Download PDF
Synthesis 2022; 54(13): 2976-2990
DOI: 10.1055/a-1806-4513
short review

Mechanisms of the Nickel-Catalysed Hydrogenolysis and Cross-Coupling of Aryl Ethers

Andryj M. Borys∗
,
Eva Hevia  ∗

We thank the Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (SNSF, 188753) and the University of Bern for their generous sponsorship of this research.
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Abstract

The Ni-catalysed hydrogenolysis and cross-coupling of aryl ethers has emerged as a powerful synthetic tool to transform inert phenol-derived electrophiles into functionalised aromatic molecules. This has attracted significant interest due to its potential to convert the lignin fraction of biomass into chemical feedstocks, or to enable orthogonal reactivity and late-stage synthetic modification. Although the scope of nucleophiles employed, and hence the C–C and C–heteroatom bonds that can be forged, has expanded significantly since Wenkert’s seminal work in 1979, mechanistic understanding on how these reactions operate is still uncertain since the comparatively inert Caryl–O bond of aryl ethers challenge the involvement of classical mechanisms involving direct oxidative addition to Ni(0). In this review, we document the different mechanisms that have been proposed in the Ni-catalysed hydrogenolysis and cross-coupling of aryl ethers. These include: (i) direct oxidative addition; (ii) Lewis acid assisted C–O bond cleavage; (iii) anionic nickelates, and; (iv) Ni(I) intermediates. Experimental and theoretical investigations by numerous research groups have generated a pool of knowledge that will undoubtedly facilitate future discoveries in the development of novel Ni-catalysed transformations of aryl ethers.

1 Introduction

2 Direct Oxidative Addition

3 Hydrogenolysis of Aryl Ethers

4 Lewis Acid Assisted C–O Bond Cleavage

5 Anionic Nickelates

6 Ni(I) Intermediates

7 The ‘Naphthalene Problem’

8 Conclusions and Outlook


#

Key words

nickel - catalysis - cross-coupling - mechanism - phenol electrophiles
1

Introduction

Zoom Image
Andryj M. Borysreceived his B.Sc. (2015) and Ph.D. (2018) from the University of Kent (UK), the latter under the supervision of Dr Ewan Clark. He was a postdoctoral researcher with Dr Michael Cowley at the University of Edinburgh (UK), then with Prof. Thomas Baumgartner and Prof. Chris Caputo at York University (Canada) before joining the group of Prof. Eva Hevia at Universität Bern (Switzerland) in 2020. His current research explores the synthesis and catalytic applications of s- and d-block heterobimetallic complexes. Andryj is also the creator of the Schlenk Line Survival Guide.
Eva Hevia received her received both her M.Sci. degree in Chemistry and her Ph.D. degree from the Universidad de Oviedo (Spain) in 1998 and 2002, respectively. After a three-year position at the University of Strathclyde (Glasgow, UK) working as a Marie Curie Fellow with Professor Robert Mulvey, in 2006 she took up a Royal Society University Research Fellowship and Lectureship there. Subsequently she was promoted to Full Professor in 2013. In February 2019, Eva moved to the University of Bern to take up a full professorship in Inorganic Chemistry. Research in her group focuses on polar organometallic chemistry at the crossroads of inorganic, organic, and green chemistry. Eva is an elected fellow of the European Academy of Sciences and her research has been recognized with several awards including the 2017 RSC Corday-Morgan Prize and the 2021 Arfvedson-Schlenk Prize by the GDCh.

Nickel catalysis has emerged as versatile tool to construct a diverse range of carbon–carbon and carbon–heteroatom bonds.[1] [2] Beyond simply being a cheaper, earth-abundant alternative to palladium, the metal of choice to catalyse many cross-coupling reactions, nickel presents several other key advantages that have made it not only a unique catalyst but also a metal of continued fundamental and theoretical interest. These include the propensity of nickel to engage in single electron reactivity and manoeuvre across many accessible oxidation states [from Ni(0) to Ni(IV)], which in turn enables unique bond-forming strategies.[3] Mechanisms involving a Ni(I)/Ni(III) catalytic cycle have been identified in several coupling reactions,[4,5] and significant effort has been made to develop efficient routes to Ni(I) and Ni(III) species.[6] [7] Furthermore, nickel catalysts can activate more challenging and inert substrates outside the scope of Pd allowing unconventional, yet readily available, electrophiles to serve as viable cross-coupling reaction partners.[8] [9] As early as 1979, Wenkert discovered that aryl ethers can undergo Caryl–O bond cleavage and Kumada–Corriu type cross-coupling with Grignard reagents to furnish biaryls (Scheme [1]).[10] However, only since 2000 have the benefits of phenol-derived electrophiles been recognised and utilised in a range of cross-coupling reactions.[11] Phenol-derived electrophiles are particularly attractive due to their ready accessibility and natural abundance, as well as the minimal (high atom-economy) and benign (non-halogenated) byproducts, particularly for phenols, aryl methyl ethers, and simple aryl esters. Since Pd catalysts are generally unable to react with all but the more activated phenol-derived electrophiles, such as triflates and tosylates,[9] nickel catalysis, therefore, provides a unique opportunity for orthogonal reactivity or the late-stage functionalisation of aromatic molecules containing OR (R = alkyl, aryl) substituents.[12]

Zoom Image
Scheme 1 Wenkert’s seminal reaction on the Ni-catalysed cross-coupling of aryl ethers

Whilst Wenkert’s seminal work has now evolved to convert aryl ethers into many other C–C and C–heteroatom coupled products using a range of nucleophiles beyond Grignard reagents,[12] detailed mechanistic understanding on how Ni enable these transformations have lagged behind. This is complicated by the awareness that different mechanisms are likely to be at play depending on the choice of nucleophile (and hence reaction conditions), with milder nucleophilic coupling partners, such as boronic esters, for example, typically requiring high catalyst loadings, elevated reaction temperatures and strongly donating ligands.[13] [14] Firstly, unlike activated phenol-derived electrophiles in which the Caryl–O bond can be readily cleaved and undergo direct oxidative addition to Ni(0),[8,9,11] the Caryl–O bond in aryl ethers is strong (~100 kcal mol–1 for anisole)[15] leading to very high and often inaccessible calculated barriers for reaction pathways following a classical Ni(0)/(II) cross-coupling mechanism (Scheme [2a]). Furthermore, stoichiometric studies have revealed that the proposed Ni(II)-OMe intermediates are unstable species and undergo β-hydride elimination to give Ni(0)-CO complexes.[16] [17] These findings suggest that non-classical mechanisms may be in operation, particularly for cross-coupling reactions that occur under mild conditions. This includes the proposed alternative anionic pathway which instead relies on the formation of an electron-rich Ni(0)-ate complex derived from the Ni(0) catalyst and organometallic nucleophile (Scheme [2b]). This pathway has been investigated by theoretical methods[18] [19] and proposed by several research groups, but only recently has direct experimental evidence for this alternative mechanism been elucidated.[20] Other mechanisms and variations of the two extremes shown in Scheme [2] have also been proposed, and this review aims to document the collective experimental and theoretical insights that have been reported to date.

Zoom Image
Scheme 2 Key examples of proposed catalytic cycles in the Ni-catalysed cross-coupling of aryl ethers; (a) neutral Ni(0)/Ni(II) pathway; (b) anionic Ni(0)/Ni(II) pathway

# 2

Direct Oxidative Addition

Many developments in the nickel-catalysed cross-coupling of aryl ethers are based on the working principle that a classical mechanism involving concerted oxidative addition, transmetalation, and reductive elimination are operative (see Scheme [2a]). However, this is challenged by the inert nature of the Caryl–OMe bond (~100 kcal mol–1 for anisole)[15] which ultimately means that direct oxidative addition to Ni(0) is thermodynamically and kinetically unfavourable. Combined experimental and theoretical studies by Gómez-Bengoa and Martin,[17] focusing on the reductive cleavage of aryl ethers with silanes,[21] have demonstrated that direct oxidative addition of 2-methoxynaphthalene to Ni(COD)2/PCy3 is not experimentally observed (Scheme [3a]) and the proposed Ni(II)-OMe intermediate cannot be accessed rationally via anion metathesis from (Cy3P)2Ni(Ar)Cl (Scheme [3b]). In the latter case, naphthalene was formed directly and quantitatively, and the nickel-containing species was identified as (Cy3P)2Ni-CO 1. Naphthalene was also formed when repeating the anion metathesis with NaOEt, but this time the Ni species was identified as (Cy3P)2Ni{HC(=O)Me} 2, in which an ethanal molecule is coordinated to Ni(0) in an η2-fashion. These studies reveal that even if direct oxidative addition was possible at elevated temperatures, that the Ni(II)-OR (R = Me, Et) intermediates themselves are highly unstable, even at room temperature, and undergo potentially unproductive β-hydride elimination. Since no naphthalene is observed in the absence of silane (the reducing agent, vide infra) under catalytic conditions, this further suggests that a catalytic manifold involving Ni(II)-OMe intermediates is unlikely under these circumstances.

Zoom Image
Scheme 3 Mechanistic insights into the oxidative addition of 2-methoxynaphthalene to Ni(COD)2/PCy3

Complementary theoretical studies reveal that the total activation barrier for the oxidative addition of 2-methoxynaphthalene to Ni(COD)2/PCy3 is +40.4 kcal mol–1, suggesting that concerted oxidative addition is very unlikely, even at elevated temperatures.[17] A slightly lower barrier of +35.1 kcal mol–1 was calculated when starting directly with Ni(PCy3)2, however, under experimental conditions this catalyst system performs worse than Ni(COD)2/PCy3 indicating that COD may play an important role as an auxiliary ligand during catalysis. This can add an extra layer of complexity to the mechanistic elucidation since the ligated Ni catalyst is often generated in situ, and the exact composition of the on-cycle intermediates is rarely validated by experimental means. In addition to experimental challenges associated with mechanistic elucidation, there has been shown to be a significant functional dependence with theoretical studies, with some functionals and basis-sets providing considerably lower activation barriers that often contradict experimental findings.[14] [17]

Further mechanistic insights into the oxidative addition of aryl ethers to Ni(0) and β-hydride elimination from Ni(II)-OMe species has been gathered by Agapie and co-workers.[16] Using the diphosphine aryl methyl ether pincer ligated Ni(0) complex 3, oxidative addition to first give 4 is observed after 12 hours at 45 °C, which readily converts into the Ni(II)-H 5 on further heating (Scheme [4]). Attempts to access the Ni(II)-OMe 4 by treatment of a Ni(II)-halide with NaOMe also gave a mixture of 4 and 5, amongst other unidentified species, highlighting that whilst direct oxidative addition to Ni(0) can occur, and is evidently more facile for intramolecular systems, that the Ni(II)-OMe species are highly unstable. Heating 3 to 100 °C liberates ca. 1 equivalent of combustible gas to give compound 6, which was identified as a Ni(0)-CO complex. This can be viewed as the product of reductive elimination from 5 to form a new Caryl–H bond, along with decarbonylation of formaldehyde to generate H2. These findings have important implications for nickel-catalysed hydrogenolysis of aryl ethers with H2 (vide infra)[22] since it illustrates that the new Caryl–H bond may originate from the aryl (alkyl) ether substrate itself and not H2. Indeed, when performing the hydrogenolysis of deuterium-labelled ArOCD3 under H2 catalysed by an NHC-Ni complex, Caryl–O bond cleavage occurs with 90% deuterium incorporation in the newly formed arene. Tobisu and Chatani have further exploited this realisation to develop catalytic protocols for the reductive cleavage of a range of aryl alkyl ethers in the absence of external reductants.[23]

Zoom Image
Scheme 4 Stoichiometric insights into the oxidative addition of ArOMe to Ni(0) and subsequent reductive elimination and decarbonylation reactions

An additional structural feature that has been proposed to precede (and ultimately facilitate) oxidative addition of the aryl ether to Ni(0) is coordination of the arene π-system to Ni(0),[14] [24] typically via η2-interaction from the double bond adjacent to the oxygen. This has been identified computationally for other, more activated phenol-derived electrophiles[25,26] and can be observed experimentally in compound 3 (Scheme [4]).[16]

Zoom Image
Scheme 5 Comparison of calculated activation barriers for the oxidative addition of 2-methoxynaphthalene to Ni(0) using PCy3 (in the absence or presence of CsF) or ICy ligands

Detailed theoretical studies have also been reported for the nickel-catalysed cross-coupling of aryl ethers using arylboronic esters as nucleophilic coupling partners.[14] These reactions are proposed to occur via oxidative addition of the aryl ether to Ni(0), and whilst the calculated barriers are high (and therefore account for the elevated reaction temperatures), the nature of the ligand and presence of an external base can significantly alter the barriers and make them more energetically favourable. Using PCy3 as the ligand, direct oxidative addition of 2-methoxynaphthalene to Ni(PCy3)2 has a calculated barrier of +32.9 kcal mol–1 (Scheme [5a]), which is considerably higher than the barriers previously determined for other phenol-derived electrophiles including aryl esters (+22.9 kcal mol–1)[25] and aryl carbamates (+13.5 kcal mol–1).[26] Notably, pathways involving a mono-PCy3 ligated Ni complex or using P t Bu3 were both found to be less energetically favourable. The addition of CsF, which was found to be essential for the reaction to proceed,[13] lowers the calculated activation barrier for oxidative addition to +28.8 kcal mol–1 (Scheme [5b]) and delivers an intermediate that is now energetically favoured with respect to the starting materials (–0.4 kcal mol–1 relative to Ni(COD)2/PCy3). The subsequent transmetalation step occurs with a comparably high barrier since dissociation of one of the PCy3 ligands is necessary to access a four-membered transition state. This barrier is again significantly reduced in the presence of CsF, however, unlike traditionally accepted mechanisms, in which the external base accelerates transmetalation by generating a four-coordinate borate species,[27] [28] the transmetalation pathway is more energetically favoured when CsF coordinates directly to the Ni(II) centre. Despite the high barriers calculated for transmetalation, these were nevertheless found to be lower and thus more favourable than competing β-hydride elimination reactions from the Ni(II)-OMe intermediates, a process that has been established by earlier stoichiometric studies (vide supra).[16,17] QTAIM and energy decomposition analyses reveal that it is weak interactions between CsF (as part of an adduct with the boronic ester) and the aryl ether, via both Cs···CAr and Cs···OMe contacts, that play a pivotal role in facilitating C–O bond cleavage (Scheme [5b]). Using the N-heterocyclic carbene (NHC) ligand ICy (ICy = 1,3-dicyclohexylimidazol-2-ylidene) in the absence of base (CsF or NaO t Bu), the calculated activation energy from the bis-ligated Ni(0) complex is +25.5 kcal mol–1 and, therefore, marginally lower than using PCy3 with or without CsF (Scheme [5c]).


# 3

Hydrogenolysis of Aryl Ethers

In 2011, Hartwig reported the use of a NHC-Ni catalyst for the selective hydrogenolysis of aryl ethers with H2.[22] This is a particularly important process since it can convert oxygen-rich lignocellulosic biomass into commercial arene feedstocks. Unlike previous heterogeneous catalysts that require high temperatures and pressures and afford a mixture of products (including cyclic hydrocarbons and cycloalkanols),[29] the homogenous NHC-Ni(0) catalyst operates selectively under relatively mild conditions (120 °C, 1 bar H2) without concurrent reduction of the arene ring or aliphatic C–O bonds. Unsurprisingly, this remarkable finding sparked significant efforts to elucidate the mechanism of this transformation,[30] [31] [32] and whilst several subsequent studies have demonstrated that the reductive cleavage of aryl alkyl ethers can occur in the absence of an external reductant (vide supra),[16,17,23] these previously discussed studies fail to justify the selective hydrogenolysis of diaryl ethers or benzyl ethers, the latter of which requires AlMe3 as an additive.

DFT studies by Surawatanawong and co-workers have provided mechanistic insights into the hydrogenolysis of diaryl ethers catalysed by Hartwig’s NHC-Ni system.[30] Starting with a mixture of Ni(COD)2 (COD = cycloocta-1,5-diene), SIPr (SIPr = 1,3-bis(2,6-diisopropylphenyl)imidazolidin-2-ylidene; generated in situ from SIPr·HCl and excess NaO t Bu) and diphenyl ether, the precursor (SIPr)Ni(η6-PhOPh) initially forms and then readily rearranges to (SIPr)Ni(η2-PhOPh) 7, which was identified as the active on-cycle catalyst (Scheme [6]). The formation of this species is justified by literature precedence which reports that a mixture of Ni(COD)2 and NHC in arene solution under a H2 atmosphere readily gives (NHC)Ni(η6-arene) complexes.[33] Notably however, whilst Ni(SIPr)2 is considerably lower in energy than 7 by 14.2 kcal mol–1, attempts to locate the corresponding diphenyl ether oxidative addition product were unsuccessful. Compound 7 undergoes oxidative addition to give the Ni(II) complex 8 via a three-centred transition state with an energy barrier of +24.0 kcal mol–1, making it the rate-determining step of the catalytic cycle. From 8, two possible catalytic cycles were identified, both of which have comparable energy barriers suggesting that both pathways may be possible. In cycle A, intermediate 8 first rearranges to 9 in which the phenolate is now trans to the vacant site on Ni, which subsequently binds H2 to give the σ-complex 10. This binding interaction involves both a charge transfer from the HOMO (the dxy orbital) of 9 to the σ*-orbital of H2, and charge transfer from the σ-orbital of H2 to the LUMO (the dx 2 -y 2 orbital) of 9. Whilst intermediate 9 is considerably less stable than 8 (+13.5 kcal mol–1), the subsequent steps in this pathway are energetically favoured. Intermediate 10 proceeds via facile σ-complex assisted metathesis (σ-CAM), depicted in transition state 11, to eliminate benzene in an exergonic process (–19.3 kcal mol–1 from 10) and give the T-shaped nickel complex 12. Reductive elimination of phenol from 12 occurs with a moderate energy barrier of +19.4 kcal mol–1 and regenerates the active Ni(0) catalyst 7.

Zoom Image
Scheme 6 Surawantanawong’s proposed mechanism for the hydrogenolysis of diphenyl ether with NHC-Ni catalysts

In the second pathway (cycle B, Scheme [6]), the initial diphenyl ether oxidative addition product 8, in which the phenyl is trans to the vacant site on Ni, can similarly bind H2 to give the corresponding σ-complex 13. This then undergoes σ-complex assisted metathesis, albeit with a moderate energy barrier (+19.3 kcal mol–1 from 8), with release of phenol to give intermediate 15. In contrast to cycle A, reductive elimination of benzene from 15 is barrierless, and the Ni(0) catalyst 7 is again regenerated. Alternatively, intermediate 8 can rearrange to a marginally more stable intermediate in which the SIPr ligand is now trans to the vacant site causing the phenolate to coordinate to Ni in an η3-OC ipso C ortho fashion. This nevertheless delivers intermediate 15 via similar H2 binding and σ-CAM, but now with an overall slightly higher energy barrier of +22.0 kcal mol–1.

Studies were also conducted to rationalise the selectivity observed during the hydrogenolysis of substituted diaryl ethers.[22] Using p-CF3-C6H4OPh as the substrate, η2-coordination at Ni(0) to form a complex akin to 7 is slightly favoured (1.0 kcal mol–1) at the electron-withdrawing arene ring containing the CF3 group compared to the phenyl ring. C–O bond activation occurs preferably at the carbon on the electron-withdrawing ring (+21.6 kcal mol–1 vs. +25.3 kcal mol–1 for the unsubstituted phenyl ring) to give the corresponding oxidative addition product, (SIPr)Ni(OPh)(p-CF3-C6H4), and ultimately deliver phenol and CF3-C6H5, consistent with experimental findings.[22] Furthermore, since the energy barrier for oxidative addition when using electron-withdrawing diaryl ethers is lower compared to diphenyl ether (cf. +24.0 kcal mol–1; see Scheme [6]), this justifies the lower reaction temperatures (100 °C vs. 120 °C) that can be employed for these substrates. Additional studies justifying the overall selectivity for the hydrogenolysis of aryl ethers found that the hydrogenation of benzene using the NHC-Ni catalyst has an inaccessible barrier of +39.0 kcal mol–1 and is overall an endergonic process.[30]

Subsequent DFT studies by Chung and co-workers instead focused on the overlooked role of additives in Hartwig’s NHC-Ni catalysed hydrogenolysis of aryl ethers.[31] Contrastingly to Surawatanawong’s studies,[30] Ni(SIPr)2 was determined to be the key starting point for catalysis given that ligand exchange from Ni(COD)2 by two NHC ligands is exergonic by –11.0 kcal mol–1. Oxidative addition of diphenyl ether, via (SIPr)2Ni(η2-PhOPh), however has an inaccessible barrier of +47.6 kcal mol–1 due to unfavourable steric repulsion of two cis NHC ligands. Alternatively, oxidative addition can proceed through a mono-NHC-ligated pathway (i.e., intermediates 7 and 8 in Scheme [6]), as suggested by Surawatanawong,[30] albeit with a high reaction barrier of +36.6 kcal mol–1 from Ni(SIPr)2. Given these high barriers, an alternative pathway which considers the role of excess NaO t Bu base was therefore explored by Chung and co-workers.[31] Starting with Ni(SIPr)2 16, ligand exchange of one NHC ligand by t BuO– is favourable (–1.2 kcal mol–1) and generates the anionic nickel species 17 (Scheme [7]). Coordination of diphenyl ether affords the η2-(PhOPh) intermediate 18 which undergoes oxidative addition to give 19 with a total barrier of +33.4 kcal mol–1 (relative to 17). Whilst this energy barrier is still high, it is lower than the unassisted base-free mechanism by 3.2 kcal mol–1, and furthermore the presence of the electron-rich anionic base strongly stabilises the Ni(II) oxidative addition product 19.

Zoom Image
Scheme 7 Anionic-base pathway for the oxidative addition to NHC-Ni(0)

Following C–O bond cleavage and oxidative addition, dissociation of PhO– from 19 is exergonic by 8.6 kcal mol–1 and generates a stable three-coordinate neutral intermediate. Subsequent H2 binding and σ-complex assisted metathesis proceeds with a barrier of +16.1 kcal mol–1, and release of t BuOH followed by reductive elimination of benzene occurs readily to give a stable (SIPr)Ni(η6-C6H6) intermediate. Finally, coordination of t BuO– and displacement of benzene regenerates 17, the active on-cycle species. The excess base employed in the reaction therefore assists in several different ways: (i) it reduces the barrier for oxidative addition and stabilises the corresponding Ni(II) intermediate through the formation of anionic nickel species; (ii) it facilitates the dissociation of PhO– which generates a more stable neutral species that promotes H2 activation with a lower barrier; (iii) it coordinates to Ni(0) to assist in the release of benzene and regenerate the active species; and (iv) it greatly increases the exergonicity of the reaction due to the formation of a stronger O–H bond in t BuOH vs. PhOH.

Studies assessing the selectivity observed for substituted diaryl ethers are in agreement with Hartwigs’ experimental[22] and Surawatanawong’s theoretical findings,[30] and cleavage of the electron-deficient Caryl–O bond occurs favourably due to lower activation barriers for oxidative addition. Chung and co-workers have also reassessed the competing β-hydride elimination that has been experimentally observed for aryl alkyl ethers.[16] [17] Attempts to locate classical β-hydride elimination transition states from Ni(II)-OMe species, however, were unsuccessful and instead a low-barrier pathway involving direct hydrogen transfer was identified. This is proposed to occur via a concerted pathway through a more stable five-centred transition-state which is lower in energy than the four-centred transition-state in the corresponding σ-CAM pathway.

In 2017, Hartwig reinvestigated the NHC-Ni catalysed hydrogenolysis of diaryl ethers to provide experimental clarification over previous, somewhat conflicting, mechanistic studies.[32] Whilst these demonstrate the crucial role of NaO t Bu, as proposed by Chung and co-workers,[31] Hartwig’s experimental studies instead suggest that it facilitates catalyst turnover. Firstly, the role of NaO t Bu was found to have a more extensive role than simply generating the free SIPr carbene in situ from the hydrochloride salt. When the hydrogenolysis of diaryl ethers is performed using the free carbene ligand in the absence of NaO t Bu, only 19% conversion is observed whilst quantitative conversion occurs in the presence of excess (2.5 equiv) base. Similar effects are observed when performing the reaction with both isolated Ni(SIPr)2 16 (Scheme [7]) and (SIPr)Ni(η6-arene)[33] as the catalyst, with the latter complex being identified as the catalyst resting state by NMR and UV-Vis spectroscopy. Despite experimental validation that the presence of excess base is necessary for improved yields, kinetic studies reveal a 0th order dependence in NaO t Bu indicating it is not involved in the rate-limiting step of the catalytic cycle. Instead, a first order dependence in the concentration of both (SIPr)Ni(η6-arene) and diaryl ether was determined and this is consistent with C–O bond cleavage and oxidative addition to Ni(0) being rate determining.

Zoom Image
Scheme 8 Stoichiometric reactivity of (SIPr)Ni(η6-C6H6) 20

Subsequent steps in the catalytic cycle were probed by using (SIPr)Ni(η6-C6H6) 20 (Scheme [8]) as the starting point for stoichiometric studies. Treatment of 20 with excess PhOPh affords (SIPr)Ni(η6-PhOPh) 21 in which the substrate is bound η6 to the Ni(0) centre. 2D EXSY NMR studies show that the bound and unbound aryl rings in 21 are in rapid exchange at room temperature. Furthermore, a minor species that also exists in equilibrium with 21 was identified, however neither of these complexes were believed to exhibit η2-coordination of the diaryl ether substrate as previously proposed by Surawantanawong.[30] Since attempts to access the proposed three-coordinate (SIPr)Ni(OAr)(Ar) intermediates by treatment of (TMEDA)Ni(OAr)(Ar) with SIPr were unsuccessful, alternative routes to related complexes were envisioned. The three-coordinate Ni(II) metallacycle 23 could be synthesised independently through the reaction of 20 with substituted phenoxide 22. This can be viewed as the formal oxidative addition product of dibenzofuran to 20, although efforts to access 23 via this route led to no reaction below 80 °C and decomposition above 120 °C. Metallacycle 23 is catalytically competent for the hydrogenolysis of diaryl ethers and the direct reaction of 23 with H2 affords the corresponding η6-κ1-hydroxybiphenyl Ni(0) complex 24, along with small amounts of 20. Complex 24 can also be independently prepared by treating 20 with 2-hydroxybiphenyl (25). The formation of 24 from 23 demonstrates that NaO t Bu is not directly needed for the reaction of the aryl nickel phenoxide with H2, and it was instead found that NaO t Bu deprotonates the hydroxybiphenyl, thereby facilitating the displacement of the corresponding phenoxide from Ni(0) to regenerate 20, the catalyst resting state.

Zoom Image
Scheme 9 Hartwig’s proposed mechanism for the NHC-Ni catalysed hydrogenolysis of diaryl ethers

Further experimental studies were conducted to probe how the phenol or phenoxide products from the hydrogenolysis of diaryl ethers interact with the on-cycle Ni(0) intermediates. No reaction is observed between 20 and PhONa, consistent with the preferential binding of neutral arenes to the NHC-Ni(0) species. Contrastingly, when 20 is treated with PhOH, a Ni(I) phenoxide-bridged dimer [(SIPr)Ni(OPh)]2 is obtained. This is proposed to form via initial oxidative addition of PhOH to Ni(0) to afford (SIPr)Ni(OPh)(H) that undergoes subsequent dissociation to (SIPr)Ni(OPh)2 and (SIPr)NiH2 followed by reductive elimination of H2 from the Ni(II) hydride to give a Ni(0) complex which comproportionates with (SIPr)Ni(OPh)2 to give the Ni(I) dimer. This finding reveals that under base-free conditions, the formation of PhOH during the hydrogenolysis of diaryl ethers results in the undesirable formation of the catalytically inactive Ni(I) dimer. This species however can re-enter the catalytic cycle when both H2 and NaO t Bu are present. Based on these comprehensive experimental studies, a mechanism for the NHC-Ni catalysed hydrogenolysis of diaryl ethers was proposed (Scheme [9]).[32] The (SIPr)Ni(η6-arene) complex A was determined to be the catalyst resting state and forms readily from Ni(COD)2 and SIPr in arene solvents under H2.[33] This then reacts reversibly with the diaryl ether substrate B to give the corresponding (SIPr)Ni(η6-PhOPh) complex C which is primed for oxidative addition. C–O bond cleavage was established to be the rate-limiting step of the reaction by kinetic studies, and should afford the three-coordinate Ni(II) complex D akin to 23 (Scheme [8]). Complex D reacts with H2 to give phenol E and the arene-ligated Ni(0) complex F, presumably via Ni(II) hydride intermediates. In the absence of NaO t Bu, the phenol reacts with Ni(0) (A, C, or F) to generate Ni(I) dimers that are off-cycle species. When NaO t Bu is present, deprotonation of phenol E affords the corresponding phenoxide H, which does not react with Ni(0) species and, therefore, does not poison the catalytic activity. Finally, replacement of the arene product with solvent or substrate to give (SIPr)Ni(η6-arene) (A or C) closes the catalytic cycle. This proposed mechanism closely resembles the steps computed by Surawantanawong,[30] and the experimentally determined ΔG for the oxidative addition and C–O bond cleavage is consistent with theoretical studies. No evidence for the formation of anionic nickel species could be observed experimentally, as proposed by Chung and co-workers,[31] however the role of excess NaO t Bu was determined to be critical for catalyst turnover despite not being involved in the rate-determining step.


# 4

Lewis Acid Assisted C–O Bond Cleavage

Since the oxidative addition of aryl ethers has been calculated to proceed with high, often inaccessible, energy barriers, alternative mechanisms have been proposed to rationalise the selective C–O bond cleavage enabled by Ni catalysts. In Hartwig’s NHC-Ni catalysed hydrogenolysis of aryl ethers with H2, the more challenging benzyl alkyl ether substrates require AlMe3 as an additive to facilitate the reaction.[22] Theoretical studies by Chung and co-workers reveal that the barrier for oxidative addition of benzyl methyl ether to Ni(0) is significantly reduced from +35.0 kcal mol–1 to +24.7 kcal mol–1 in the presence of AlMe3.[31] The coordination of the ethereal oxygen to the Lewis acidic Al centre stabilises the negative charge on the oxygen atom, thus stabilising the corresponding oxidative addition product. Chung further proposed that this Lewis acid assisted mechanism may also facilitate the cross-coupling of aryl ethers when other organometallic nucleophiles (R–M; M = Li, MgX, ZnX, BX2) are employed, and, indeed, the coordination of a Lewis acidic metal to the ethereal oxygen is frequently identified in experimental and theoretical studies.[14] [18] [19]

Rueping has expanded on these findings to develop a Ni-catalysed alkylation protocol for aryl ethers using trialkylaluminium reagents as the nucleophilic coupling partner.[34] Surprisingly, no undesirable β-hydride elimination is observed under the optimised conditions, and whilst a diverse range of trialkylaluminium nucleophiles could be employed, this method is not compatible with cyano, ester, and amide functional groups. Furthermore, this reaction only works using the bidentate phosphine ligand, dcype [1,2-bis(cyclohexylphosphino)-ethane], and still requires harsh conditions (100–120 °C, 12–72 hours) in combination with electron-deficient or π-extended aryl ethers. 1H and 27Al NMR studies support that coordination occurs between the ethereal oxygen of the substrate and trialkylaluminium. Complementary computational studies reveal that this coordination dramatically reduces the barrier for oxidative addition of 2-methoxynaphthalene from +40.0 kcal mol–1 (in the absence of AlEt3) to +18.6 kcal mol–1 (Scheme [10]), consistent with Chung’s studies.[31] Contrastingly to previous computational studies however in which C–O bond cleavage is the rate-determining step, the subsequent transmetalation to give (dcype)Ni(Ar)(Et) followed by reductive elimination were both predicted to have higher energy barriers (ca. +19 kcal mol–1 and 27.4 kcal mol–1, respectively), although the latter was predicted to being marginally lower (3.6 kcal mol–1) than competing β-hydride elimination.[16] [17] Stoichiometric reactions between [dppf)Ni(Ar)(Cl) (dppf = 1,1′-bis(diphenylphosphino)ferrocene] and AlEt3 demonstrate that the trialkylaluminium can act as a suitable nucleophile for the transmetalation at Ni(II), although the ligands and substituents on Ni, as well as the reaction conditions, differ significantly to catalytic conditions.[34]

Zoom Image
Scheme 10 Oxidative addition of 2-methoxynaphthalene to (dcype)Ni(COD) in the absence or presence of AlEt3

Agapie and co-workers have utilised their diphosphine aryl ether pincer scaffold (see compound 3 in Scheme [4]) as a model system to study the effect of Lewis acid additives on Caryl–OMe bond cleavage.[35] An increased rate of oxidative addition to Ni(0), and subsequent in situ transmetalation with (TMEDA)MgMe2, was observed in the presence of MeMgBr, AlMe3, Al i Bu3, AlEt3, and AlPh3. The transmetalating reagent itself, (TMEDA)MgMe2, was found to not affect the rate of oxidative addition since the bidentate TMEDA ligand prevents coordination of the ethereal oxygen to magnesium. Furthermore, the use of a non-polar solvent such as toluene is key, since THF can coordinate to the Lewis acid itself more favourable than the aryl ether. Additional spectroscopic, kinetic, and theoretical studies were employed to investigate the role of AlMe3 in more detail. Low temperature NMR spectroscopy supports that the Ni(0) aryl ether complex coordinates to AlMe3; this intermediate is stable at low temperatures and only converts into the corresponding Ni(II)-Me species upon warming. Kinetic studies revealed that although the rate of Caryl–OMe bond activation rises with increasing concentration of AlMe3, this effect is relatively­ small in comparison to the significant rate enhancement observed upon initial addition of Lewis acid (ca. 105–fold increase from 0–10 equivalents). This suggests that two distinct mechanisms may be in operation; the major primary mechanism involves coordination of the ethereal oxygen to a single molecule of AlMe3, whilst the minor mechanism, which requires additional AlMe3, is proposed to further activate the C–OMe bond or generate a more reactive, lower-energy transition-state (cis vs. trans coordination). Complementary DFT studies on a partially simplified system shows that AlMe3 coordination (i.e., the major mechanism) lowers the barrier for oxidative addition by at least 5.4 kcal mol–1; both trans and cis isomers were found to have similar calculated energies, but the latter has a lower barrier for oxidative addition (+12.5 kcal mol–1 vs. +17.0 kcal mol–1). These theoretical findings are in good agreement with spectroscopic and kinetic studies, and provide convincing evidence for a mechanism occurring via Lewis acid assisted oxidative addition of aryl ethers to Ni(0).


# 5

Anionic Nickelates

Whilst there is a pool of experimental and theoretical evidence to support that the oxidative addition of aryl ethers to Ni(0) can occur, albeit generally under harsh conditions and with suitable ligands or additives, this still fails to account for the mild conditions employed when using organolithium,[36] [37] [38] Grignard,[39–41] or organozinc[42] reagents as the nucleophilic coupling partner. Non-classical mechanisms have been proposed that instead invoke the formation of electron-rich, anionic nickelate species which are anticipated to be more reactive towards the inert aryl ether substrate. Wang and Uchiyama have provided theoretical support for the involvement of anionic nickelates through DFT studies,[18] focusing specifically on the Kumada–Corriu-type reaction between aryl ethers and Grignard reagents (PhMgBr) under Ni catalysis, which closely resembles Wenkert’s seminal report.[10] Although it was proposed that the Grignard reagent itself could potentially act as Lewis acid to assist in C–O bond cleavage[31] (vide supra), the calculated barriers for oxidative addition, whilst significantly lower than direct oxidative addition in the absence of a Lewis acid, were still found to be prohibitively high (+31.4 kcal mol–1) to account for experimental reaction conditions. An alternative pathway involving Ni(0)-ate intermediates was identified and found to have a significantly lower activation barrier of +13.6 kcal mol–1 (Scheme [11]). The two substrates and Ni(0) catalyst first combine together to form association complex 26, which is endergonic by 9.9 kcal mol–1 with respect to the starting components. Migration of the phenyl group from Mg to Ni proceeds via transition state 27 with a reasonable energy barrier (+12.7 kcal mol–1) to smoothly afford anionic Ni(0)-ate complex 28, the key intermediate. Cleavage of the Caryl–OMe bond is facilitated by coordination of the ethereal oxygen to magnesium (i.e., Lewis acid activation) and takes place though transition state 29 with a barrier of +13.6 kcal mol–1 to give the corresponding Ni(II) species 30. Release of MeOMgBr(solv) (the byproduct), and reductive elimination of biphenyl (the cross-coupled product) via 31 proceeds readily (+5.6 kcal mol–1), to regenerate the Ni(0) catalyst. Notably, whilst C–OMe bond cleavage is still the rate-determining step of the reaction, this pathway avoids the formation of unstable Ni(II)-OMe intermediates which are prone to undesirable β-hydride elimination.[16] [17]

Zoom Image
Scheme 11 Proposed reaction pathway for the Ni-catalysed cross-coupling of anisole and PhMgBr

Subsequent theoretical studies by Wang and Uchiyama identified comparable mechanisms involving Ni(0)-ate intermediates in both Murahashi- (organolithium) and Negishi-type (organozinc) cross-couplings of aryl ethers.[19] Notably, pathways operating via Lewis acid assisted oxidative addition were again found to be energetically inaccessible for both PhLi(solv) and PhZnCl(solv) (+42.7 and +34.5 kcal mol–1, respectively). The computed mechanism for PhLi(solv) closely resembles the pathway shown in Scheme [11] for PhMgBr(solv), and proceeds with a reasonable activation barrier of +24.2 kcal mol–1. Surprisingly, this is notably higher than for PhMgBr(solv) (cf. +13.6 kcal mol–1), despite the expected increased nucleophilicity of PhLi(solv). Using PhZnCl(solv), similar intermediates and transition states are again identified, but a high activation barrier energy of +33.2 kcal mol–1 is now calculated, indicating that this reaction is unlikely to occur at ambient temperature. Experimentally, PhZnCl was indeed found to be unreactive towards aryl ethers under Ni catalysis and the use of dianionic zincates was necessary to enable smooth ethereal cross-couplings under mild conditions.[42] This is attributed to the less polar C–Zn bond and reduced nucleophilicity of PhZnCl, in comparison to PhLi and PhMgBr, and, hence, decreased σ-donor ability, which is a key interaction in the stabilisation and formation of the anionic nickelate intermediate.

Anionic Ni(0)-ate intermediates have been loosely proposed in many other Ni-catalysed cross-couplings of C–OMe bonds using organolithium and Grignard reagents,[12] [43] but support for these claims is largely based on the mild reaction conditions that are employed which ultimately rules out a concerted oxidative addition pathway (vide supra). Evidence for nickel(0)-ate intermediates has been identified by theoretical means in the Ni-catalysed homocoupling of aryl ethers using Mg-anthracene as the reductant.[44] This study proposed that the cooperativity between the Lewis acidic magnesium and electron-rich nickel centre is key to facilitating smooth C–O bond cleavage. Similar heterobimetallic mechanisms have been recently observed using a well-defined Rh/Al catalyst for the reduction and borylation of aryl ethers.[45]

Beyond C–C bond forming reactions using organometallic nucleophiles, anionic nickelates have also been proposed in the ipso-silylation of aryl ethers.[46] [47] [48] These reactions proceed under very mild conditions using Et3Si–BPin and KO t Bu with Ni(COD)2 (1 mol%) as the catalyst and without the need for external ligands.[46] Furthermore, this reaction also works with unbiased, non-π-extended, anisole derivatives which contrasts with many other transformation of aryl ethers (vide infra). Empirical studies into the reaction mechanism ruled out single electron transfer (SET) processes or the involvement of silyl radicals, and instead pointed towards the generation of silyl anions, formed in situ from Et3Si–BPin and KO t Bu. This is supported by the observation that PhMe2SiLi works directly as a nucleophilic coupling partner in the absence of KO t Bu. A discrete Ni(0)-ate complex, [(COD)Ni–SiEt3]–K+, was proposed as the key intermediate, but attempts to isolate this species or obtain direct spectroscopic evidence failed. This is unsurprising however given the extreme sensitivity and short lifetimes that have been previously reported for anionic nickelates derived from Ni(0)-olefin complexes and polar organometallic reagents.[49] [50] [51] [52] [53] [54] [55]

Subsequent DFT studies by Yu and Fu have been reported which systematically investigate the mechanism for the Ni-catalysed ipso-silylation of aryl ethers.[47] The formation of a silyl anion from Et3Si–BPin and KO t Bu was first probed and identified 32 as the resting state of the reaction (Scheme [12]). Cleavage of the silicon–boron bond proceeds with a barrier of +14.2 kcal mol–1 to give 33 that can deliver the silyl anion to Ni(0) and undergo π-coordination of the aryl ether substrate (anisole) to give either intermediate 34 or 35, which have comparable energies. Unlike, the discrete Ni(0)-ate complex, [(COD)Ni–SiEt3]–K+, proposed by Martin,[46] intermediates 34 and 35 both retain ( t BuO)BPin within the structure, which essentially solvates the potassium cation. A related pathway using THF as the solvent was also examined, and whilst a reasonable route could be identified, the overall energy barriers were marginally higher (~3 kcal mol–1), accounting for the reduced yields observed experimentally in THF compared to toluene.[46] The ipso-silylation reaction was also found to be extremely particular to the base employed with both KOMe and NaO t Bu proving completely ineffective. Computationally, replacing KO t Bu with KOMe affords a much more stable silyl boronate (akin to 32) that ultimately increases the barrier of the reaction by ~9 kcal mol–1. For NaO t Bu the overall impact was less dramatic, however, (~2 kcal mol–1) and this was attributed to subtleties relating to non-covalent interactions.[47]

Zoom Image
Scheme 12 Computed pathway for the formation of anionic Ni(0)-silyl complexes

Several different pathways were explored from intermediates 34 and 35, but mechanisms involving direct OMe abstraction or nucleophilic aromatic addition by the silyl anion were calculated to have inaccessible barriers that do not account for the mild conditions employed experimentally. Instead, Caryl–OMe bond cleavage occurs via an oxidative addition pathway at Ni and can proceed with comparable energies from either 34 and 35 to deliver the same Ni(II) intermediate, 36 (Scheme [13]). A marginally lower energy pathway (+22.9 vs. +23.2 kcal mol–1) is located from 34 since the K···OMe interaction polarises the Caryl–OMe bond and enhances the leaving ability of the OMe group. Furthermore, since this non-classical oxidative addition process simultaneously constructs a Caryl–Ni bond whilst capturing the OMe group by potassium, the formation of unsupported Ni(II)-OMe intermediates is avoided.[16] [17] Subsequent reductive elimination from 36 proceeds with a low barrier of +8.3 kcal mol–1 to give Ph–SiEt3 and [( t BuO)BPin(OMe)]–K+ and regenerate Ni(COD)2.

Zoom Image
Scheme 13 Lowest energy pathway for the oxidative addition of anisole to 34
Zoom Image
Scheme 14 Nucleophilic aromatic substitution pathway for the ipso-silylation of 2-methoxynaphthalene

A separate DFT study into the ipso-silylation of aryl ethers has been reported by Avasare and co-workers,[48] which identifies alternative pathways and instead focuses on 2-methoxynaphthalene as the model substrate. The formation of a silyl anion from Et3Si–BPin and KO t Bu was again the starting point for investigations. Contrastingly to previous theoretical studies,[47] the second equivalent of KO t Bu (which is required under experimental conditions) is taken into consideration, and proposed to facilitate the release of Et3SiK through formation of [( t BuO)2BPin]–K+. Co-complexation of Et3SiK with Ni(COD)2 then affords the anionic nickelate complex, [(η2-COD)2Ni–SiEt3]–K+ 37, in good agreement with Martin’s proposal.[46] From this key Ni(0)-ate intermediate, two different mechanisms were explored. In the first pathway, a redox-neutral nucleophilic aromatic substitution reaction is proposed (Scheme [14]) and proceeds with a calculated barrier of +22.9 kcal mol–1. Displacement of COD and η2-coordination of 2-methoxynaphthalene to the anionic nickelate affords intermediate 38, which is suitably primed for nucleophilic attack by the silyl anion at the ipso-carbon as shown in transition state 39. This results in a loss of aromaticity in intermediate 40 that is subsequently restored on cleavage of the C–OMe bond in 41 to give 42, which then releases the product and byproduct (KOMe) and regenerates Ni(COD)2. Complexation between the ethereal oxygen and K+ is proposed to facilitate C–OMe bond cleavage, as well as assisting in the retention of aromaticity, however specific details are unclear. A non-classical oxidative addition pathway involving Ni(0)-ate intermediates, which bears similarities to the mechanism proposed by Wang and Uchiyama[18] [19] (see Scheme [11]), was also examined but was found to have an activation barrier of +31.7 kcal mol–1, indicating it is less favourable than nucleophilic aromatic substitution for 2-methoxynaphthalene. Using anisole as the aryl ether substrate, nucleophilic aromatic substitution proceeds with a prohibitively high energy barrier of +41.9 kcal mol–1; this is inconsistent with experimental findings, however,[46] raising some doubts over the true mechanism involved in these transformations.

Anionic nickelates derived from silyl anions have more recently been implicated in the enantiospecific silylation of benzyl methyl ethers.[56] [57] 31P NMR spectroscopy studies support that an Ni(0)-ate complex of the constitution [LnNi–SiMe3]–[MgI(solv)]+ is formed on addition of Me3SiMgI(TMEDA) to NiBr2(diglyme)/PCy3, or by direct treatment of the Ni(0) precursor, Ni(PCy3)2, with Me3SiMgI(TMEDA). This same signal is also observed in the reaction mixture, and cyclic voltammetry and UV-Vis spectroscopy studies suggest that this Ni(0)-ate is the predominant resting state.

Despite significant theoretical support for the involvement of anionic nickelates intermediates in the Ni-catalysed cross-coupling of aryl ethers,[18] [19] direct experimental evidence including the isolation and characterisation of these species remained a significant synthetic challenge for many years. Cornella has proposed the role of Ni(0)-ates in the low-temperature Kumada–Corriu sp2–sp3 cross-coupling of aryl or vinyl halides with alkyl Grignard reagents,[58] [59] however the isolated species contain catalytically irrelevant supporting ligands, thus potentially masking the true constitution of on-cycle intermediates. These studies however, along with earlier work on the Lewis acidity of Ni–olefin complexes,[49] [50] [51] [52] [53] [54] [55] do, nevertheless, illustrate that the formation of anionic nickelates is best viewed as an equilibrium that is often pushed towards the starting materials, meaning that careful control of the Ni(0) precursor and organometallic nucleophile would be necessary to facilitate their isolation and characterisation.

Borys and Hevia have capitalised on these findings and previous computational studies to seek experimental insights into the anionic pathway in the Ni-catalysed cross-coupling of aryl ethers.[20] By systematically assessing the co-complexation between Ni(COD)2 and PhLi, four different classes of lithium nickelates could be unambiguously characterised by multinuclear NMR spectroscopy and single-crystal X-ray diffraction (Scheme [15]). Notably, the simplest 1:1 lithium nickelate 43 only exists as a minor component in concentrated THF solutions and readily dissociates to Ni(COD)2 and the 2:1 lithium nickelate 44. This is particularly surprising given that a 1:1 lithium nickelate can be isolated from Ni(0), PhLi, and TMEDA under an ethylene atmosphere,[58] demonstrating that subtle ligand intricacies can have significant consequences. Complex 44 slowly loses half an equivalent of COD to afford the bridging complex 45, which can further co-complex with additional PhLi to give the 3:1 octanuclear species 46. Compounds 44 and 45 can also be characterised or isolated with different donors, namely TMEDA and PMDETA.[52] Whilst these Ni(0)-ates were found to be extremely sensitive to air, moisture and temperature, they can be prepared reliably in good to excellent yields, providing a unique platform for onward mechanistic studies.

Zoom Image
Scheme 15 Anionic nickelates derived from Ni(COD)2 and PhLi

Using the Ni(COD)2 catalysed cross-coupling of 2-methoxynaphthalene and PhLi as the model reaction,[37] a series of stoichiometric, catalytic, and kinetics studies were performed to elucidate the role of anionic nickelates. The choice of solvent and donor was found to have a profound impact on the reaction, since competing ortho-lithiation of 2-methoxynaphthalene is observed in bulk THF or with donor additives (DME, TMEDA, and PMDETA). Stoichiometric studies between in situ prepared 44 and 2-methoxynaphthalene also reveal that the choice of solvent/donor greatly impacts the rate of the cross-coupling reaction, with the best results obtained in non-donor arene solvents or with weaker donors such as Et2O, consistent with many other reports using polar organometallic nucleophiles.[36] [37] [38] [39] , [42] [43] This not only supports the involvement of anionic nickelates such as 44 in the cross-coupling reaction, but also suggests that the ability for the aryl ether substrate to coordinate to the lithium cation(s) in the lithium nickelate is crucial. Specifically, the coordination of the ethereal oxygen to Li would bring the substrate into proximity to the electron-rich Ni centre as well as polarising the Caryl–OMe bond to facilitate oxidative addition.[18] [19] This hypothesis was supported by the isolation of an adduct between PhLi and p-Tol-OMe, as well as the significant donor effects observed in catalysis.[20]

Kinetic studies reveal a 1st order dependence in catalyst but 0th order dependence in both PhLi and 2-methoxynaphthalene, indicating that all three components associate together prior to Caryl–OMe bond cleavage, the rate-determining step of the reaction. Monitoring the catalytic reaction by 1H NMR spectroscopy supports that a complex containing all three components is formed and that it is the initial resting state of the reaction when the concentration of both PhLi and aryl ether is high. The isolated lithium nickelates 44–46 showed comparable catalytic activity to Ni(COD)2, supporting their involvement in the catalysis, but displayed a different kinetic profile. This was attributed to an off-cycle donor dissociation equilibrium from the lithium cation(s) that is necessary to enable coordination of the aryl ether substrate,[44] and is again supported by a dramatic influence on rate when comparing THF and TMEDA solvates of 44 as catalysts for the reaction. Taking these structural, spectroscopic and kinetic data into consideration, along with related computational studies,[18] [19] a mechanism for the Ni-catalysed cross-coupling of aryl ethers was proposed (Scheme [16]). Starting from Ni(COD)2, the addition of PhLi can form either a 1:1 (J) or 2:1 lithium nickelate (M) which suggests that two pathways, an anionic (A) and dianionic (B), may be operative. It was not possible to distinguish between these cycles based on kinetic data, however the observation that the 1:1 lithium nickelate 43 only exists under specific conditions in tandem with the expected increased nucleophilicity of the 2:1 species point towards the dianionic pathway being favoured. From intermediate J in path A, coordination of the aryl ether to the lithium cation brings the substrate into proximity to the Ni centre to give intermediate K that undergoes non-classical oxidative addition to give the Ni(II) intermediate L which can subsequently reductively eliminate to give the cross-coupled product. Specific experimental insight into Caryl–OMe bond cleavage is still unclear, but this was proposed to occur via a bimetallic mechanism as proposed by Wang and Uchiyama (see transition state 29, Scheme [11]).[18] [19] The elementary reaction steps in path B are identical to path A, except here one of the phenyl carbanions acts as a strong σ-donating ligand to both increase the nucleophilicity of Ni(0) and stability of Ni(II) intermediates. Additional stoichiometric studies suggest that the formation of Ni(0)-ates is less likely for nickel-phosphine complexes due to quenched Lewis acidity at nickel, and illustrate that π-accepting ligands such as COD often play an overlooked role in catalysis.

Zoom Image
Scheme 16 Proposed anionic (A) and dianionic (B) pathways in the Ni-catalysed cross-coupling of aryl ethers

# 6

Ni(I) Intermediates

Whilst many mechanistic proposals have now been investigated in relation to the Ni-catalysed cross-coupling and hydrogenolysis of aryl ethers, these nevertheless generally operate via a Ni(0)/Ni(II) pathway, with nucleophilic aromatic substitution being the primary exception.[48] Combined experimental and theoretical studies by Gómez-Bengoa and Martin into the reductive cleavage of aryl ethers with silanes have identified Ni(I) species as the key reaction intermediates.[17] An initial induction period is observed when Ni(COD)2/PCy3 is combined with Et3SiH and 2-methoxynaphthalene, which precedes the formation of naphthalene (the product of the reaction). Given that no intermediates could be observed by NMR spectroscopy, this suggested that Ni(I) species are instead formed on addition of Et3SiH to Ni(COD)2/PCy3 and this was supported by EPR spectroscopy which gave characteristic spectra consistent with Ni(I) species. These are proposed to form via oxidative addition of Et3SiH to Ni(0), followed by comproportionation with Ni(PCy3)2 to give a dimeric Ni(I) hydride 47 and monomeric Ni(I)-SiEt3 species 48 (Scheme [17]).

Zoom Image
Scheme 17 Proposed formation of Ni(I) species from Ni(COD)2/PCy3 and Et3SiH
Zoom Image
Scheme 18 Proposed mechanism for the Ni-catalysed reductive cleavage of aryl ethers with silanes

Two different mechanisms were envisioned and investigated depending on the Ni(I) source. The high stability of the dimeric Ni(I) hydride 47 suggested that a reaction involving these intermediates must be initiated by a monomeric Ni(I) hydride, which itself has to overcome a significant energy barrier of +15.5 kcal mol–1. Based on this, and other experimental factors established from kinetic data and isotope labelling studies, mechanisms involving Ni(I)-H species were ruled out. A pathway involving the Ni(I)-silyl species 48 as the key intermediate and active catalyst was instead favoured (Scheme [18]). Coordination of 2-methoxynaphthalene and displacement of PCy3 forms intermediate 49, which undergoes Ni–SiEt3 migratory insertion and dearomatisation to give 50. Subsequently, elimination of Et3SiOMe and 1,2-migration of Ni affords 51, which then undergoes σ-bond metathesis with Et3SiH to give naphthalene and regenerate 48. The rate-determining step of the reaction was established as migratory insertion, with a total activation barrier of +32.9 kcal mol–1 consistent with the high operating temperatures required.[17] Notably, this mechanism bears some similarity to the aromatic nucleophilic substitution pathway proposed involving Ni(0)-ates in the Ni-catalysed ipso-silylation of aryl ethers (see Scheme [14]).[48]


# 7

The ‘Naphthalene Problem’

Whilst the range of nucleophiles employed in the coupling of aryl ethers has expanded significantly in recent years,[12] the scope of aryl ethers is still often limited to π-extended derivatives, a feature that is prevalent amongst Ni catalysis using inert electrophiles.[8] [9] [11] In some cases, the choice of ligand can broaden the aryl ether substrate scope, with alkylphosphines such as PCy3 or NHCs enabling non-π-extended anisole derivatives to undergo Kumada–Corriu-type cross-coupling with Grignard reagents,[39–41] thereby overcoming limitations in the original Wenkert reaction.[10] These same workhorse ligands have also played a pivotal role in broadening the scope of nucleophiles that can employed in the functionalisation of aryl ethers.[12]

Several proposals have been put forward to justify observed reactivity trends (or general lack of reactivity for anisole derivatives), and insights have been aided by a combination of experimental and theoretical studies. The coordination of the aryl ether substrate to Ni(0) via π-arene interactions is frequently proposed to precede Caryl–OMe bond cleavage.[14] , [16] [17] [18] [19] , [30] [31] [32] , [47] [48] Since π-extended aryl derivatives and electron-deficient arenes coordinate to electron-rich Ni(0) centres more favourably than neutral or electron-rich arenes[24] (i.e., unbiased anisoles), this hypothesis provides a reasonable explanation for the improved reactivity. The coordination of the aryl ether substrate to Ni(0), however, often relies on displacement of a strong σ-donating ligand (phosphines or NHCs) which is often endergonic, suggesting that this is unlikely to be the primary contributing factor to account for the ‘naphthalene problem’. The requirement to use π-extended aryl ether substrates is more apparent if a nucleophilic aromatic substitution pathway is operative,[17] [48] since these mechanisms involve dearomatised intermediates that would be considerably more stable for π-extended derivatives which help to retain partial aromaticity. Theoretical studies by Avasare[48] into the ipso-silylation of aryl ethers have shown that whilst the calculated activation barrier for 2-methoxynaphthalene via nucleophilic aromatic substitution is +22.9 kcal mol–1 (see Scheme [14]), this barrier increases significantly to +41.9 kcal mol–1 for anisole. Whilst not proceeding through formally dearomatised intermediates, π-extended aryl ethers or those bearing electron-withdrawing substituents are nevertheless expected to significantly stabilise the key transition state in which the Caryl–O bond is cleaved and hence facilitate oxidative addition to Ni(0). The use of electron-withdrawing substituents ortho to OMe has successfully broadened the scope of non-π-extended aryl ether substrates in the Ni-catalysed ipso-borylation,[60] however the mechanistic understanding of how these reactions operate is still unclear.


# 8

Conclusions and Outlook

The cross-coupling of aryl ethers has evolved significantly since Wenkert’s seminal discovery in 1979, enabling a range of nucleophiles to be employed to forge new C–C and C–heteroatom bonds from unactivated phenol-derived electrophiles. The inert nature of aryl ethers and strong Caryl–O bond has raised questions regarding the mechanisms involved in these transformations, implying that non-classical pathways which deviate from canonical Ni(0)/Ni(II) catalytic cycles are in operation. Combined experimental and theoretical investigations have identified several plausible possibilities with current evidence suggesting that different mechanisms are likely at play depending on the choice of nucleophile. For mild nucleophiles such as organoboron reagents and H2, pathways involving direct oxidative to Ni(0) are most probable, although alternative mechanisms have also been proposed. The calculated energy barriers for direct oxidative addition are consistently high, but this is nevertheless consistent with the elevated reaction temperatures, high catalyst loadings, and requirement for strongly donating ligands employed in these transformations. Stoichiometric studies into the oxidative addition of aryl methyl ethers indicates that the Ni(II)-OMe intermediates that are formed can be highly unstable and prone to β-hydride elimination. The use of Lewis acids can assist in the oxidative addition process through coordination of the ethereal oxygen thus polarising the C–O bond. This has been shown to dramatically lower the energy barrier for oxidative addition and is believed to be operative when using AlMe3 as an additive in the hydrogenolysis of benzyl methyl ethers, or using trialkylaluminium reagents directly as nucleophiles in the Ni-catalysed cross-coupling of aryl ethers. The use of polar organometallic nucleophiles such as organolithiums, Grignards, and organozincates typically operate under mild conditions, ruling out pathways involving direct or Lewis acid assisted oxidative addition to Ni(0). Theoretical and experimental studies instead support that anionic nickelates, formed by co-complexation of the Ni(0) catalyst with a polar organometallic, are the key intermediates that can still participate in a Ni(0)/(II) catalytic cycle. Lastly, nucleophilic aromatic substitution has been proposed and identified as the lowest energy pathway in certain transformations of aryl ethers, either via Ni(0)-ate intermediates or Ni(I) species.

Whilst there is still mechanistic ambiguity, it is hoped that the collective knowledge presented in this review will help facilitate the discovery and systematic development of new protocols that unlock the full synthetic potential of the Ni-catalysed cross-coupling and reductive cleavage of aryl ethers.


#
#

Conflict of Interest

The authors declare no conflict of interest.


Corresponding Authors

Andryj M. Borys
Departement für Chemie, Biochemie und Pharmazie, Universität Bern
Freiestrasse 3, 3012 Bern
Switzerland   

Eva Hevia
Departement für Chemie, Biochemie und Pharmazie, Universität Bern
Freiestrasse 3, 3012 Bern
Switzerland   

Publication History

Received: 24 February 2022

Accepted after revision: 23 March 2022

Accepted Manuscript online:
23 March 2022

Article published online:
11 May 2022

© 2022. Thieme. All rights reserved

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


   Permissions and Reprints All articles of this category
Zoom Image
Andryj M. Borysreceived his B.Sc. (2015) and Ph.D. (2018) from the University of Kent (UK), the latter under the supervision of Dr Ewan Clark. He was a postdoctoral researcher with Dr Michael Cowley at the University of Edinburgh (UK), then with Prof. Thomas Baumgartner and Prof. Chris Caputo at York University (Canada) before joining the group of Prof. Eva Hevia at Universität Bern (Switzerland) in 2020. His current research explores the synthesis and catalytic applications of s- and d-block heterobimetallic complexes. Andryj is also the creator of the Schlenk Line Survival Guide.
Eva Hevia received her received both her M.Sci. degree in Chemistry and her Ph.D. degree from the Universidad de Oviedo (Spain) in 1998 and 2002, respectively. After a three-year position at the University of Strathclyde (Glasgow, UK) working as a Marie Curie Fellow with Professor Robert Mulvey, in 2006 she took up a Royal Society University Research Fellowship and Lectureship there. Subsequently she was promoted to Full Professor in 2013. In February 2019, Eva moved to the University of Bern to take up a full professorship in Inorganic Chemistry. Research in her group focuses on polar organometallic chemistry at the crossroads of inorganic, organic, and green chemistry. Eva is an elected fellow of the European Academy of Sciences and her research has been recognized with several awards including the 2017 RSC Corday-Morgan Prize and the 2021 Arfvedson-Schlenk Prize by the GDCh.
Zoom Image
Scheme 1 Wenkert’s seminal reaction on the Ni-catalysed cross-coupling of aryl ethers
Zoom Image
Scheme 2 Key examples of proposed catalytic cycles in the Ni-catalysed cross-coupling of aryl ethers; (a) neutral Ni(0)/Ni(II) pathway; (b) anionic Ni(0)/Ni(II) pathway
Zoom Image
Scheme 3 Mechanistic insights into the oxidative addition of 2-methoxynaphthalene to Ni(COD)2/PCy3
Zoom Image
Scheme 4 Stoichiometric insights into the oxidative addition of ArOMe to Ni(0) and subsequent reductive elimination and decarbonylation reactions
Zoom Image
Scheme 5 Comparison of calculated activation barriers for the oxidative addition of 2-methoxynaphthalene to Ni(0) using PCy3 (in the absence or presence of CsF) or ICy ligands
Zoom Image
Scheme 6 Surawantanawong’s proposed mechanism for the hydrogenolysis of diphenyl ether with NHC-Ni catalysts
Zoom Image
Scheme 7 Anionic-base pathway for the oxidative addition to NHC-Ni(0)
Zoom Image
Scheme 8 Stoichiometric reactivity of (SIPr)Ni(η6-C6H6) 20
Zoom Image
Scheme 9 Hartwig’s proposed mechanism for the NHC-Ni catalysed hydrogenolysis of diaryl ethers
Zoom Image
Scheme 10 Oxidative addition of 2-methoxynaphthalene to (dcype)Ni(COD) in the absence or presence of AlEt3
Zoom Image
Scheme 11 Proposed reaction pathway for the Ni-catalysed cross-coupling of anisole and PhMgBr
Zoom Image
Scheme 12 Computed pathway for the formation of anionic Ni(0)-silyl complexes
Zoom Image
Scheme 13 Lowest energy pathway for the oxidative addition of anisole to 34
Zoom Image
Scheme 14 Nucleophilic aromatic substitution pathway for the ipso-silylation of 2-methoxynaphthalene
Zoom Image
Scheme 15 Anionic nickelates derived from Ni(COD)2 and PhLi
Zoom Image
Scheme 16 Proposed anionic (A) and dianionic (B) pathways in the Ni-catalysed cross-coupling of aryl ethers
Zoom Image
Scheme 17 Proposed formation of Ni(I) species from Ni(COD)2/PCy3 and Et3SiH
Zoom Image
Scheme 18 Proposed mechanism for the Ni-catalysed reductive cleavage of aryl ethers with silanes