁻ (748) to 1500 equiv of [H₂NPh₂]OTf and 1800 equiv of Cp*₂Co, respectively, led to a yield of 120 ± 11 equiv of NH₃ per Os.

The reaction of 748 with HBAr^F₄ at −78 °C afforded a mixture of the osmium hydrides [(P^iPr₃^Si)Os(N₂)(H)] (984) and [(P^iPr₃^Si)Os(H)₃] (985, Scheme 137b). These two hydride compounds were independently synthesized and reacted with acid and reductant under the N₂-to-NH₃ conditions (HBAr^F₄ or [H₂NPh₂]OTf, 46 equiv; KC₈ or Cp*₂Co, 50 equiv; 1 atm, −78 °C, Et₂O) without appreciable formation of ammonia. Further mechanistic tests also indicated that hydrides 984 and 985 form during the catalysis and are catalytically inactive states. In turn, reaction between 748 and HOTf (3 equiv) in thawing 2-MeTHF (−135 °C) generated the hydrazido(2−) complex [(P^iPr₃^Si)Os=NNH₂]OTf (986), which was structurally authenticated. Treatment of 986 with 46 equiv of [H₂NPh₂]OTf and 50 equiv of Cp*₂Co at low temperature afforded 2.6 equiv of ammonia per osmium center, illustrating that it facilitates the generation of ammonia in this system and is a viable intermediate in the N₂-to-NH₃ conversion. Notably, while the analogous iron- or ruthenium-based systems require stronger reductants to drive the reduction of N₂, the osmium system operates with the milder Cp*₂Co, but only at low temperature.

3.3. Activation of CO and CO₂

Carbon monoxide and carbon dioxide are fundamentally important molecules in industry and nature. While carbon monoxide is widely used as a C1-feedstock in industry,^155,594 CO₂ is an inert and thermodynamically stable molecule. Nevertheless, the chemical valorization of CO₂ is attracting enormous interest due to its role as a greenhouse gas.^595,596 It has long been known that the nucleophilic character of transition metalate complexes can be harnessed in such processes. In a pioneering study published in 1988, Focchi investigated the reaction of K[V(C₆H₆)₂] (987) with CO and CO₂.⁷⁰ In both cases, a range of C₁ and C₂ products were detected, including HCO₃^–, HCOO^–, HC₂O₄^2–, C₂O₄^2–, and minor amounts of HOCH₂COO^–. CO, H₂, and C₂H₂ were not detected in the headspace of the reactions. By contrast, Corella and Cooper observed the addition of CO₂ to coordinated benzene upon reaction of CO₂ with [Cr(η⁴-C₆H₆)(CO)₃]^2– (988), resulting in the formation of substituted cyclohexadienyl complexes.⁵⁹⁷ Furthermore, the reductive disproportionation of CO₂ into carbonate and carbon monoxide by dianionic carbonyl metalates based on vanadium, group 6 metals (Cr, Mo, W) and group 8 metals (Fe, Ru, Os) is well-documented.^598,599

3.3.1. CO Activation by Ferrates, Cobaltates, and Molybdates

CO and CO₂ activation by metalates has increasingly gained interest in recent years, including some catalytic applications. The carbonylation of tertiary amines using low-valent iron catalysts was described in 2019.⁶⁰⁰ The series of low valent carbonyl-iron complexes K₂[Fe(CO)₄] (K₂[1]), Fe(CO)₅ (18), [Fe₃(CO)₁₂] (22) and [(η⁵-Cp)Fe(CO)₂]₂ (651) were evaluated in the catalysis. The carbonylation of C–N bonds in tertiary amines was initially studied using the following conditions: N,N-dimethylaniline as the model substrate, 6 mol % catalyst loading, 55 bar of CO and MeI as a promoter (0.8 equiv) at 200 °C for 15 h in acetonitrile. Under these conditions, the selective carbonylation of one N–Me bond was achieved in moderate yields (40–72%) using the neutral complexes 18, 22, or 651 as catalysts. By comparison, the iron metalate K₂[1] afforded quantitative yields of the product N-methylacetanilide. To achieve comparable yields with catalyst 18 or 651, the reaction time had to be extended to 60 h. At lower pressures (8 bar), a decrease in the catalytic activity was observed for all the catalysts (17–58% yield), though the carbonylferrate K₂[1] still exhibited the highest activity (58% yield). Therefore, K₂[1] was specifically examined in further catalytic screening (Scheme 138). To overcome the loss of activity observed with the lower pressure of CO, the use of Lewis acids additives was evaluated. Upon addition of either Sc(OTf)₃ or Nd(OTf)₃ (60 mol %), the carbonylation by ferrate K₂[1] led to quantitative generation of N-methylacetanilide under 8 bar of CO. The reaction scope was found to extend to methylamine substrates with electron withdrawing or electron donating groups, as well as ethylamine substrates (10 examples in total). Yields varied substantially across the range of substrates (28 to 99%).

Scheme 138. Carbonylation of Tertiary Amines Catalyzed by a Low-Valent Ferrate, K₂[Fe(CO)₄] (K₂[1])⁶⁰⁰.

As mentioned above (section 2.3.3.2), the iron-carbyne complex [(P^iPr₃^Si)Fe≡C-OSiMe₃] (744, Scheme 89, vide supra) was obtained from the the anionic carbonylferrate [(P^iPr₃^Si)Fe{CONa(thf)₃}] (741) and a silyl electrophile.¹⁴¹ In a similar vein, the Peters group reported the reduction of carbonyl ligands at related examples of anionic iron and cobalt complexes.^601−603 Reduction of the dicarbonyliron complex [(PhP₂^B)Fe(CO)₂] [989, PhP₂^B = PhB(o-iPr₂PC₆H₄)₂] with potassium in the presence of benzo[15]crown-5 affords ferrates in two distinct oxidation states: the monoanionic compound [K(benzo[15]crown-5)₂][(PhP₂^B)Fe(CO)₂] (990) was obtained with one equivalent of potassium, whereas an excess of reducing agent gave the complex [K(benzo[15]crown-5)₂]₂[(PhP₂^B)Fe(CO)₂] (991, see Scheme 139).⁶⁰¹

Scheme 139. Reduction of an Iron-Dicarbonyl Complex in the Absence and in the Presence of a Silyl Electrophile: Synthesis of a CO-Derived Iron Dicarbyne⁶⁰¹.

The increase in electron density at the metal atom is reflected by the significant red-shift of the stretching bands for the carbonyl ligands in 989–991 [ν_CO(989) = 1908 cm^–1, ν_CO(990) = 1857 cm^–1, ν_CO(991) = 1738 cm^–1]. The low energy C–O stretch of 991 indicated that the functionalization of the CO ligands should be feasible. In line with this, the reduction of the dicarbonyl compound 989 by potassium in the presence of trimethylsilyl triflate (Me₃SiOTf) afforded the silylated product [(PhP₂^B)Fe(COSiMe₃)₂] (992, Scheme 139), which can be described as a double terminal iron-carbyne complex. The characterization data of 992 (X-ray diffraction, NMR, and ⁵⁷Fe Mössbauer spectra) point to extensive Fe–C multiple bonding and an additional C–B interaction in one of the carbyne units. Based on an analysis of the structural parameters and DFT calculations on a simplified model complex (Me₃P)₂Fe(COSiH₃)₂, the first three resonance forms shown in Figure 23 (from left to right) should be more representative of the bonding situation of 992 than the one in the right. In addition, compound 992 reacted with H₂ (1 atm, r.t.) via hydrogenative C–C coupling to afford the olefin Z-Me₃SiOCH=CHOSiMe₃, a functionalized CO-derived product (Scheme 139).

Figure 23.

Relevant resonance structures of the iron dicarbyne complex [(PhP₂^B)Fe(COSiMe₃)₂] (992).

The previously reported carbonyl complex [(P₃^B)Fe(CO)] (721)¹³⁹ reacts with H₂ to afford the bridged-hydride species [(P₃^B)(μ-H)Fe(H)(CO) (993).^586,602 Complex 993 undergoes two-electron reduction to afford the dianionic hydridocarbonyl compound 994 featuring a significantly activated CO ligand (ν_CO = 1575 cm^–1; see Scheme 140). Compound 994 is protonated with traces of water to afford an unstable anionic trihydride complex [K(thf)_n][(P₃^B)(μ-H)Fe(H)₂(CO)] (995), which loses a molecule of hydrogen to yield the hydrocarbonylferrate [K(thf)_n][(P₃^B)FeH(CO)] (996). Moreover, the dianionic ferrate 994 is silylated at the oxygen atom upon reaction with Me₃SiOTf (>2 equiv) to afford [(P₃^B)Fe-CH₂OSiMe₃] (997, Scheme 140). In complex 997 the hydride ligands have migrated to the carbonyl carbon atom. While 997 reacts with different Brønsted acids to yield the bridged hydride 993, the functionalized −CH₂OSiMe₃ fragment is released as CH₃OSiMe₃ or PhSiH₂CH₂OSiMe₃ in the presence of hydrogen (hydrogenolysis) or silanes, respectively (Scheme 140, bottom). The functionalization reaction of the CO thus results in a net 4-electron reduction of a carbonyl ligand.⁶⁰²

Scheme 140. Formation of Hydrido(carbonyl)Iron Complexes and Reduction of an Iron-Bound Carbonyl Ligand Followed by Release of the CO-Functionalized Product^586,602.

A trisphosphine supported carbonylcobaltate complex can also be O-functionalized to afford a carbyne motif.⁶⁰³ The cobalt complex [(P₂^P)CoBr(CO)] (998, P₂^P = PhP(o-iPr₂PC₆H₄)₂) undergoes reduction by sodium to yield the Co(−I) compounds [NaL_n][(P₂^P)Co(CO)] [L_n = (solv)_x (999), ([12]crown-4)₂ (1000); Scheme 141]. The carbonyl ligands in the reduced complexes present a high degree of activation, according to the stretching vibrations [ν_CO(999) = 1731 cm^–1; ν_CO(1000) = 1752 cm^–1]. Treatment of 999 with silyl electrophiles (R₂R’SiOTf) afforded the carbyne complexes [(P₂^P)Co≡C-OSiR₂R′] [R = R′ = iPr (1001), R = Ph, R′ = tBu (1002)] in a high yield. These species were identified as terminal carbyne complexes via structural characterization of the tBuPh₂ derivative 1002, which features a significantly shortened Co–C bond (1.640(4) Å) and an elongated C–O distance (1.260(5) Å).

Scheme 141. Synthesis of a (P₂^P)Co(−I)-Carbonyl Complex and Its O-Functionalization to Afford Terminal Carbyne Compounds⁶⁰³.

In 2016, Agapie and Buss described a remarkable transformation taking place at a molybdenum complex bearing a terphenyl–diphosphine ligand (tpdp = 1,4-bis(2-(diisopropylphosphino)phenyl)benzene).⁶⁰⁴ The formally zerovalent complex [Mo(tpdp)(CO)₂] (1003) undergoes four-electron deoxygenative reductive coupling of carbon monoxide in the presence of four equivalents of an appropriate electrophile (E⁺, ECl = iPr₃SiCl or Me₃SiCl) under a nitrogen atmosphere to afford the molybdenum dinitrogen compound [Mo(tpdp)N₂] (1004), along with products EOC≡CE and E₂O resulting from rearrangement and functionalization of the carbonyl ligands (Scheme 142). This transformation is enabled by the architecture of the ligand scaffold, which allows both the formation of a highly anionic overall complex and the adoption of different coordination modes, stabilizing the several intermediates involved in the reaction pathway. In this context, the authors identified and characterized the anionic species [K₂(thf)₃][Mo(tpdp)(CO)₂] (1005) and [K₃(thf)_2.5][Mo(tpdp)(CO)₂] (1006), formally containing Mo(−II) (1005) and Mo(−III) (1006) metal centers, upon reducing the neutral precursor 1003 with two equivalents or an excess of KC₈, respectively. In the solid state, both 1005 and 1006 are potassium-ion-bridged polynuclear clusters, where the cations interact with one phosphine arm, the oxygen atoms of the carbonyl donors, and the ligand π-systems.

Scheme 142. Deoxygenative CO-Upgrade at a Molecularly Defined Mo-tpdp Scaffold: The tpdp Motif Allows for the Storage of Two (1005) or Three (1006) Redox Equivalents, Therefore Activating the Coordinated CO⁶⁰⁴.

The molybdenum-arene hapticity changes from η⁶-coordination in 1003 to either η⁴-coordination in compound 1005 or η³-coordination in 1006, causing significant distortion from planarity in the central arene. For this motif, the C–C bond metrics are indicative of a considerable delocalization of anionic charge into the terphenyl moiety and the adoption of a partial cyclohexyldienyl dianion character. In this extreme case, 1005 can be described as a Mo(0) anion and 1006 as a Mo(−I) species. Across the reduction series [Mo(tpdp)(CO)₂]ⁿ⁻ (n = 0, 1, 2), the X-ray data also reveal the elongation of C–O and contraction of Mo–CO bond lengths, consistent with the attenuation of the observed CO stretching frequencies in the IR spectra and the significantly deshielded ¹³CO resonances visible by ¹³C{¹H} NMR spectroscopy. Altogether, the experimental evidence points to strongly activated CO ligands, susceptible to electrophilic attack as demonstrated with R₃SiCl reagents (R = Me, iPr). In fact, 1005 and 1006 react with four equivalents of iPr₃SiCl to yield the CO functionalized product iPr₃SiOC≡CSiiPr₃ and compound 1003 in 50% and 75% yields, respectively. This scenario suggested a redox imbalance, where 1005 or 1006 also acted as sacrificial reductants. Addition of KC₁₀H₈ and iPr₃SiCl to the mixture of 1006 and iPr₃SiCl led to almost quantitative formation of 1004, along with products iPr₃SiOC≡CSiiPr₃ and iPr₃SiOSiiPr₃, with no apparent formation of 1003.

The same authors presented a detailed mechanistic study for this intricate reaction sequence, establishing the elementary reaction steps involved in the CO cleavage and coupling at a single molybdenum site.⁶⁰⁵ These works rationalize the design principles for CO activation at monometallic complexes. In connection with the reductive catenation of C1 oxygenates, Agapie reported the anionic carbide K[Mo(tpdp)(C)(CO)] (1007), which is an open-shell species.⁶⁰⁶

Within the remarkable chemistry of the Mo-tpdp system, in 2018 the Agapie group also discovered that the low-valent anionic nitride complex M[Mo(tpdp)(N)] (1008, M = Na, K) undergoes N-atom transfer upon addition of CO, affording a cyanate anion and a mixture of neutral carbonyl complexes [Mo(tpdp)(CO)_n] [n = 1 (1009) or 3 (1010); Scheme 143].⁶⁰⁷ Two-electron reduction of the Mo(IV) precursor [Mo(tpdp)(N)(OTf)] (1011) yielded the nitride compound 1008. Characterization by multinuclear NMR spectroscopy and X-ray diffraction indicated the presence of a terminal nitride ligand, an η⁶ Mo–arene binding mode to the central aromatic ring of the ligand scaffold, and one dissociated phosphine arm of the terphenyl chelate. For K[1008] a tetrameric N₄K₄ cubane arrangement was observed in the solid-state molecular structure, which precluded Mo–Mo bonding (Mo–Mo distance is ca. 6.5 Å). As seen with [K₂(thf)₃][Mo(tpdp)(CO)₂] (1005) and [K₃(thf)_2.5][Mo(tpdp)(CO)₂] (1006),⁶⁰⁴ the potassium cation interacts with the π-system in the ligand scaffold and with the nitride unit.

Scheme 143. Synthesis of a Reduced Anionic (Nitride)molybdenum Complex and N^– Group Transfer Reactivity⁶⁰⁷.

The electronic structure of the anion 1008 was investigated by DFT, revealing the important role of the flexible terphenyl diphosphine ligand in stabilizing this unusual terminal nitrido complex with high formal d electron count. The hemilability of tpdp allows for variable binding modes that prevent the population of antibonding orbitals with respect to the Mo–N interaction. DFT calculations also indicated that the lone pair on the nitride is relatively high in energy (HOMO–2), making it nucleophilic, in agreement with the experimental observation of its carbonylation in solution under an atmosphere of CO. This work provides a useful strategy for promoting productive group transfer reactivity in stable early transition-metal nitrides. Recently, Mazzanti, Maron, and Agapie showed that the nitride ligand in Na[1008] is partially transferred to the uranium(III) precursor [U(OSitBu₃)₃(thf)₂] (1012), affording the first example of a transition metal-capped uranium nitride Na[U(OSitBu₃)₃(μ-N)Mo(tpdp)] (1013).⁶⁰⁸ The experimental and computational data point to an N-transfer accompanied by a two-electron transfer from uranium to molybdenum, formally resulting in the nitride triply bonded to U(V) and singly bonded to Mo(0). The isolated compound 1013 displays a U–Mo interaction in the solid state.

3.3.2. CO₂ Activation by Metalates Containing Sc, Mn, Fe, Co, and Ni

Reactions of CO₂ with [K([18]crown-6)][Sc{N(SiMe₃)₂}₃] (419, Scheme 51, vide supra) afford mixtures of the oxalate complex, {K₂([18]crown-6)₃}{[{(Me₃Si)₂N}₃Sc]₂(μ-C₂O₄-κ¹O:κ¹O″)} (1014), and a CO₂^– radical anion complex, [{(Me₃Si)₂N}₃Sc(μ-OCO-κ¹O:κ¹O′)K([18]crown-6)]_n (1015, Scheme 144).³²⁶ Complex 1015 is similar to [K₂([18]crown-6)₂]{[(μ-CO₂)({(Me₃Si)₂N})₃Y(μ-CO₂)] (1016), which was previously obtained from the reaction of [Y{N(SiMe₃)₂}₃]/K with CO₂ in the presence of [18]crown-6.⁶⁰⁹

Scheme 144. CO₂ Activation by Scandate(II) Amide Complex 419.

Manganese complexes are useful alternatives to the rhenium analogues for the electrochemical reduction of CO₂.^610−613 In studies on the electrochemical reduction of CO₂ to CO using manganese catalysts, Kubiak and co-workers identified several anionic species as key intermediates. The system formed by catalyst fac-[MnBr(tbpy)(CO)₃] (1017, tbpy = 4,4′–di–tert–butyl–2,2′–bipyridine) in a Brønsted acid (2,2,2-trifluoroethanol, TFE, 1.4 M) is selective for the production of CO from CO₂ with TOF of 340 s^–1.⁶¹⁰ In this system, three distinct manganese species were detected, in different oxidation states. The first corresponds to catalyst 1017, which undergoes one-electron reduction with loss of the bromide ligand to afford a Mn–Mn dimer, [Mn(tbpy)(CO)₃]₂ (1018). Further reduction leads to an anionic complex, [Mn(tbpy)(CO)₃]⁻ (1019), which was identified as the catalytically active species.

The structure of the anionic complex was confirmed through X-ray crystallography, facilitated by direct synthesis of [K([18]crown-6)][1019] by reduction of 1017 with KC₈. The complex features a manganese center in a pentacoordinate environment with a partially reduced bipyridine ligand. Moreover, substantial metal-to-ligand backbonding to the carbonyl units is evidenced by the low energy CO stretching frequencies observed by infrared spectroelectrochemistry [IR-SEC, ν_CO = 1907 and 1807 cm^–1]. These findings indicate that the behavior of the manganese catalyst is comparable to that of previously studied rhenium systems,^{171,614−616} which have shown remarkable activity in the electrochemical reduction of CO₂.

To circumvent the problem of the dimerization of the one-electron reduction product, the same group developed a related manganese complex featuring a bulkier bipyridine ligand, 6,6′-dimesityl-2,2′-bipyridine (mesbpy).⁶¹¹ Like in the previous case, the catalyst fac-[MnBr(mesbpy)(CO)₃] (1020) is electrochemically reduced to produce a singly and a doubly reduced species, [Mn(mesbpy)(CO)₃] (1021) and [Mn(mesbpy)(CO)₃]⁻ (1022) respectively. The formation of both species occurs at approximately the same potential. By contrast to the tbpy-based system, the neutral species 1021 does not dimerize owing to the steric protection provided by the bulky ligand. The reduced compounds, 1021 and 1022 were synthesized by reaction with KC₈ and characterized in the solid state [ν_CO = 1917 and 1815 cm^–1]. The anionic species 1022 binds CO₂, yielding a Mn(I)-COOH intermediate in the presence of H⁺, which requires further reduction to initiate the catalytic production of CO. The high selectivity and improved catalytic activity for CO₂ reduction in the presence of elevated concentrations of acid were partly explained by the noninnocence of the bpy ligand: the electron density on the bpy ligand favors the charge transfer to CO₂ through both σ and π interactions. Furthermore, a 10-fold increase in the reaction rate was achieved by the addition of Mg²⁺ cations as a Lewis acid instead of TFE.⁶¹² The resulting intermediate is capped by a Mg²⁺ cation, weakening a C–O bond in the coordinated CO₂ unit. Upon addition of a second CO₂ molecule the activated C–O bond is broken and CO₃^2– is released as MgCO₃. This generates a cationic Mn(I)-tetracarbonyl species, electrochemical reduction of which releases CO and regenerates 1022.

In the search of a similar system featuring redox-inactive ligands, the mixed carbonyl-isocyanide complexes BrMn(CO)₃(CNAr^Dipp2)₂ (40) and [Mn(thf)(CO)₂(CNAr^Dipp2)₃]OTf (1023) were investigated by Kubiak, Figueroa, and co-workers.⁶¹³ While the analysis of the electronic structure of the bipyridine-based systems could be complicated by redox events at the ligand, the electronic description of the mixed-ligand complexes 40 and 1023 is comparably straightforward (vide supra, section 2.2.3.1).^121,122 The pentacoordinate anionic intermediate [Mn(CO)₃(CNAr^Dipp2)₂]⁻ (39) is a stable analogue of the homoleptic carbonyl complex [Mn(CO)₅]⁻ (37), and the redox chemistry of these mixed-ligand complexes occurs at the metal center. The electrochemical behavior of 40 (and that of the analogous Cl^– and I^– complexes) toward CO₂ was consistent with the reactivity observed for the anionic compound 39 with CO₂. Stoichiometric reaction between 39 and CO₂ afforded the zerovalent compound [Mn(CO)₃(CNAr^Dipp2)₂] (49, Scheme 3, section 2.2.3.1, vide supra), dimers Mn₂(CO)₇(CNAr^Dipp2)₃ (1024) and Mn₂(CO)₈(CNAr^Dipp2)₂ (1025, see Scheme 145a), free isocyanide ligand and unreacted 39. Nonetheless, by using a 1:10 ratio of 39:CO₂, complete consumption of 39 with formation of CO₃^2– and dimers 1024 and 1025 was observed. Therefore, metalate 39 (2 equiv) promoted the reductive disproportionation of CO₂.

Scheme 145. (a) Structures of Dimers Mn₂(CO)₇(CNAr^Dipp2)₃ (1024) and Mn₂(CO)₈(CNAr^Dipp2)₂ (1025) and (b) Proposed Mechanism of Reductive Disproportionation of CO₂ Promoted by Metalate Na[Mn(CO)₃(CNAr^Dipp2)₂] (39)⁶¹³.

A mechanism for the transformation of CO₂ to CO and CO₃^2– by metalate 39 was described (Scheme 145b). According to this proposal, complex 39 reacts with CO₂ to form a metallocarboxylate intermediate, Na[Mn(CO₂)(CO)₃(CNAr^Dipp2)₂] (int-1026), which conducts a nucleophilic attack on another molecule of CO₂, forming a bound −CO(O)CO₂ ligand at the Mn center (int-1027). Int-1027 decomposes into CO₃^2–, generating a cationic tetracarbonyl complex, [Mn(CO)₄(CNAr^Dipp2)₂]⁺ (int-1028). This species is reduced, affording the zerovalent compound 49 with release of CO. In an independent reaction between compound 49 and CO, the rapid formation of the dimers 1024 and 1025 was identified. Therefore, 49 was proposed to act as a trap for the CO formed in the reductive disproportionation of CO₂. Due to this generation of various off-cycle species, the system cannot engage in catalytic turnover.

In another example of this reactivity, the bulky tetraisocyanide dianion Na₂[Fe(CNAr^Mes2)₄] (51; Scheme 4, vide supra)¹²³ activates CO₂ to produce CO and carbonate (CO₃^2–), via reductive disproportionation.⁶¹⁷ Treatment of a solution of metalate 51 with 1 atm of CO₂ yielded a new diamagnetic species. Crystallographic analysis revealed it to be the carbonyl complex Fe(CO)(CNAr^Mes2)₄ (1029, Scheme 146). The reactivity toward CO₂ parallels that of the carbonyl dianion complex Na₂[Fe(CO)₄] (Na₂[1]), which affords Fe(CO)₅ (18).^598,599 Moreover. the five-coordinate complex 1029 is isostructural with the previously described dinitrogen complex Fe(N₂)(CNAr^Mes2)₄ (61, Scheme 5, vide supra).¹²³

Scheme 146. Reactivity of the Dianionic Complex Na₂[Fe(CNAr^Mes2)₄] (51) toward CO₂⁶¹⁷.

By treating 51 with CO₂ in the presence of silyl triflates (R_nMe_3-nSiOTf), a silylated CO₂ adduct was obtained, Na[Fe≡CN{(Ar^Mes2)(C(O)OSiMe₃)}(CNAr^Mes2)₃] (1030, for Me₃SiOTf, see Scheme 146), which presents as a contact-ion pair. X-ray crystallography indicated that this four-coordinate product is a terminal iron carbyne compound (Fe–C = 1.658(10) Å), in which the C_Carbyne atom is substituted by an aryl carbamate group. This product is the formal result of silylative trapping of a molecule of CO₂. The −C(O)OSiR₃ group directly binds to an isocyano nitrogen atom, prior to electronic rearrangement. DFT calculations performed on a truncated model of the carbyne complex reproduced the experimental structure. Furthermore, the formation of the carbyne product seems in response to the bulk of the silyl triflate reagent: while treatment with PhMe₂SiOTf also exclusively afforded the carbyne-carbamate Na[Fe≡CN{(Ar^Mes2)(C(O)OSiMe₂Ph)}(CNAr^Mes2)₃] (1031), the reaction with the more bulky Ph₂MeSiOTf led to the corresponding product Na[Fe≡CN{(Ar^Mes2)(C(O)OSiMePh₂)}(CNAr^Mes2)₃] (1032) in only 10% yield, along with significant amounts of the carbonyl compound 1029. This disparity was attributed to changes in the steric profile of the silane, making it kinetically inefficient to trap the CO₂ adduct of 51. In the absence of the silane, the CO₂ adduct engages in a traditional mechanism of reductive disproportionation, thereby producing the carbonyl complex 1029.

As previously mentioned (section 2.2.6.1), Müller, Wolf and co-workers reported an anionic iron complex featuring a pyridyl-phosphinine motif, namely [K([18]crown-6)][Cp*Fe(L)] [323, L = 2-(2′-pyridyl)-4,6-diphenylphosphinine], which exists as two isomeric forms in solution depending on the coordination mode adopted by the ligand: 323-π and 323-σ (vide supra).²⁷⁴ Given the coordinative flexibility and noninnocence of the ligand in 323-π and 323-σ, the reactivity of this species (in solution) toward CO₂ was investigated. An immediate color change was observed, accompanied by the replacement of the resonances of 323-π and 323-σ by two new resonances in the ³¹P{¹H} NMR spectrum (δ = 97.0 and 116.0 ppm), suggesting that both isomers had reacted with CO₂. Fractional crystallization allowed the structural characterization of both species.

Isomer 323-σ reacted with CO₂ to generate complex 1033 (Scheme 147), in which one C=O bond is fully cleaved at room temperature. The cleavage of the CO₂ molecule thereby generated an Fe-CO species and a P–O moiety at the σ-coordinated bidentate ligand. The continued coordination of the bidentate ligand to the metal center was attributed to electronic flexibility of the phosphinine moiety. The π-coordinated isomer, 323-π, reacted with CO₂ to give complex 1034 (Scheme 147) the product of addition of an intact CO₂ molecule to a carbon atom of the phosphinine unit. The resulting ligand can be described as a carboxylate-substituted phosphacyclohexadienyl motif, which coordinates the metal center in an η⁵-fashion. The carboxylate unit in the ligand is κ²-coordinated to the potassium cation (not shown in Scheme 147). For the reaction of both isomers with CO₂, DFT calculations indicated that only small energy barriers must be overcome to reach the products of activation (323-σ → 1033 = 3.5 kcalmol^–1; 323-π → 1034 = 5.5 kcalmol^–1). In both cases, these barriers are considerably lower than that for the isomerization between 323-σ and 323-π (27.0 kcalmol^–1, vide supra).²⁷⁴

Scheme 147. Activation of CO₂ by the Anionic Species [K([18]crown-6)][Cp*Fe(L)] (323)²⁷⁴.

The anionic species are stabilized by [K([18]crown-6)]⁺ cations.

The highly reduced Mabiq complex Na(OEt₂)[Fe(Mabiq)] (1035, Mabiq = 2–4:6–8-bis(3,3,4,4-tetramethyldihydropyrrolo)-10–15-(2,2′-biquinazolino)-[15]-1,3,5,8,10,14-hexaene1,3,7,9,11,14-N₆) developed by Hess and co-workers is a noteworthy precatalytic intermediate in the selective electrocatalytic reduction of CO₂ to CO.⁶¹⁸ Detailed spectroscopic and computational studies indicate that the complex features an intermediate spin Fe(II) center coupled to a ligand biradical, resulting in a S = 1 spin state for the complex. Since the metal atom does not feature a “low-valent” metal center, we would like to refer the reader to the original literature for further details on the promising catalytic properties and intricate reaction mechanism as well as the closely related Co-Mabiq catalysts.⁶¹⁹

As described in section 3.1, remarkable heterobimetallic catalysts for CO₂ hydrogenation were developed by Lu and co-workers based on metalates involving group 13 elements and Ni and Co.^522−525 The reactivity of the nickelates [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [M = Al (766), Ga (767)] toward carbon dioxide was also studied.^78,620 These complexes reacted with CO₂ (1 atm) within minutes, forming K₂CO₃ and a 1:1 mixture of the corresponding neutral compound [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [M = Al (762), Ga (763)] and carbonyl complex [(CO)NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] [M = Al (1036), Ga (1037); see Scheme 148]. The presence of carbonyl ligands was evidenced by infrared spectroscopy [ν_CO(1036) = 1953 cm^–1; ν_CO(1037) 1962 cm^–1]. These observations suggested that the CO₂ molecule underwent two-electron reduction, promoted by the anionic complex [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (2 equiv), to generate CO₃^2– and CO, with the latter being trapped by the corresponding neutral compound [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}] to afford 1036 or 1037.

Scheme 148. Reactivity of the Heterobimetallic Nickelates [NiM{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ [M = Al (766), Ga (767)] toward Carbon Dioxide⁶²⁰.

Complexes 1036 and 1037 were independently synthesized by treating the neutral compounds 762 or 763 with ethyl formate and paraformaldehyde, respectively (Scheme 148). In the proposed mechanism for the reductive disproportionation of CO₂ to carbonate and carbon monoxide, the anionic complexes 766/767 react with CO₂ to produce a Ni^I-metallacarboxylate, which is further reduced by another equivalent of 766/767, affording a formally dianionic nickel diolate. Insertion of a second molecule of CO₂ produces a Ni intermediate with a di-CO₂ arrangement (Ni–C(OK)–O–C(O)-OK), which decomposes into CO₃^2– and a bound CO unit in 1036/1037. The stability of the carbonyl compounds unfortunately hampers a possible turnover via CO release.

Inspired by the structure of the C-cluster of the Ni carbon monoxide dehydrogenase (CODH), Lu and co-workers developed a related family of nickel–iron heterobimetallic compounds, including reduced species that engage in the reductive disproportionation of CO₂ to carbonate and carbon monoxide.⁶²¹ The NiFe complex [NiFe{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (1038) reacted with KC₈ to afford the reduced complex [K(thf)₃NiFe{N[o-(NCH₂PiPr₂)C₆H₄]₃}] (K[1039], Scheme 149a). Crystallographic analysis revealed the formation of a contact ion pair by interaction of the K(thf)₃⁺ unit with two phenyl rings from the ligand. An analogue with a noninteracting cation was obtained by performing the reduction in the presence of the phosphazenium cation, [P(NMeCy)₄]⁺, which afforded [P(NMeCy)₄][NiFe{N[o-(NCH₂PiPr₂)C₆H₄]₃}] ([P(NMeCy)₄][1039]).

Scheme 149. (a) Bioinspired Heterobimetallic Nickel–Iron Complexes and (b) Their Reactivity toward Reductive Disproportionation of CO₂⁶²¹.

The carbonyl derivatives of both the neutral and the reduced species [(CO)NiFe{N[o-(NCH₂PiPr₂)C₆H₄]₃}]ⁿ [n = 0 (1040), 1– (1041), Scheme 149a] were also synthesized by direct treatment of either 1038 or 1039 with CO, contrasting with the analogous treatment of 762 or 763 which afforded intractable mixtures.⁶²⁰ Furthermore, addition of LiHBEt₃ to complex 1040 also yields the anionic carbonyl complex. The reduced nature of the anionic complex is reflected in the FT-IR spectrum [ν_CO(1040) = 1954 cm^–1; ν_CO(1041) 1930 cm^–1]. Crystallographic, spectroscopic (NMR, EPR, UV–vis, ⁵⁷Fe Mössbauer) and DFT analyses indicated that the nickel center is zerovalent, with the reduction events exclusively affecting the iron center in the supporting ligand, Fe(III)→Fe(II). The Ni⁰Fe^II anionic species 1039 reductively disproportionates CO₂ into carbonate and metal-bound CO as shown in Scheme 149b. A CO₂-bound species, proposed to be an intermediate, was identified through electrochemical investigations. The neutral complex 1038 did not behave similarly toward CO₂.

3.4. Activation of White Phosphorus

3.4.1. General Remarks on White Phosphorus Activation

The reports on the activation and functionalization of P₄ by transition metal complexes, including examples by metalates, have been extensively reviewed.^{149,622−624} Thus, this review will focus only on the most recent and relevant contributions to this area that have been reported since the year 2020 and which were therefore not covered by the aforementioned review. Prior examples will be discussed in a generalized manner for context.

One early landmark example is the decaphosphatitanocene compound [K([18]crown-6)]₂[Ti(η⁵-cyclo-P₅)₂] (1042, Scheme 150) reported by Ellis and co-workers.⁶²⁵ This complex, which was the first example of an entirely inorganic, “carbon-free” all-phosphorus metallocene, was obtained by treatment of a highly reactive naphthalenetitanate(−II) (150, generated in situ by reduction of TiCl₄(thf)₂ with potassium naphthalenide⁶²⁶) with P₄ (2.5 equiv). The resulting metallocene has a formally zerovalent titanium center.

Scheme 150. Synthesis of the Carbon-Free Sandwich Complex [(cyclo-P₅)₂Ti]^2– (1042) by Ellis and Co-workers⁶²⁵.

Considering the variety of polyphosphorus ligands obtained by interaction of the P₄ tetrahedron with transition metals, a useful way to organize the contributions in this area is by the degree of degradation or aggregation at the metal center. Examples of the reactivity of white phosphorus with metalloradicals are beyond the scope of the present article. However, these have been reviewed elsewhere.^149,622

3.4.2. P₄ Activation by Ferrates

Examples of products of activation of P₄ at low-valent iron metalates are summarized in Figure 24. These results show that reduced P₂, P₄, and P₇ fragments are accessible from the P₄ molecule and low-valent ferrate complexes.

Figure 24.

Products of activation of P₄ (degradation) at low-valent iron complexes/ferrates.

A recent example of the generation of a reduced P₄ ligand was published by Lammertsma and co-workers, who isolated the complexes Li[Cp*Fe(CO)₂(η¹-P₄·BAr₃)] (1043, Figure 24)⁶²⁷ by reaction of the nucleophilic iron metalate, Li[Cp*Fe(CO)₂] (1044) with white phosphorus in the presence of Lewis acids (BAr₃; Ar = C₆F₅, Ph). The outcome of the reaction resembles that of P₄ with other organic or p-block element-based nucleophiles.⁶²⁸ A bicyclo[1.1.0]tetraphosphabutanide anion with a “butterfly” structure is obtained, which has been trapped by the Lewis acid, generating a metalphosphido-borane interaction. 1043 can be protonated at its nucleophilic P site, with loss of the borane, to yield metal-substituted η¹-P₄H ligands, though these decompose within 24 h.

Wolf and co-workers reported the formation of a tetraphosphacyclobutadiene (P₄^2–) ligand, as well as P₄ chains coordinated to iron centers.⁶²⁹ Reduction of [Cp^RFe(μ-Br)]₂ (1045, Cp^R = C₅(C₆H₄-4-Et)₅) with potassium naphtalenide (KC₁₀H₈, 4 equiv), followed by addition of P₄ (2 equiv), afforded the cyclo-P₄ compound [{Cp^RFe(η⁴-cyclo-P₄)}]⁻ (1046, Figure 24). In the solid state, this anion forms a contact ion pair with the [K([18]crown-6)]⁺ cation. DFT calculations performed on a truncated model confirmed that, in 1046, an iron center in a d⁶ configuration is coordinated to a delocalized planar cyclo-P₄ unit. By contrast, reduction of 1045 with an excess of sodium amalgam in the presence of P₄ yielded the highly air sensitive catenated species [Na₂(thf)₅{(Cp^RFe)₂(μ,η^4:4-P₄)}] (1047, Figure 24). While 1047 is readily protonated by Et₃N·HCl, [H(Et₂O)₂][BAr^F₄], or trace amounts of water, even simple dissolution of the complex in THF slowly results in decomposition to give the protonated ferrate [Na(thf)₃{(Cp^RFe)₂(μ,η^4:4-P₄)(H)}] (1048, Figure 24). Reactions of 1047 with suitable electrophiles, such as an excess of Me₃SiCl, resulted in the functionalization of P₄ chains with release of the functionalized polyphosphido fragments from the metal center (mixture of P₇(SiMe₃)₃, P(SiMe₃)₃, and PH(SiMe₃)₂ in a 10:1:1 ratio).

Heterobimetallic complexes with highly reduced P₄ ligands, derived from iron and cobalt metalates, were reported in 2020 by Wolf, Weigand and co-workers.⁶³⁰ The metalates [M(η⁴-C₁₄H₁₀)₂]⁻ [M = Fe (52); Co (14)] reacted with the gallium tetraphosphido species [(nacnac)Ga(κ²-P₄)] (1049, nacnac = CH[CMeN(2,6-iPr₂C₆H₃)]₂), which contains an activated (butterfly) P₄ unit.⁶³¹ The strategy afforded the heterobimetallic complexes [(η⁴-C₁₄H₁₀)M(μ–κ⁴:κ²-P₄)Ga(nacnac)]⁻ [M = Fe (1050), Co (1051); Figure 24]. These compounds contain a bridging catena-P₄ ligand in which the transition metal centers are coordinated to the four phosphorus atoms, and the gallium center only coordinates the two terminal P atoms of the open chain. The reaction of cobaltate 14 with an excess of 1049 did not result in the substitution of both anthracene ligands, affording only 1051.

Notably, an early, related example of reaction between a ferrate complex and white phosphorus was reported for the heteroleptic metalate [K([18]crown-6){Cp*Fe(η⁴-C₁₀H₈)}] (201).²¹⁶ By treating 201 with white phosphorus, a metal-mediated fragmentation-reaggregation of the P₄ tetrahedron occurs, and two major products are isolated after fractional crystallization: the anionic cluster [K([18]crown-6)(thf)₂][(Cp*Fe)₃(P₃)₂] (1052, Figure 24) and the mononuclear complex [{K([18]crown-6)}₂(Cp*FeP₇)] (1053, Figure 24). Compound 1053 is the product of aggregation of white phosphorus, and showcases a P₇ norbornadiene-like framework, whereas in cluster 1052 the fragmentation of P₄ leads to two cyclo-P₃ scaffolds binding three Cp*Fe units.

A related contribution by Driess and co-workers reported the fragmentation of P₄ at a low-valent β-diketiminato-iron(I) complex, [(nacnac)Fe·toluene] (1054, nacnac = CH[CHN(2,6-iPr₂C₆H₃)]₂).⁶³² The selective cleavage of the P₄ tetrahedron at 1054 afforded the neutral iron(III) compound [{(nacnac)Fe}₂(μ₂:η²,η²-P₂)₂] (1055, see Figure 24), bearing two bridging dumbbell P₂^2– ligands. Reduction of 1055 by metallic potassium afforded the rare mixed-valence iron(II,III) anion [{(nacnac)Fe}₂(μ₂:η²,η²-P₂)₂]⁻ (1056), isostructural with 1055. DFT calculations along with Mössbauer spectroscopic studies support a delocalized mixed-valent configuration for the iron centers.⁶³²

3.4.3. P₄ Activation Mediated by Cobaltates

A summary of the products of P₄ activation at low-valent cobalt complexes, reported up to late 2020,^{149,622−624} is illustrated in Figure 25. Several examples involve the use of diimine/diketimine ligand scaffolds.^383,385,633 For instance, both the highly reduced cobaltate [K(OEt₂)₂{Co(^DippBIAN)(η⁴-1,5-cod)}] (540) and the analogous complex [K([18]crown-6)(thf)_1.5][Co(PHDI)(η⁴-1,5-cod)] (542; Scheme 65, above) were suitable platforms to activate white phosphorus. These complexes reduced P₄ to cyclo-P₄ moieties with different oxidation states.^383,385540 reacted with P₄ to yield the dianionic complex [(^DippBIAN)Co}₂(μ-η⁴:η⁴-P₄)]^2– (1057).³⁸³ X-ray diffraction analysis on 1057 revealed that the P₄ unit in this complex is best described as a superposition of two mesomeric structures consisting of two P₂^2– dumbbell units and a P₄^4– motif. Oxidation of compound 1057 by [Cp₂Fe]BAr^F₄ resulted in monoanionic and neutral tetraphosphido complexes to complete the series [(^DippBIAN)Co}₂(μ-η⁴:η⁴-P₄)]ⁿ⁻ [n = 2 (1057), n = 1 (1058), n = 0 (1059); Figure 25 shows only the dianionic species 1057], which feature either cyclic P₄^4– units or open-chain P₄^4– ligands. Thus, the oxidation process, and the strength of the interaction with the potassium counterion, impact the structure rather than the oxidation state of the polyphosphorus ligand. The presence of the redox-active BIAN ligand appears crucial to achieve a high degree of reduction of the P₄ units.

Figure 25.

Complexes obtained by activation of white phosphorus at low-valent cobalt compounds.

By contrast, P₄ activation by the cobaltate 542 results in a mononuclear complex [(PHDI)Co(η⁴-P₄)]⁻ (1060, Figure 25).³⁸⁵ Structural authentication for complex 1060 and metric parameters of the P₄ ligand were obtained through the adduct [K([18]crown-6)][(PHDI)Co(μ²:η¹,η⁴-P₄)W(CO)₅] (1061). The almost perfectly square cyclo-P₄^2– unit observed in the molecular structure is best described as featuring six delocalized π-electrons. The bond lengths of the motif lie between the values expected for P–P single and P=P double bonds (av. 2.147(7) Å). Selective functionalization of the cyclo-P₄ ring in complex 1060 with diorganochlorophosphanes resulted in the quantitative formation of the corresponding neutral ring-expansion η⁴-cyclo-P₅R₂ products. The latter species react with cyanide salts (2 equiv) to fragment the cyclo-P₅R₂ into P₃ and P₂ units, namely the anionic (cyclo-P₃)cobaltate complex [(PHDI)Co(CN)(cyclo-P₃)]⁻ (1062, Figure 25) and a series of 1-cyanodiphosphan-1-ide anions.

A similar planar cyclo-P₄ unit and a cyclo-P₃ complex were obtained by activation of P₄ at the formal dicobalt(0) complex [K₂{(nacnac)Co}₂(μ₂:η¹,η¹-N₂)] (1063, nacnac = CH[CMeN(2,6-Me₂C₆H₃)]₂).⁶³³ The products were the monoanionic compounds [{(nacnac)Co}₂(μ₂:η⁴,η⁴-P₄)]⁻ (1064) and [{(nacnac)Co}₂(μ₂:η³,η³-P₃)]⁻ (1065). Extrusion of a P atom from the P₄ ligand of 1064 leads to the constitution of the cyclo-P₃ motif observed in 1065. 1064 can also be accessed by reduction of its neutral congener, [{(nacnac)Co}₂(μ₂:η⁴,η⁴-P₄)] (1066), prepared from a suitable Co(I) precursor. Analogously, thermally induced extrusion of a P atom from 1066 yields the neutral congener of 1065. Magnetic measurements (Evans method and SQUID) indicated that 1064 is a mixed-valent Co^I–Co^II species, while in 1065 the cyclo-P₃^3– unit bridges two Co(II) centers.⁶³³ Driess and co-workers described the synthesis of further examples of cobalt complexes featuring cyclo-P₄^2– motifs, similar to 1064. The two-step procedure first involves the activation of P₄ at Co(I) precursors featuring related diketimine ligands (nacnac = CH[CHN(2,6-iPr₂C₆H₃)]₂ or CH[CMeN(2,6-Et₂C₆H₃)]₂), to generate analogous dinuclear cobalt compounds with a neutral tetraphosphacyclobutadiene bridging ligand. The latter species were then reduced by KC₈ to afford [{(nacnac)Co}₂(μ₂:η⁴,η⁴-P₄)]⁻ (1067/1068).⁶³⁴Figure 25 shows the generic structure of complexes 1064, 1067 and 1068 and the structure of compound 1065.

The complexes [K(OEt₂){Co(^ArBIAN)(η⁴-1,5-cod)}] [^ArBIAN = ^DippBIAN (540), ^MesBIAN (541)] were treated with [(nacnac)Ga(κ²-P₄)] (1049) by a synthetic protocol similar to that which yielded [(η⁴-C₁₄H₁₀)M(μ-η⁴:κ²-P₄)Ga(nacnac)]⁻ [M = Fe (1050), Co (1051); Figure 25, vide supra].^384,630 This procedure afforded the heterodinuclear cobalt–gallium compounds [K(dme)₂{(^ArBIAN)Co(μ-η⁴:κ²-P₄)Ga(nacnac)}] [^ArBIAN = ^DippBIAN (1069), ^MesBIAN (bis(mesitylimino)acenaphthenediimine) (1070)], featuring catena-P₄ ligands. Reactivity studies performed on 1070 showed that these complexes can be functionalized with alkylchlorophosphanes (R_nPCl_3–n) to afford polyphosphorus cyclo-P₅R₂ compounds. In the case of the dialkylchlorophosphanes R₂PCl₂ (R = iPr, Cy, and tBu) the functionalization is accompanied by release of the Ga(nacnac) fragment.

Since late 2020,^{149,425,475,622−624} further reports on the activation of P₄ by low-valent cobalt complexes have appeared. As previously mentioned, Hey-Hawkins, Wolf, and co-workers, described the reactions of the bis(phosphide)carborane complex [K([18]crown-6)(thf)][Co{1,2-(PMes)₂C₂B₁₀H₁₀}(η⁴-cod)] (rac-330) with cyclohexyl isocyanide (Cy-N≡C) and tert-butylphosphaalkyne (tBu-C≡P), which afford the 16 VE bis(isocyanide) and diphosphacyclobutadiene complexes rac-333 and rac-334, via ligand substitution (Scheme 42; section 2.2.6.2, vide supra).¹³² The reactivity of rac-330 toward white phosphorus was also evaluated. Depending on the rac-330:P₄ ratio used, two distinct products were obtained (Scheme 151). A 1:1 reaction afforded complex [K([18]crown-6)][meso-1071] (Scheme 151), resulting from the fragmentation of the P₄ tetrahedron into P₃ and P₁ motifs.¹³² This behavior contrasts, for example, with the formation of cyclo-P₄ rings at the similar anionic species [Co(^DippBIAN)(η⁴-1,5-cod)]⁻ (540) or [Co(PHDI)(η⁴-1,5-cod)]⁻ (542), bearing nitrogen-chelates.^383,385

Scheme 151. White Phosphorus Activation at Complex [K([18]crown-6)(thf)][Co{1,2-(PMes)₂C₂B₁₀H₁₀}(η⁴-cod)] (rac-330; Cations Are Omitted for Clarity)¹³².

The Co(I) complex [meso-1071]⁻ is a 18 VE species containing a cyclo-P₃ and a η³-triphospholanido ligand, probably resulting from intramolecular abstraction of a P⁺ fragment by the nucleophilic phosphanido moieties. Theoretical calculations provided a detailed explanation on the coordination and inner-sphere fragmentation of the P₄ molecule at rac-330, and support the proposed abstraction by the phosphanido ligand. In contrast, by using a 2:1 rac-330:P₄ ratio and adding [18]crown-6, the reaction product is a bimetallic complex, [K([18]crown-6)]₂[rac-1072] (Scheme 151), bearing a bridging catena-P₄ ligand, μ–η⁴:η⁴-P₄^2–, i.e., a cis-1,3-tetraphosphabutadienediide motif. The short intermetallic distance (2.651(1) Å) indicates the presence of a Co–Co single bond. BS-DFT calculations revealed that [rac-1072]^2– contains two low-spin 3d⁷ Co(II) atoms, which are antiferromagnetically coupled via one of the terminal P atoms of the P₄^2– chain. By means of time-resolved ³¹P{¹H} NMR spectroscopic monitoring, it was established that [rac-1072]^2– results from the reaction of [meso-1071]⁻ with a second equivalent of the bis(phosphanido)cobaltate rac-330.¹³²

Wolf and co-workers studied the activation of P₄ by a heteroleptic magnesium cobaltate sandwich complex.⁴²⁵ Treatment of [(^Depnacnac)MgCo(P₂C₂tBu₂)(η⁴-cod)] (688, vide supra) with white phosphorus (P₄) in toluene (75 °C, 18 h) led to the formation of the cyclo-P₄ sandwich complex [(^Depnacnac)MgCo(cyclo-P₄)(P₂C₂tBu₂)]₂ (1073, Scheme 152) via substitution of the 1,5-cyclooctadiene ligand.

Scheme 152. Synthesis of a Cyclo-P₄ Heteroleptic Sandwich Complex from a Magnesium Cobaltate⁴²⁵.

The solid-state structure shows that each magnesium moiety is coordinatively saturated by binding to a P atom of each cyclo-P₄ unit, with no interaction between the magnesium and cobalt centers. Two of the P–P bond lengths of the cyclo-P₄ unit are slightly longer (P1–P2/P1–P4 distances, 2.2083(5)/2.2184(5) Å, respectively) than the other two (P3–P2/P3–P4 2.1641(5)/2.1740(5) Å, respectively). The cyclo-P₄ ligand is similar to the other examples reported with metals such as iron or cobalt.¹⁴⁹ However, the phosphorus rich compound 1073 represents the first example of an anionic cobaltate sandwich compound featuring a cyclo-P₄ ligand, according to the authors.⁴²⁵

Heavy tetraphospholide anions formed at heterometallic cobaltates were later reported by the same group.⁴⁷⁵ These were obtained when the complexes [Ar^Dipp2TtCo(P₂C₂tBu₂)(η⁴-cod)] [Tt = Ge (693), Sn (694), Pb (695)] and Sn[Co(P₂C₂tBu₂)(η⁴-cod)]₂ (696) (Scheme 82, vide supra) reacted with white phosphorus. While reaction between the germanium derivative and white phosphorus was unsuccessful even upon heating, analogous treatment of the tin compounds 694 or 696 with P₄ afforded the triple decker compound [(P₂C₂tBu₂)₂Co₂(μ,η⁵:η⁵-SnP₄)] (1074, Scheme 153) in each case. Using 696 proved to be more efficient, yielding 1074 in a more selective manner. Likewise, 695 reacted with P₄ to the corresponding lead complex [(P₂C₂tBu₂)₂Co₂(μ,η⁵:η⁵-PbP₄)] (1075). The latter can also be accessed in a one-pot reaction between the magnesium cobaltate 688, Pb(acac)₂ and P₄. However, the product in this case is obtained with (^Depnacnac)Mg(acac) (1076) as a contaminant.

Scheme 153. Formation of Heavy Tetraphospholide Anions by Reaction of Heterometallic Cobaltates with White Phosphorus⁴⁷⁵.

In both cases, the triple decker complexes feature a planar cyclo-TtP₄ middle deck, as evidenced through the X-ray characterization. The lead derivative, though, resulted unstable at room temperature, depositing metallic lead, and forming a different triple decker compound, namely the cyclo-P₄ complex [(P₂C₂tBu₂)₂Co₂(μ,η⁴:η⁴-P₄)] (1077). The combined structural and calculated data indicate that the heavy tetraphospholide anions TtP₄^2–, analogous to the known cyclo-P₅ ligand P₅^–, possess a planar structure with almost identical P–P bond lengths, and therefore are aromatic. According to IBO analyses on the TtP₄^2– ring (TPSS D3BJ/def2-SVP level) the delocalized π-system involves the Tt, P1/P3, P2/P4 centers, and its interaction with the transition metal atoms corresponds to a typical π-coordination.⁴⁷⁵

3.4.4. P₄ Activation by Molybdenum and Tungsten Complexes

Anionic polyphosphido complexes based on 4d/5d metals have also been described. For instance, Figueroa and co-workers described the synthesis of low-valent Mo-isocyanide complexes featuring η⁴-end-deck cyclo-P₄ ligands (Figure 26).^149,635,636 The neutral compounds (η⁴-cyclo-P₄)MoI₂(CO)(CNAr^Dipp2)₂ [1078, Dipp = 2,6-(iPr)₂C₆H₃] and (η⁴-cyclo-P₄)Mo(CO)₂(CNAr^Dipp2)₂ (1079) were obtained by activation of the P₄ at isocyanide-Mo precursors.⁶³⁶ These complexes bear the six π-electron [cyclo-P₄]^2– dianion ligand. To further investigate the nature of the aromatic [cyclo-P₄]^2– ligand, the bonding situation of the complexes and the influence of the formal oxidation state of the metal on the ligand, the reduction of the complexes 1078 and 1079 was studied. The Mo(IV) (d²) compound 1078 was treated with an excess KC₈, followed by addition of dibenzo[18]crown-6, to give the low-valent complex {K₂(dibenzo[18]crown-6)}[(η⁴-cyclo-P₄)Mo(CO)(CNAr^Dipp2)₂] (K₂[1080]) after four-electron reduction. The collective characterization data (³¹P NMR, FT-IR, X-ray diffraction) suggested that the cyclo-P₄ ligand remained unaltered after reduction and indicated that the three-legged piano stool complex has a highly reduced metal center, forming a contact ion pair between the Mo-anion and potassium cations.

Figure 26.

4d and 5d metalates featuring polyphosphido ligands.

Similar reduction of complex 1079 with KC₈ (excess) afforded a mixture of complex 1080 and the dicarbonyl compound K₂[(η⁴-cyclo-P₄)Mo(CO)₂(CNAr^Dipp2)] (1081). Upon reduction, compound 1079 loses either one CO ligand to afford complex 1080 or one isocyanide ligand yielding 1081. By treating 1080 with CO (1 atm), the complex was fully converted to 1081. Again, the cyclo-P₄ motif in 1081 has a common electronic structure with 1080 or the parent compounds 1078 and 1079, according to the metric parameters. Furthermore, the increased electron density in 1081 is also evidenced, as it was for 1080. The experimental data and computational calculations support the formulation of both complexes as including aromatic [cyclo-P₄]^2– dianions bound to zerovalent metal fragments.⁶³⁵

As mentioned in section 2.2.6.4, Ruiz and co-workers reported that the triply bonded anions [M₂(η⁵-Cp)₂(μ-PCy₂)(μ-CO)₂]⁻ [M = Mo (358), W (359); vide supra] activate white phosphorus under mild conditions to afford the anionic diphosphorus complexes [M₂Cp₂(μ-PCy₂)(CO)₂(μ-κ²:κ²-P₂)]⁻ [M = Mo (351), W (352), see Figure 26).^302,637 The molybdenum analogue features a Mo₂P₂ tetrahedral core with the P₂ unit bridging the two metal centers. The tungsten analogue could not be isolated due to its extreme air-sensitivity, but indirect structural proof was obtained through characterization of the methylated derivative [W₂Cp₂(μ-PCy₂)(CO)₂(μ-κ²:κ²-P₂Me)] (354).

3.5. Reactions of Transition Metalates with Unsaturated Organophosphorus Compounds

3.5.1. Metalate-Mediated Phosphaalkyne Oligomerization

The metal-mediated cyclodimerization of phosphaalkynes constitutes a useful route to access π-coordinated diphosphacyclobutadiene ligands.⁶³⁸ Early examples of the application of this strategy include the synthesis of complexes [Cp^RM(P₂C₂tBu₂)] (1082, M = Co, Rh, Ir; Cp^R = Cp, Cp*), featuring η⁴-coordinated 1,3-diphosphacyclobutadiene ligands,^639−641 some Ti(0), Mo(0) and Fe(0) species,^642−647 and the first example of a homoleptic bis(1,3-diphosphacyclobutadiene) complex, the neutral [Ni(P₂C₂tBu₂)₂] (1083).^648,649 The steric hindrance on the carbon atom of the parent phosphaalkyne determines whether the 1,2- or 1,3-diphosphacyclobutadiene ligand is obtained, with bulky substituents kinetically favoring the 1,3-isomer.⁶³⁸

In this context, Wolf and co-workers examined the reactions of phosphaalkynes RC≡P (R = tBu, Ad, tPent) with the labile bis(anthracene)metalate(−I) [M = Fe (52); Co (14)] complexes.^85,208,209 The homoleptic diphosphacyclobutadiene complexes [M(η⁴-P₂C₂R₂)₂]⁻ [M = Fe, R = tBu (188), M = Co, R = tBu (189), tPent (190), Ad (191); Scheme 22, vide supra] were isolated, characterized, and their reactivity extensively evaluated (Scheme 154).^85,208,209 As in the case of the parent metalates [M(η⁴-C₁₄H₁₀)₂]⁻ [M = Fe (52); Co (14)], the iron compound 188 is more prone to oxidation than the cobalt analogue 189.⁴⁷³ Cyclic voltammetry studies revealed that upon oxidation of the anionic complexes, the neutral species [M(P₂C₂tBu₂)₂] [M = Fe (1084), Co (1085)] can be obtained, and the process is reversible. The M(0) compounds were isolated by oxidation of the anionic complexes with ferrocenium hexafluorophosphate, [Cp₂Fe]PF₆ (Scheme 154).^211,4731084 is a phosphorus analogue of the elusive sandwich bis(η⁴-cyclobutadiene)iron(0). According to calculations carried out for the simplified diamagnetic 18-electron dianion [Fe(P₂C₂Me₂)₂]^2– (1086′), further reduction of complex 188 is theoretically viable.²⁰⁹ Nonetheless, all attempts to further reduce 188 to form the dianion [Fe(P₂C₂tBu₂)₂]^2– (1086) were unsuccessful. The latter species was not observed by cyclic voltammetry, nor could it be prepared by using strong reducing agents such as potassium graphite or potassium naphthalenide.

Scheme 154. Synthesis of Homoleptic 1,3-Diphosphacyclobutadiene Sandwich Complexes (Top) And Follow-up Chemistry to Yield Either Substituted 1,3-Diphosphacyclobutadiene Ligands (Middle and Bottom Right) or 1,3-Diphosphacyclobutene Derivatives (Bottom Left-Center)^208,209.

O-based substrate = ethers, epoxides, lactones, aldehydes, ketones, isocyanates.

3.5.2. Reactivity of Diphosphacyclobutadiene Cobaltates and Ferrates

Further reactivity studies on the chemistry of 189 afforded a range of functionalized complexes and demonstrated its ability to act as a versatile building block for phosphaorganometallic species. For instance, reaction with electrophiles (see Scheme 154, center) allows for the functionalization of the ligand through one of its phosphorus atoms. Using this methodology, the cobalt hydride complex [Co(P₂C₂tBu₂)₂H] (1087) and the ring-expansion product [Co(P₃C₂tBu₂Ph₂)(P₂C₂tBu₂)] (1088) were obtained.²⁰⁸ Compound 189 reacted with MeI, ClSnPh₃ or Me₃SiCl to afford the corresponding P-substituted ligands, thus forming the neutral complexes [Co(P₂C₂tBu₂R)(P₂C₂tBu₂)] [R = Me (1089), SnPh₃ (1090), SiMe₃ (1091)]. The hydride complex 1087 reacts with lithium organyls (R′Li, R′ = Ph, nBu, tBu) to generate rare 1,3-diphosphacyclobutene complexes [R′ = Ph (1092), nBu (1093), tBu (1094); Scheme 154, bottom left-center]. Reaction of the anionic complex 1093 with Me₃SiCl led to the quantitative formation of [Co(η³-P₂C₂tBu₂HnBu)(η⁴-P₂C₂tBu₂SiMe₃)] (1095), in which both P₂C₂ rings have been functionalized.⁶⁵⁰ In addition, oxygen-containing substrates (ethers, epoxides, lactones, aldehydes, ketones or isocyanates) insert into the Si–P bond of the silyl-derivative 1091, thereby generating new P–O bonds (1096, Scheme 154, bottom right).⁶⁵¹

The phosphorus atom on the diphosphacyclobutadiene ligands of 189–191 can further coordinate other metal centers, acting as metalloligands, to obtain oligonuclear complexes. This type of chemistry is of significance in catalysis and supramolecular chemistry. When the homoleptic complexes 189–191 were treated with metal salts from groups 8–11, different types of interactions were identified.^{210,652−654} The expected σ-coordination of metal cations from group 11 (M = Cu, Ag, Au) to the cobaltates 189–191 as metalloligands (Scheme 155, top), afforded the series of neutral dinuclear compounds [Co(η⁴-P₂C₂R₂ML_n)(η⁴-P₂C₂R₂)] [R = tBu, tPent, Ad; ML_n = Cu(PPh₃)₂ (1097), Ag(PMe₃)_x (1098), Au(PMe₃)₂ (1099)].^652,653

Scheme 155. Chemistry of the Anionic Complexes [Co(η⁴-P₂C₂R₂)₂]⁻ (189–191) toward Coinage Metals: Assembly of Oligonuclear Complexes through σ-Coordination^{210,652−654}.

Furthermore, the phosphorus atom of the diphosphacyclobutadiene units of 190 or 191 served as a platform for the synthesis of [M{Co(η⁴-P₂C₂Ad₂)₂}₂]_x [M = Ag (1100); M = Au (1101)], which exist as coordination polymers in the solid state, and of the trinuclear complexes [Co{η⁴-P₂C₂tPent₂[Ag(PMe₃)_x]}₂] (1102) or [Au{Co(η⁴-P₂C₂R₂)₂}₂]⁻ [R = tBu, Ad (1103)], shown in Scheme 155. This type of complex showcases phosphorus atoms of at least one of the diphosphacyclobutadiene ligands σ-bonded to coinage metals, or connecting two sandwich motifs as in 1103.²¹⁰ It was thereby demonstrated that the coordination of the phosphaorganometallic ligand can occur through more than one of its phosphorus atoms.^652,653 Being able to functionalize multiple phosphorus atoms on 189 was also crucial in the self-assembly of the unprecedented octanuclear molecular square [Au{Co(η⁴-P₂C₂tBu₂)₂}₂]₄ (1104, Scheme 155), an example of a metallosupramolecular compound which can be obtained from different gold precursors in the presence of 4 equiv of anion 189.⁶⁵⁴

The flexidentate coordination behavior of 189 or 191 was demonstrated with group 8–10 metals (Scheme 156).^474,655 Coordination studies showed that the diphosphacyclobutadiene moiety of 189 also acts as a σ-donor ligand toward the cobalt(I) or nickel(II) complexes (C₄Me₄)CoI(CO)₂ (1105) or CpNiBr(PPh₃) (1106) to afford [Co(η⁴-P₂C₂tBu₂ML_n)(η⁴-P₂C₂R₂)] [ML_n = (C₄Me₄)Co(CO)₂ (1107)] and [Co(η⁴-P₂C₂tBu₂ML_n)(η⁴-P₂C₂R₂)] [ML_n = CpNi(PPh₃) (1108) (Scheme 156, left).⁶⁵⁵ The latter two complexes are analogous to the coinage metal derivatives 1097–1099 mentioned above.^652,653 However, whereas simple σ-coordination of a P atom occurs in any of these cases, cobaltates 189 or 191 engage in rearrangement, insertion and fragmentation reactions toward other transition metal complexes, thereby creating new oligonuclear compounds with unprecedented structures.⁴⁷⁴

Scheme 156. Reactivity of 189 and 191 toward Metal Precursors from Groups 8–10.

In section 2.3.2, the formation of diphosphanickelacyclopentadiene ligands via insertion of Ni fragments into intact 1,3-diphosphacyclobutadiene moieties was discussed (vide supra). Complex 189 and the heteroleptic sandwich complex 688 behave similarly toward Ni fragments, yielding the aforementioned diphospha-nickelacyclopentadiene motifs (Scheme 81b, vide supra). Compounds 689 and 692 constitute, to the best of our knowledge, the sole examples of such insertion into diphosphacyclobutadiene ligands.^425,474

Likewise, upon reaction of 189 with the palladium(II) precursor cis-[PdCl₂(PPh₃)₂] (1109), the diphosphacyclobutadiene ligand of one molecule of cobaltate 189 is transformed to generate compound [Pd₃(PPh₃)₂{Co(η⁴-1,3-P₂C₂tBu₂)₂}₂] (1110, Scheme 156, bottom right). The resulting motif constitutes a rare 1,4-diphospha-2-butene ligand, proposed to have formed by isomerization of a 1,3-diphosphabutadiene unit to the 1,2-isomer, followed by activation of the resulting P–P bond at the central palladium center. This P–P bond is weakened, as observed from the crystallographic metric parameters of 1110 (P–P = 2.401(1) Å). Compound 1110 is a pentanuclear complex, with a chain of three palladium atoms. The central palladium atom completes its coordination environment with the phosphorus atoms of the isomerized diphosphacyclobutadiene unit, and with a chelating molecule of cobaltate 189 (Scheme 156).

Compound 189 also behaves as a chelate toward ruthenium, as identified by reaction with the halide cluster, [Cp*RuCl]₄ (230, Scheme 156, bottom). The product is the tetranuclear complex 1111, in which every phosphorus atom in the homoleptic cobaltate is coordinated to ruthenium centers through σ-bonds. This coordination mode necessitates adoption of an eclipsed conformation by the diphosphacyclobutadiene sandwich compound. Prolonged reaction of the cobaltate 189 with the ruthenium cluster 230 caused the P₂C₂tBu₂ ligand to fragment, in a process possibly involving the chelating tetranuclear complex 1111, to generate the dinuclear ruthenium-“P₂C₂” species 1112. Finally, reactions of the Rh(I) dimer [RhCl(cod)]₂ (401) with cobaltate 189 or 191, demonstrated π-coordination of the 1,3-P₂C₂tBu₂ unit. The products, complexes [Rh(cod){Co(P₂C₂R₂)₂}] [R = tBu (1113), Ad (1114); Scheme 156, upper right), present bridging μ,η⁴:η⁴ diphosphacyclobutadiene ligands. The reactivity of the π-bound complexes toward additional phosphaalkyne substrate suggested that the synthesis of a genuine triple-decker complex might be feasible in the case of the adamantyl derivative.⁴⁷⁴

3.5.3. Reactivity of Di-tert-butyldiphosphatetrahedrane toward Metalates

Diphosphacyclobutadiene complexes were also obtained from reactions of a phosphaalkyne dimer, di-tert-butyldiphosphatetrahedrane, (tBuCP)₂ (1115) with various metalates.⁶⁵⁶ Compound 1115, reported by Wolf and co-workers in 2019, was obtained by nickel catalyzed dimerization of the corresponding phosphaalkyne monomer.⁶⁵⁷ The reactivity of the diphosphatetrahedrane toward [Fe(η⁴-C₁₄H₁₀)₂]⁻ (52, the [K([18]crown-6)]⁺ salt), cobaltate [Co(η⁴-cod)₂]⁻ (239),⁷ or complexes [(^DippBIAN)M(η⁴-cod)]⁻ [M = Fe (537), Co (540)] was investigated (Scheme 157). The reaction between 52 and 1115 afforded the heteroleptic compound [K([18]crown-6)][Fe(1,2-tBu₂C₂P₂)(η⁴-C₁₄H₁₀)] (1116), featuring a rare 1,2-diphosphacyclobutadiene ligand (Scheme 157a). Similar treatment of 239 with (tBuCP)₂ furnished, instead, a mixture of compounds, of which the major product was characterized as the homoleptic 1,2-diphosphacyclobutadiene-containing complex [K([18]crown-6)(1,4-dioxane)₄][Co(1,2-tBu₂C₂P₂)₂] (1117) and the minor species corresponds to the known 1,3-isomer [Co(1,3-P₂C₂tBu₂)₂]⁻ (189, see Scheme 157b).²⁰⁸ The observed reactivity indicates that the diphosphatetrahedrane (tBuCP)₂ serves as a precursor to the rare 1,2-tBu₂C₂P₂ ligand, as opposed to the 1,3-isomer which generally results from head-to-tail cyclodimerization of the parent phosphaalkyne tBuC≡P.

Scheme 157. Reactivity of Di-tert-butyldiphosphatetrahedrane toward Different Metalates^656,657.

The 1,2-diphosphacyclobutadiene ligands are, therefore, the result of P–C bond cleavage in the C₂P₂ tetrahedron of 1115. To confirm this hypothesis, the diiminometalates [(^DippBIAN)M(η⁴-cod)]⁻ [M = Fe (537), Co (540)] were independently treated with tBuC≡P (2 equiv) or with 1115 (1 equiv) (see Scheme 157c). As in the case of the homoleptic alkene/arene metalates, the interaction between [(^DippBIAN)M(η⁴-cod)]⁻ and tBuC≡P yielded the terminal 1,3-diphosphacyclobutadiene complexes [(^DippBIAN)M(1,3-P₂C₂tBu₂)]⁻ [M = Fe (1118), Co (1119)], while reactions of the metalates with the diphosphatetrahedrane molecule generated the 1,2-isomers [(^DippBIAN)M(1,2-tBu₂C₂P₂)]⁻ [M = Fe (1120), Co (1121)]. After prolonged heating (60 °C, THF-d₈, 5 d), compound 1119 isomerizes to 1121. Comparative reactivity studies were performed on the BIAN cobaltate complexes, to gain insights into the reactivity of these isomeric ligands.⁶⁵⁶

Clear differences were observed after reaction with electrophiles: While treatment of the 1,3-isomer 1119 with HCl, Me₃SiCl or Cy₂PCl afforded mainly the dimeric compound [(^DippBIAN)Co(μ-Cl)]₂ (1122), compound 1121 reacted with these electrophiles to yield a bridged 1,2-diphosphacyclobutadiene dinuclear complex, [{(^DippBIAN)Co}₂(μ,η⁴:η⁴-1,2-P₂C₂tBu₂)] (1123, see Scheme 158, top). Furthermore, it was also possible to obtain a product of ring expansion from 1121 and Cy₂PCl (Scheme 158, bottom), which led to the isolation of the neutral [(^DippBIAN)Co(η⁴-P₃C₂tBu₂Cy₂)] (1124) and confirms the synthetic utility of the 1,2-diphosphacyclobutadiene moiety. In the case of the 1,3-isomer 1119, analogous treatment with Cy₂PCl resulted in an unselective reaction. For all the BIAN-containing metalates, the characterization data indicate the presence of dianionic BIAN^2– ligands, with bonding best described as between the two possible extreme electronic cases: neutral π-accepting P₂C₂tBu₂ ligands with low-spin M(I) centers, or dianionic π-donating P₂C₂tBu₂ ligands at low-spin M(III) centers.⁶⁵⁶

Scheme 158. Initial Reactivity Studies of a Coordinated 1,2-Diphosphacyclobutadiene Ligand⁶⁵⁶.

4. Summary and Outlook

The chemistry of low-valent transition metalates has developed tremendously over the last one and a half decades. Previously, metalates stabilized by carbonyl, alkene and arene ligands were principally known, while research efforts focused on the synthesis, characterization, and only basic reactivity of new types of metalates. Early work in the area revealed the high synthetic utility of carbonylate anions as nucleophiles in organic chemistry and as synthons for the preparation of oligonuclear transition metal carbonyl compounds. Furthermore, it was shown that alkene and arene metalates containing labile alkene or arene ligands may serve as powerful “sources of transition metal anions” (J. E. Ellis), which enable the synthesis of a large variety of low oxidation-state transition metal complexes. This pioneering work laid the foundation for recent advances in metalate chemistry, by intimating that the high reactivity of these complexes could be applied in a broader spectrum of synthesis, and even catalysis.

To date, the chemistry of the transition metal carbonylate anions continues to fascinate as the field is revisited with more sophisticated spectroscopic and structural techniques, enabling the remaining “gaps” in the field to be closed–as exemplified by Korber’s characterization of the tricarbonyl nickelate [Ni(CO)₃]^2– (26). Likewise, the development of bulky isocyanide ligands by Figueroa has enabled the isolation of complexes with unique coordination environments and unusual reactivities (as demonstrated, for example, by the recent characterization of the fluoroborylene-iron complex 67 and the P₂-iron complex 73).

In comparison with the well-established carbonyl and isocyanide complexes, alkene and polyarene ligands in metalates are usually much more labile. As a result, alkene and arene metalates are potent reagents–by serving as precursors, a wide range of organometallic complexes have become accessible. In addition, metalates acting as sources of metal atoms to generate unique heterobimetallic clusters have opened another highly important avenue of investigation in the field (e.g., compounds 236, 237, and 242).

While the synthesis of alkene, arene and carbonyl metalates is relatively well-explored, metalates with sophisticated new architectures have been prepared by introducing carbon-, nitrogen- and phosphorus-based ligands. The variety of metalates stabilized by carbenes, phosphines, amides, and imines is simply astonishing. Indeed, much of the recent progress in the field is due to advances in ligand design, which offers precise control over the coordination and electronic properties of the metal atom. This has also provided ligand scaffolds which, through their flexibility, can adapt to support metal centers as they engage in further reactivity, illustrated by the varying hapticity of phosphinine ligands in complexes 308, 309, and 311–316. The development of sophisticated metalates stabilized by tailored ligands translates into ever more successful applications in small molecule activation and catalysis. For instance, the P₃^B ligand, featured in complexes 718–728, has been demonstrated to continue its stabilization of an iron center as it undergoes a series of oxidation state changes enforced by activation of small molecules, by altering its coordination mode. In addition, interesting molecular properties which offer specialized applications, such as magnetic anisotropy, can be incorporated into new complexes.

The significant influence of ion-pairing effects on the stability and reactivity of metalates was largely neglected until recently. However, there are a growing number of examples showing that the countercation (usually an alkali or alkaline earth metal cation) substantially influences their reactivity and catalytic properties. In particular, ion-pairing interactions are crucial for stabilizing highly charged metalate anions and enable unusual coordination geometries (e.g., in the hydride complexes 706–716).

Furthermore, the interaction of transition metal atoms with Lewis acidic p-block elements (e.g., Al and Ga) and the synthesis of bimetallic transition metalates are successful recent strategies for tuning the electronic properties and the reactivity of metalate anions. A range of binuclear metalate anions has been developed by the groups of Peters and Lu, which are very effective in N₂ functionalization and CO₂ reduction chemistry.

Indeed, the development of reactive low-oxidation state metalate anions that efficiently and selectively transform small inorganic molecules such as H₂, CO, CO₂, N₂, and P₄ is arguably the most striking advance in the field. Prior to the period covered in this review, the potential of metalates in these areas had largely been overlooked. Moreover, the importance of metalate anions as key catalytic intermediates has become increasingly clear, and the number of catalytic transformations in which they are implicated has grown rapidly. Catalytic applications include cross-couplings, isomerizations, hydrogenations and dehydrogenations as well as CO₂ and N₂ reductions. The electron-rich nature of transition metalates is a key asset for the success of these applications.

Considering the impressive depth and the large variety of the metalate chemistry, it seems a reasonable assessment that the field is approaching maturity. Nevertheless, further exciting developments can be expected building on the results described in this review.

What is next for the field?

Although “it is very difficult to predict—especially the future” (Niels Bohr), we wish to conclude this article with a short outlook highlighting some possible future developments in the area:

• 1)Synthesis of metalates. While their molecular design has become increasingly sophisticated, the synthetic routes employed have remained limited to classical approaches (see Figure 1 in section 1), most commonly the reduction of suitable precursor compounds with a strong reducing reagent such as KC₈. New innovative synthesis techniques (e.g., ball milling) might enable more selective reactions and give access to previously unknown metalate species. The continued improvement of experimental and theoretical characterization techniques will also provide greater understanding of the fundamental nature of new classes of metalates, as they emerge.
• 2)Small molecule activation. The ability of highly reduced metalate complexes to mediate challenging transformations of unreactive molecules will continue to be of prime interest. Increasingly sophisticated ligand design should enable more effective and selective transformations.
• 3)Homogeneous catalysis. The potential of low-valent metalates in homogeneous catalysis is increasingly being realized. In particular, metalates have much potential in reductive catalysis, e.g. in hydrogenations and (as the reverse reaction) dehydrogenations. Surprisingly, to the best of our knowledge there are no applications of metalate anions in asymmetric catalysis; this is likely an area of interest in the future. Reported insights into the mechanistic minutiae surrounding metalate-mediated transformations will also permit greater rational design in the synthesis of metalate complexes intended to act as effective precatalysts.
• 4)Ion-pairing effects and metal–metal bonding. Recent studies have revealed the crucial influence of ion-pairing on the reactivity of metalate anions. In the future, the interaction of metalates with Lewis acidic metal cations should be studied in more detail to delineate the characteristics of ion-pairing in such species. Using these interactions, the reactivity of metalates can be tuned, which will have an impact on the development of catalytic processes.
• 5)Synthesis of new metal clusters and inorganic materials. The use of metalates as metal atom sources for metal clusters and heterometallic materials is still in its infancy. However, this strategy will become increasingly important in the future and might eventually be used to introduce transition metal atoms into molecular materials. Furthermore, some low-coordinate metalates show very interesting magnetic properties, e.g., magnetic anisotropy and slow magnetic relaxation, which are naturally essential for the design of magnetic materials.

So far, the chemistry of low-valent transition metalates has mainly been of fundamental academic interest. However, considering the plethora of significant developments described in this review, it seems very likely that many new applications of metalate anions will appear in diverse contexts. It is the aim of this review to enlighten and inspire future researchers to develop new applications of these fascinating and often highly reactive species. Based on the impressive, rapid advancement of the field over the last 15 years, the future of transition metalate chemistry looks very bright!

Glossary

Abbreviations

AB

ammonia borane (NH₃BH₃)

acac

acetylacetonate

Ad

adamantyl

AE

alkaline earth

AltraPhos

Al[N{o-(NCH₂PiPr₂)C₆H₄}₃]

AM

alkali metal

anthr

anthracene

Ar

aryl

Ar^DArF2

2,6-bis[3,5-bis(trifluoromethyl)phenyl]-4-fluorophenyl

Ar^Dipp2

2,6-bis(2,6-diisopropylphenyl)phenyl

Ar^Mes2

2,6-bis(2,4,6-trimethylphenyl)phenyl

Ar^Tripp2

2,4,6-tris(2,6-diisopropylphenyl)phenyl

BAr^F₄

tetrakis[(3,5-bis(trifluoromethyl)phenyl]borate

Bcat

boronic acid catechol ester

bda

benzylideneacetone

BDFE

bond dissociation free energy

^DippBIAN

1,2-bis(2,6-diisopropylphenylimino)acenaphthene

^MesBIAN

1,2-bis(2,4,6-dimethylphenylimino)acenaphthene

BINC

(bis(2-isocyanophenyl)phenylphosphonate)

bp

benzophenone

bpa

bis(2-picolyl)amine

^DippBPDI

2,6-(2,6-iPr₂–C₆H₃N=CPh)₂C₅H₃N

bpy

2,2′-bipyridine

BS-DFT

broken symmetry DFT

BTMSA

bis(trimethylsilyl)acetylene

CAAC

cyclic(alkyl)(amino)carbene

CASSCF

complete active space self-consistent field

CHD

cyclohexadiene

CKphos

1-((3aR,8aR)-2,2-dimethyl-4,4,8,8-tetrakis(perfluorophenyl)tetrahydro-[1,3]dioxolo[4,5-e] [1,3,2]dioxaphosphepin-6-yl)pyrrolidine

CNAr₃NC

2,2′′-diisocyano-3,5,3′′,5′′-tetramethyl-1,1′:3′,1′′-terphenyl

cod

1,5-cyclooctadiene

CODH

carbon monoxide dehydrogenase

Cp

cyclopentadienyl

Cp*

pentamethylcyclopentadienyl

Cp‴

η⁵-C₅H₂-1,2,4-tBu₃

CP MAS NMR

cross-polarization magic angle spinning NMR

[2.2.2]cryptand

N(CH₂CH₂OCH₂CH₂OCH₂CH₂)₃N

CV

cyclic voltammetry

Cy

cyclohexyl

dach

1,2-diaminocyclohexane

dad

1,4-diazabutadiene

dae

1,2-diaminoethylene

dap

1,3-diaminopropane

DBU

1,8-diazabicyclo [5.4.0] undec-7-ene

dct

dibenzo[a,e]cyclooctatetraene

Dep

(2,6-diethyl)phenyl

DFT

density functional theory

DHC

dehydrocoupling

Dipp

(2,6-diisopropyl)phenyl

DMAB

dimethylaminoborane (Me₂NHBH₃)

DMAP

p-dimethylaminopyridine

^Dipp2dmdad

N,N′-bis(2,6-diisopropylphenyl)butane-2,3-diimine

dme

dimethoxyethane

dmf

dimethylformamide

dmpe

1,2-bis(dimethylphosphino)ethane

dppe

1,2-bis(diphenylphosphino)ethane

dvtms

divinyltetramethyldisiloxane

en

ethylenediamine

ENDOR

electron nuclear double resonance

EPR

electron paramagnetic resonance

ESI

electrospray ionization

Et

ethyl

EXAFS

extended X-ray absorption fine structure

EXSY

exchange spectroscopy

Fc

ferrocene

FT-IR

Fourier-transform infrared

HAT

hydrogen atom transfer

HBpin

pinacolborane

HERFD-XANES

high-energy-resolution fluorescence-detected XANES

HMPA

hexamethylphosphoramide

HOMO

highest-occupied molecular orbital

IBO

intrinsic bond orbital

ICy

1,3-dicyclohexylimidazol-2-ylidene

IMAr^Mes2

5-(Ar^Mes2-imino)furanone

IMes

1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene

ipa

iminopropenamide

IPr

1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene

iPr

isopropyl

IR-SEC

infrared spectroelectrochemistry

L^Dipp

1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene

LIFDI-MS

liquid injection field desorption mass spectrometry

Me

methyl

Me₂dad

1,4-dimethyl-diazabuta-1,3-diene, 2-MeTHF, 2-methyltetrahydrofuran

Mes

mesityl (1,3,5-trimethylphenyl)

mesbpy

6,6′-dimesityl-2,2′-bipyridine

MesIm

1-mesitylimidazol-2-ylidene

MLC

metal–ligand cooperation

MO

molecular orbital

MS

mass spectrometry

nacnac

CH[C(tBu)N(2,6-iPr₂C₆H₃)]₂

^Arnacnac

CH(ArNCMe)₂

^Dippnacnac

HC{C(Me)N(2,6-C₆H₃iPr₂)}₂

^Me3nacnac

MeC[C(tBu)N(2,6-iPr₂C₆H₃)]₂

nbd

norbornadiene

NBO

natural bond orbital

nBu

n-butyl

NHC

N-heterocylic carbene

NMP

N-methylpyrrolidone

NMR

nuclear magnetic resonance

NON

4,5-bis(2,6-diisopropylanilido)-2,7-di-tert-butyl-9,9-dimethylxanthene

OTf

triflate; trifluoromethanesulfonate

P₂^P

PhP(o-iPr₂PC₆H₄)₂

P₃^B

tris[2-(diisopropylphosphino)phenyl]borane

P₃^C

tris[2-(diisopropylphosphino)phenyl]methyl

P^iPr₃^Si

tris[2-(diisopropylphosphino)phenyl]silyl

P^Ph₃^Si

tris[2-(diphenylphosphino)phenyl]silyl

PCP

2,6-(CH₂PtBu₂)₂-1-yl-C₆H₃

PDI

pyridine-diimine

Ph

phenyl

PhP₂^B

PhB(o-iPr₂PC₆H₄)₂

PHDI

bis(2,6-diisopropylphenyl)phenanthrene-9,10-diimine)

PMDETA

N,N,N′,N″,N″-pentamethyldiethylenetriamine

PN^triazineP

2,6-bis(PiPr₂NH)-4-Ph-triazine

py

pyridyl

QTAIM

quantum theory of atoms in molecules

R*

2,4,6-tBu₃-C₆H₂

r.t.

room temperature

SEM

scanning electron microscopy

SOMO

singly occupied molecular orbital

tbpy

4,4′-di-tert-butyl-2,2′-bipyridine

tBu

tert-butyl

tBuIm

1-tert-butylimidazol-2-ylidene

TEM

transmission electron microscopy

TFA

trifluoroacetate

TFE

2,2,2-trifluoroethanol

thf

tetrahydrofuran

TM

transition metal

TMEDA

1,2-bis(dimethylamino)ethane

tol

tolyl

TolNCNTol

di(p-tolyl)carbodiimide

tPent

tert-pentyl

tpdp

1,4-bis(2-(diisopropylphosphino)phenyl)benzene

TPP

2,4,6-triphenylphosphinine

tpy

2,2’:6′,2’’-terpyridine

trop

5H-dibenzo[a,d]cyclohepten-5-yl

trop₂NH

bis(5Hdibenzo[a,d]cyclohepten-5-yl)amine

Tt

tetrel

VE

valence electron

VtCXES

valence-to-core X-ray emission spectroscopy

XANES

X-ray absorption near-edge structure spectroscopy

XPS

X-ray photoelectron spectroscopy

Xyl

xylyl (2,6-dimethylphenyl)

Biographies

Vanessa R. Landaeta graduated in chemistry from Universidad Simón Bolívar, Caracas (Venezuela) and received her doctoral degree from the Instituto Venezolano de Investigaciones Científicas (IVIC). After postdoctoral Marie Curie research training in the Consiglio Nazionale delle Ricerche (Italy), she started her independent research career at Universidad Simón Bolívar, where she subsequently became full professor. After a period as guest researcher with AntonioTogni (ETH Zurich, Switzerland), she joined the Wolf group in Regensburg from 2020 to 2023 as a postdoctoral researcher. Currently, she is a researcher at the Istituto di Chimica dei Composti Organometallici (ICCOM-CNR) in Florence, Italy. Her research interests include inorganic/organometallic chemistry, ligand design, and homogeneous catalysis.

Thomas M. Horsley Downie received his MChem (Hons) in Chemistry from Newcastle University (UK) in 2017. Undertaking his PhD at the University of Bath (UK) under the supervision of Dr. David Liptrot, he completed his thesis in 2021 on the chemistry of copper(I) complexes with main group functionalities. Since 2022, he has conducted postdoctoral research in the group of Prof. Robert Wolf at the University of Regensburg (Germany), developing methods for the functionalization of white phosphorus. His research interests include main group chemistry, small molecule activation, and organometallic synthesis and catalysis.

Robert Wolf started his research career at the University of Cambridge, UK, under the guidance of Dominic S. Wright. He was awarded a PhD from Leipzig University for work in phosphorus chemistry supervised by Evamarie Hey-Hawkins. After postdoctoral research with Philip P. Power (UC Davis, USA) and Koop Lammertsma (VU Amsterdam, The Netherlands), he started his independent career at University of Münster (mentor: Werner Uhl). He became Professor of Inorganic Chemistry at the University of Regensburg in 2011, where he was recently promoted to a chair. His research interests lie at the crossroads of low-oxidation state transition metal chemistry, homogeneous (photo)catalysis, and phosphorus chemistry.

Author Present Address

^▽ Istituto di Chimica dei Composti Organometallici, Consiglio Nazionale delle Ricerche (ICCOM-CNR). Via Madonna del Piano 10, 50019 Sesto Fiorentino (Italy)

Author Contributions

V.R.L. and R.W. developed the concept. V.R.L. gathered the data and wrote the original draft. R.W. conceived and supervised the project and acquired funding. T.M.H.D. edited and revised the draft. All authors have given approval to the final version of the manuscript. CRediT: Vanessa R. Landaeta conceptualization, data curation, writing-original draft, writing-review & editing; Thomas M. Horsley Downie writing-review & editing; Robert Wolf conceptualization, funding acquisition, supervision, writing-review & editing.

The authors declare no competing financial interest.

Dedication

This paper is dedicated to John E. Ellis (University of Minnesota, USA), a pioneer in transition metalate chemistry, on the occasion of his retirement.

The following relevant literature came to our attention after the manuscript was accepted. Korber and co-workers reported the crystallographic characterization of [A([18]crown-6]₂[Pt(CO)₃] · 10 NH₃ (1125), representing the first mononuclear platinum carbonylate, and the Chini type cluster [K([2.2.2]-cryptand)]K[Pt₃(μ²-CO)₃(PPh₃)₃] · 3NH₃ (1126, A = K, Rb, [18]crown-6).^658,659 The compounds were serendipitously obtained as single crystals from reaction mixtures of the Zintl phase K₆Rb₆Ge₁₇ with Pt(CO)₂(PPh₃)₂, [18]crown-6 and [2.2.2]-cryptand in liquid ammonia. The trigonal planar carbonyl platinate dianion observed in 1125 is analogous to the [Ni(CO)₃]^2– anion in compound 26 (section 2.2.1.). The structure of 1126 displays a monomeric Chini-type dianion [Pt₃(μ²-CO)₃(PPh₃)₃]^2– containing terminally coordinated triphenylphosphine ligands on each Pt atom. It was speculated that the stabilizing effect of liquid ammonia towards negatively charged species might be a crucial factor for the formation of 1126. In work related to the chemistry described in section 3.1.5, Lu and co-workers reported the utilization of a bimetallic RhGa photoredox catalyst [RhGa{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (1127) in the catalytic hydrogenolysis of C–F bonds driven by violet (395 nm) LED light.⁶⁶⁰ An analogous RhAl complex proved to be less active. However, it is important to note that RhIn complex [RhIn{N[o-(NCH₂PiPr₂)C₆H₄]₃}]⁻ (1128), the heavier congener of 1127, catalyzes similar hydrodefluorination reactions under thermal conditions (1 atm H₂ pressure and 70–90 °C).⁶⁶¹ A purely inorganic sandwich complex [Fe(cyclo-P₄)₂]^2– (1129) was recently reported by Sun and co-workers. This complex is related to the decaphosphatitanocene dianion 1042 described in section 3.4.⁶⁶² Finally, a recent report by Ellis and co-workers on the preparation and crystallographic characterization of the remarkable complex [K(thf)₂]₂[{Cu(9,10-η²-C₁₄H₁₀)}₂] (1130, C₁₄H₁₀ = anthracene) is noteworthy.⁶⁶³ Compound 1130 is the first anionic arene complex of copper. In the solid state, the complex forms a dinuclear ion-contact structure with the copper atoms bound in a nearly linear fashion to the 9,10-carbons of the central ring of anthracene. The [K(thf)₂]⁺ cations form a sandwich type arrangement with the exo-arene rings.