‘Ring-Walking’ Aryl Cross-Coupling Reactions Involving Palladium Aryne Intermediates

Abstract

The regiospecificity of conventional cross-coupling reactions, though advantageous for its predictability and retrosynthetic simplicity, constrains chemical-space exploration. Discovery efforts have become biased toward examining substitution patterns for which coupling partners are readily obtainable. To address this problem, we have developed a migratory (‘ring-walking’) cross-coupling approach that integrates the isomerization of arylpalladium(II) intermediates into catalytic cycles. A reversible isomerization was achieved through ligand design and the use of cesium fluoride as the base, and this process was then combined with a dynamic kinetic resolution of the regioisomeric aryl halides with a broad range of oxygen- and nitrogen-centered nucleophiles. The method permits rapid access to meta-substituted arenes from para-substituted electrophiles. This account summarizes the key mechanistic principles established during the development of these reactions.

Introduction

Metal-catalyzed cross-coupling reactions are among the most general and robust synthetic tools in the organic chemist’s repertoire and, accordingly, they have become one of the most widely employed in both discovery and manufacturing applications.[1] In these transformations, the site of bond formation is almost always dictated by the position of the activating group(s). This property, termed regiospecificity, is double-edged in nature. The predictability is helpful in the design of multistep routes and retrosynthesis; however, limitations on starting-material isomers become directly translated into restrictions on product space, heavily biasing real-world discovery campaigns in the direction of chemical space that is convenient to explore.[2] A prominent example is the historical trend against the synthesis and evaluation of meta-substituted arenes, because access to the corresponding aryl (pseudo)halides is often more costly than obtaining the ortho and/or para isomers. As part of our research into new cross-coupling techniques, we became interested in an unusual approach to overcoming these existing limitations and accessing meta-coupled products from synthetically accessible para-substituted electrophiles by incorporating aryl-isomerization processes into the catalytic mechanism.

Many impactful papers have been published on methods to address deficiencies in arene regioselection, which we cannot comprehensively survey here. However, that body of work provided both confirmation of the importance of this problem and inspiration for our potential contributions. Particularly notable examples include Edelmann and Lumb’s 2024 report, providing a three-step, one-pot, 1,2-transposition of para-substituted phenols (Scheme [1]A).[3] They were able to access unusual meta-substituted phenols with good functional-group compatibility, allowing late-stage editing of complex molecules.

Some methods have combined isomerization and subsequent functionalization in one pot. A well-known study by Yamaguchi and co-workers[4] describes a so-called ester dance that proceeds under palladium catalysis to provide isomeric esters that the authors combined with diverse further transformations, including metal-catalyzed decarboxylative and decarbonylative couplings (Scheme [1]B). In another example, a metal-free halogen-dance approach was reported by Puelo and Bandar (Scheme [1]C), who showed that halogen ring-walking on substituted arenes or 3-bromopyridines could be catalyzed by an organic superbase or by a milder hydroxide reagent, respectively, leading to substitution at the 4-position by a variety of nucleophiles.[5] In this case, selectivity is driven by an inherent kinetic preference for the 4-halo isomer to undergo substitution amid a rapid reversible isomerization process. This kind of Curtin–Hammett approach inspired the strategy that we later adopted. For many arenes, however, halogen dance typically still demands harshly basic conditions that limit its applicability as an adjunct to cross-coupling. Using Pd catalysis, the Zhang group observed 1,2-migration on indoles during the preparation of 2-arylindoles from 3-bromoindole precursors (Scheme [1]D).[6] Recently, an important breakthrough was reported by Martin and co-workers, who accessed migrated functionalized products directly by using a dinuclear nickel complex intermediate formed by C–H activation, affording access to ipso, ortho, and meta (relative to the original leaving group) C–C bond formations (Scheme [1]E).[7]

Especially key to developing our central question regarding the integration of aryl (pseudo)halide isomerization and selective cross-coupling with the newly formed isomer were a series of reports by the Buchwald group about a decade ago.[8] [9] [10] In their studies, they observed the unusual formation of regioisomeric products in the fluorination of some aryl triflates and bromides. Mechanistic studies implicated a rare ortho-deprotonation (β-hydride abstraction) from the arylpalladium(II) oxidative-addition complex, leading to the formation of a putative aryne intermediate to which HF could add back to liberate a meta fluoride (Scheme [1]F).[11] Through a rational and highly instructive series of ligand modifications, Buchwald discovered new fluorinated catalysts that completely suppressed this unwanted pathway.[12]

At the outset of this research, we asked whether we could achieve the opposite: optimize intentionally to accelerate the benzyne formation and, perhaps, render this isomerization into a rapid, even reversible, process. If so, could such a process be integrated with not only C–F coupling but also C–N, C–O, C–C, and others (Scheme [1]G)?

Reversible Isomerization of Aryl Chlorides/ Bromides

We discovered that the specific combination of a diadamantyl(biaryl)phosphine ligand for Pd, CsF, and an alcohol solvent can indeed effect a reversible isomerization of aryl chlorides and bromides.[13] The special role of fluoride mirrors the studies by Buchwald, but in our case C–F bond formation is suppressed and, therefore, C–Cl(Br) reductive elimination takes place to turn over the catalyst. CsF outperforms all other fluoride bases, such as KF, on account of its greater solubility and basicity in organic solvents, and it is possible that the strict requirement for alcohol solvents is related also to solubility. Currently, it remains unexplained why fluoride is special as a base. In the transition-state structure, Pd interacts simultaneously with the fluorine atom and two carbons of the aryl ligand; so, it is possible that the small size of fluoride is critical to achieving this unusual geometry. Because primary and secondary alcohols undergo rapid O-arylation at elevated temperatures, only a few alcohols were practical options as solvents, among which tert-amyl alcohol (t-AmylOH) was found to be best. The seemingly unique effectiveness of AdBrettPhos [2-(di-1-adamantylphosphino)-3,6-dimethoxy-2′,4′,6′-tri-iso-propyl-1,1′-biphenyl] and, to a lesser extent, t-BuBrettPhos [2-(di-tert-butylphosphino)-3,6-dimethoxy-2′,4′,6′-tri-iso-propyl-1,1′-biphenyl], as a supporting ligand was curious. Even other diadamantyl phosphines such as AdBippyPhos [5-(di(adamantan-1-yl)phosphino)-1′,3′,5′-triphenyl-1′H-1,4′-bipyrazole] did not form competent catalysts at all. DFT calculations suggest that one important factor is the ability of the 1-adamantyl (Ad) groups to force, in the ground state, an ortho C–H to reside closer to the fluoride ligand (Scheme [2A]; complex I). Thus, intermediate complex I is destabilized by larger R groups, contributing to a decrease in the barrier to the isomerization. Critically, the rate of isomerization is dependent on the catalyst and ligand, but the ultimate product ratio is roughly thermodynamic and dictated by the arene substituents. For instance, experiments starting with meta- and para-chloroanisole converge to the same final isomeric ratio.

Table 1 The Effect of the Nucleophile Addition Rate on Yields and Selectivity

^a Determined by GC analysis with 1-fluoronaphthalene as internal standard.

Dynamic Kinetic Resolution of Regio­isomers for Selective Cross-Coupling

The achievement of the aryl chloride isomerization was highly encouraging to us, as was the proof-of-concept that the benzyne formation could be reversible (and was, in itself, an enabling tool for the generation of chloride isomers under mild conditions). However, from another perspective, it was also discouraging, because the ultimate product distributions of this reversible process were, of course, thermodynamically controlled, and most aryl halides equilibrated to a low ratio of isomers (<3:1; Scheme [2]B). Obviously, more would be required to achieve cross-couplings selective for the isomerized product.

At this point, we hypothesized that we could enhance the selectivity through a dynamic kinetic resolution in which this near-thermoneutral isomerization pathway would operate in parallel with a relatively slow irreversible cross-coupling reaction in a Curtin–Hammett scheme. In practice though, with most cross-couplings being very rapid and isomerizations being fairly slow, we would be far from a Curtin–Hammett regime if a typical nucleophile were present at the beginning of the reaction. For example, the inclusion of a phenol under the optimized isomerization conditions (Table [1], entry 1) resulted in only the ‘normal’ C–O coupling reaction and any sign of aryl isomerization had disappeared! But, as we slowed the addition of the nucleophile from one initial batch to a 6, 12, or 24 hour metered process, we observed an increase in the final meta/para ratio from <0.1:1 to 7.7:1 (entries 2–4). By limiting the concentration of nucleophile, the effective activation free energy of the coupling reaction was raised by –RTln[Nu], allowing for isomerization to become rapid relative to these irreversible steps. That is, the overall energy profile for this transformation shifted from one in which benzyne formation was rate-determining, to one in which cross-coupling was, allowing us to benefit from the inherent selectivity of the reductive elimination step.

In support of this theory, a direct-competition experiment involving C–O coupling of a 1:1 meta/para mixture of aryl chlorides with a phenol under the optimized coupling conditions (Table [1], entry 6) revealed an inherent reactivity bias of 3.7:1 in favor of the meta isomer. Multiplying this selectivity factor by the previously observed thermodynamic ratio of the two chloride isomers gives 11.1:1. This Curtin–Hammett prediction is in rough agreement with the 13:1 product ratio obtained from a reaction involving slow addition of nucleophile over 12 hours (entry 5). We also verified this multiplicative relationship by adding the phenol nucleophile as a single batch to a reaction starting from a 3:1 meta/para mixture of aryl chlorides (entry 7), mimicking the steady state from slow-addition runs.

Reaction Scope

The slow-addition strategy was applied to a range of oxygen and nitrogen nucleophiles, and the rate of nucleophile addition was found to critically influence the yields and selectivities. As was obvious from the aryl chloride isomerization reactions, side reactions, including decomposition of the electrophiles, could be relatively rapid in the absence of productive coupling reactions, so maintaining some concentration of nucleophile at all times was necessary. However, dosing too rapidly led to low selectivities in general, as the isomerization processes did not have time to take place reversibly. Our proposed mechanism for the productive process is shown in Scheme [3]. Once a balance between the yield and selectivity had been achieved, the reaction was quite tolerant to common functional groups. With the attenuated Brønsted basicity of fluoride in alcohol solvents, base-sensitive groups such as indazoles (3a) were well tolerated. Nucleophiles ranging from phenols (3a, 3f, 3h) to primary alcohols (3c, 3d), anilines (3e), nitrogen heterocycles (3g), and sulfonamides (3b) all converted well, including some with significant steric demand (3f). In addition to p-substituted aryl chloride electrophiles, trisubstituted arenes and certain fused heterocycles (3d) underwent successful migratory coupling.

Initially, we feared that the method would only be applicable to p-alkoxy or p-amino substrates due to the importance of the electronic bias for cross-coupling selectivity. Although the method would still be quite useful with these limitations, we were determined to address relatively electronically neutral substrates, such as p-alkyl and p-aryl halides. Because such substrates underwent comparatively inefficient isomerizations with AdBrettPhos-derived catalysts (Scheme [2]B), the discovery of new catalysts that could accelerate the process seemed to be an appropriate next step. Evaluation of over 50 derivatives of AdBrettPhos led us to L1 (Scheme [4]), which, in conjunction with Pd, enabled enhanced isomerization rates and, ultimately, improved product ratios (Scheme [4]; 3g and 3h). A key factor in these improvements was the implementation of a delayed-addition strategy, giving the system a short period of about two hours to partially isomerize before starting the addition of the nucleophile. It is worth noting that the products, as constitutional isomers, are typically separable by flash or thin-layer silica gel chromatography, so useful yields of the desired isomer can be obtained even in cases with relatively lower selectivity. Currently, electrophiles bearing electron-withdrawing groups still fail to participate in migratory cross-couplings with reasonable yields, but we have not yet prioritized overcoming this limitation as meta substitution on these substrates can often be obtained by direct electrophilic aromatic substitution.

Conclusions and Future Directions

The development of aryne-mediated metal ring-walking reactions and their incorporation into cross-coupling cycles is part of a larger research program into altered cross-coupling reactions, with the ultimate goal of accessing many related coupling products from each pair of starting materials. Migratory cross-coupling has the potential to be a powerful tool for the repurposing of conventional cross-coupling partners to make alternative isomers of products. However, several important limitations and mechanistic questions must be addressed in the near future. For example, the power of these reactions would be greatly expanded if isomerization to and from ortho positions could be rendered more facile. Deuteration experiments indicate that the formation of the 2,3-aryne complex occurs in most systems, so there must be a reluctance to transfer Pd to the 2-position, likely due to steric clashes with the ligand in the requisite benzyne rotation or protonation steps. Further, although C–O and C–N bond formations work well currently, the question remains of how to engage nucleophiles that do not couple efficiently with AdBrettPhos-modified Pd as the catalyst. In principle, the use of co-catalysis may be a promising route to expanding the nucleophile scope, and C–C bond formation will be an important next step. Our research group, by combining efforts in rational ligand design, computational methods, and mechanistic studies, continues to investigate several potential routes to solving the ortho problem and achieving more ring-walking cross-coupling reactions. These efforts include a particular focus on understanding the benzyne formation process, factors governing its kinetics, and the exact role of the ligand, fluoride, solvent, and other reaction parameters.

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgement

DFT calculations were performed on the FASRC cluster supported by the FAS Division of Science Research Computing Group at Harvard University.

• References

• 1 Brown DG, Boström J. J. Med. Chem. 2016; 59: 4443
  Search in Google Scholar
• 2 Brown DG, Gagnon MM, Boström J. J. Med. Chem. 2015; 58: 2390
  Search in Google Scholar
• 3 Edelmann S, Lumb J.-P. Nat. Chem. 2024; 16: 1193
  Search in Google Scholar
• 4 Matsushita K, Takise R, Muto K, Yamaguchi J. Sci. Adv. 2020; 6: eaba7614
  Search in Google Scholar
• 5 Puleo TR, Bandar JS. Chem. Sci. 2020; 11: 10517
  Search in Google Scholar
• 6 Yue G, Wu Y, Wu C, Yin Z, Chen H, Wang X, Zhang Z. Tetrahedron Lett. 2017; 58: 666
  Search in Google Scholar
• 7 Odena C, Gómez-Bengoa E, Martin R. J. Am. Chem. Soc. 2024; 146: 112
  Search in Google Scholar
• 8 Watson DA, Su M, Teverovskiy G, Zhang Y, García-Fortanet J, Kinzel T, Buchwald SL. Science 2009; 325: 1661
  Search in Google Scholar
• 9 Maimone T, Milner P, Kinzel T, Zhang Y, Takase M, Buchwald SL. J. Am. Chem. Soc. 2011; 133: 18106
  Search in Google Scholar
• 10 Lee HG, Milner PJ, Buchwald SL. J. Am. Chem. Soc. 2014; 136: 3792
  Search in Google Scholar
• 11 Milner PJ, Kinzel T, Zhang Y, Buchwald SL. J. Am. Chem. Soc. 2014; 136: 15757
  Search in Google Scholar
• 12 Sather AC, Lee HG, De La Rosa VY, Yang Y, Müller P, Buchwald SL. J. Am. Chem. Soc. 2015; 137: 13433
  Search in Google Scholar
• 13 Sekiguchi Y, Onnuch P, Li Y, Liu RY. J. Am. Chem. Soc. 2025; 147: 1224
  Search in Google Scholar

Corresponding Author

Publication History

Received: 13 February 2025

Accepted after revision: 11 March 2025

Article published online:
11 April 2025

© 2025. Thieme. All rights reserved

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany

• References

• 1 Brown DG, Boström J. J. Med. Chem. 2016; 59: 4443
  Search in Google Scholar
• 2 Brown DG, Gagnon MM, Boström J. J. Med. Chem. 2015; 58: 2390
  Search in Google Scholar
• 3 Edelmann S, Lumb J.-P. Nat. Chem. 2024; 16: 1193
  Search in Google Scholar
• 4 Matsushita K, Takise R, Muto K, Yamaguchi J. Sci. Adv. 2020; 6: eaba7614
  Search in Google Scholar
• 5 Puleo TR, Bandar JS. Chem. Sci. 2020; 11: 10517
  Search in Google Scholar
• 6 Yue G, Wu Y, Wu C, Yin Z, Chen H, Wang X, Zhang Z. Tetrahedron Lett. 2017; 58: 666
  Search in Google Scholar
• 7 Odena C, Gómez-Bengoa E, Martin R. J. Am. Chem. Soc. 2024; 146: 112
  Search in Google Scholar
• 8 Watson DA, Su M, Teverovskiy G, Zhang Y, García-Fortanet J, Kinzel T, Buchwald SL. Science 2009; 325: 1661
  Search in Google Scholar
• 9 Maimone T, Milner P, Kinzel T, Zhang Y, Takase M, Buchwald SL. J. Am. Chem. Soc. 2011; 133: 18106
  Search in Google Scholar
• 10 Lee HG, Milner PJ, Buchwald SL. J. Am. Chem. Soc. 2014; 136: 3792
  Search in Google Scholar
• 11 Milner PJ, Kinzel T, Zhang Y, Buchwald SL. J. Am. Chem. Soc. 2014; 136: 15757
  Search in Google Scholar
• 12 Sather AC, Lee HG, De La Rosa VY, Yang Y, Müller P, Buchwald SL. J. Am. Chem. Soc. 2015; 137: 13433
  Search in Google Scholar
• 13 Sekiguchi Y, Onnuch P, Li Y, Liu RY. J. Am. Chem. Soc. 2025; 147: 1224
  Search in Google Scholar