Striking ligand-disproportionative Cl/aryl scrambling in a simple Au(III) system. Solvent role, driving forces and mechanisms

Abstract

Aryl rearrangements triggered by Cl⁻ extraction from trans-[Au^III(Rf)₂Cl₂]⁻ (Rf = C₆F₃Cl₂-3,5), led quickly to a mixture of [Au(Rf)₃(solv)], cis-[Au(Rf)₂Cl(solv)] and [Au(Rf)Cl₂(solv)] (solv = OEt₂, OH₂). ¹⁹F NMR and X-ray diffraction studies led us to identify the species present in solution and the role of the solvent in their formation, while DFT calculations confirm the thermodynamic basis of their evolution. Very different Rf–Rf coupling rates are found from (μ-Cl)₂[cis-Au(Rf)₂]₂ or cis-[Au(Rf)₂ClL] species (L = OEt₂, NCMe, Cl⁻) depending on the coordination strength of the ligand or solvent in the fourth position.

---

Organometallic complexes bearing pentafluorophenyl (Pf) or other fluorinated aryls have been used in chemistry for decades because of the extra stability of the M–(fluoroaryl) bonds. Their use in gold chemistry in the 1970s was decisive for an explosive development of different families of gold complexes, which later have made accessible the study of properties such as aurophilicity,^1a mesomorphism,^1b and luminescence.²

A few years ago, Toste et al. reported an extremely fast aryl–aryl coupling based in the Au^III/Au^I redox couple, occurring at very low temperatures and not needing the use of ligands specially designed for such a purpose.³ This finding boosted the interest in a topic, C–C coupling from Au^III, started forty years before.⁴ Two excellent reviews of recent advances in gold(III),⁵ the first aryl–aryl coupling reported,⁶ and some recent reductive elimination studies are worth mentioning.⁷

The work presented here started because of a casual discovery when we wanted to prepare stable gold(III) complexes containing the cis-Au^III(Rf)₂ fragment (Rf = C₆F₃Cl₂-3,5). The synthesis of complexes trans-[Au^III(Pf)₂Cl₂]⁻,⁸cis-[Au^III(Pf)₂Cl₂]⁻,⁸ and (μ-Cl)₂[Au(Pf)₂]₂,⁹ was reported long ago, but we prefer to use Rf derivatives because they greatly simplify the ¹⁹F NMR spectra (there is no intra-ring ¹⁹F–¹⁹F coupling in C₆F₃Cl₂-3,5).¹⁰ This NMR simplification is useful to discern, in mixtures, signals that with C₆F₅ can overlap due to their multiplicity. For this reason, we chose to synthesize the cis- and trans-Au^III(Rf)₂ homologues of the complexes mentioned above.

In our large experience, Rf and Pf behave identically as far as chemical reactivity of their complexes is concerned. We were surprised when, applying with Rf the same synthetic procedures reported for Pf-complexes,^8,9 we observed the formation of tris-Rf and mono-Rf complexes. This complication was not mentioned for Pf complexes in the original papers or, more likely, it went unnoticed in their syntheses. Its formation reveals the occurrence of some unexpected and unreported aryl scrambling processes. These complications deserve attention because they can be a serious problem in the context of cross-coupling catalysis. In Pd, they are a frequent source of loss of selectivity, giving rise to homocoupling products and opening the door to other undesired reaction pathways. The scrambling mechanisms operating in gold(III) are poorly known and understood. In fact, to the best of our knowledge, only three aryl rearrangement reactions, incidentally involving Au^III Pf derivatives, have been reported in recent times.^11–13 They occur in very different conditions but, in all of them, either strong oxidants (nitrosyl, selectfluor, or PhI(OAc)₂) or mixtures of gold compounds with different oxidation states are involved. In the study that follows, we take advantage of the highly informative observation of Rf complexes to uncover the origin and mechanistic details of scrambling processes in gold chemistry.

Following the procedures reported in the literature for the Pf complexes, the oxidation of (NBu₄)[Au(Rf)₂],¹⁴ using a CCl₄ solution of Cl₂, led to the expected gold(III) complex (NBu₄)(trans-[Au(Rf)₂Cl₂]) (1), which was isolated and fully characterized. Also, as expected, the cis isomer (NBu₄)(cis-[Au(Rf)₂Cl₂]) (2) was obtained upon heating a solution of 1 in toluene, and was fully characterized (Scheme 1). The X-ray structures of 1 and 2 are shown in Fig. S5 and S6, ESI,† respectively. The cis isomer is thermodynamically preferred, obeying the antisymbiotic effect that avoids strongly donor ligands in mutually trans positions,¹⁵ but the trans isomer did not isomerize unless it was submitted to heating.

Scheme 1 Synthesis of Rf dichlorodiaryl gold (III) anionic complexes and evolution with AgClO₄ (1 : 1). The cation is (NBu₄)⁺.

In the literature, the Pf analogues of 1 and 2 are indistinct precursors for the synthesis of the Pf analogue of (μ-Cl)₂[Au(Rf)₂]₂ (6) by reaction with 1 equivalent of AgClO₄.⁹ However, the reaction of 1 carried out in CH₂Cl₂/Et₂O, leads to a mixture of compounds 3–5, which requires aryl rearrangement (Scheme 1). Fig. 1 shows the ¹⁹F NMR spectrum of an aliquot of this reaction after solvent evaporation and subsequent extraction with Et₂O in order to remove NBu₄ClO₄ (see experimental details in ESI†). The F_para region (upper right corner) shows five signals corresponding to the five non-equivalent Rf groups of the three species. The multiplicity of the F_ortho signals, due to through-space F_ortho–F_ortho coupling, is perfectly studied for square planar complexes with this particular aryl,¹⁰ and allows for unambiguous assignment of a tris-aryl fragment (1 : 2 quintuplet : triplet, labelled in red and blue, respectively), and a cis-diaryl fragment with two non-equivalent Rf groups (two 1 : 1 pseudotriplets labelled in green). The solv ligand (Et₂O or adventitious water, both O-donor ligands) fills the coordination vacancies, leading to [Au(Rf)₃(solv)] (3), cis-[Au(Rf)₂Cl(solv)] (4), and trans-[Au(Rf)Cl₂(solv)] (5). This ligand-disproportionative scrambling of aryl and chloro ligands replaces in these conditions the expected formation of (μ-Cl)₂[Au(Rf)₂]₂ (6).

Fig. 1 ¹⁹F NMR spectrum of an aliquot (in Et₂O, ref. acetone-d₆) of the reaction of 1 with AgClO₄: F_ortho region with integrations. F_para region in upper right corner.

If the reaction mixture of Fig. 1 is evaporated to dryness and then dissolved in non-coordinating dry CDCl₃, only one species is observed in solution. Its ¹⁹F NMR spectrum (Fig. 2, above) displays two singlets (2 : 1) assigned, respectively, to F_ortho and F_para of the desired dimer (μ-Cl)₂[Au(Rf)₂]₂ (6). Additionally, signals of the coupling product Rf–Rf start to be observed.

Fig. 2 ¹⁹F NMR spectra of 6 formed in CDCl₃ (above), and 4 formed in Et₂O (below), ref. acetone-d₆ capillary. Asterisks are signals of Rf–Rf formed from 6.

If, once more, the CDCl₃ is removed completely and Et₂O is added, the coordinating solvent (or adventitious water in it) splits the weak Cl bridges and only cis-[Au(Rf)₂Cl(solv)] (4) is formed (Fig. 2, below, see details in ESI†). This explains why, in the absence of NMR monitoring and depending on the use of solvents, often varied in synthesis, the aryl scrambling could go unnoticed in the reported syntheses of the Pf analogue of 6.⁹ We have confirmed that under our conditions, the same unreported aryl scrambling occurs with Pf.

Slow evaporation of the mixture of products in Fig. 1 led to isolation of colourless crystals of [Au(Rf)₃(OH₂)]·2Et₂O (3·OH₂) in up to 20% yield (Fig. 3 left). This is one of the few gold(III) aquo-complexes ever reported.^16,17 The fourth coordination position in the crystal is occupied by a water molecule, in this case, H-bonded to two diethyl ether molecules. In solution, fast ligand exchange between Et₂O and water is expected. The Au–C distance of the aryl group trans to OH₂ is 1.987(12) Å, significantly shorter than those found for the other two mutually trans Rf groups, with longer Au–C distances of 2.056(7) Å. This is expected from the high trans influence of Rf and the low trans influence of OH₂.

Fig. 3 X-ray structure of 3·OH₂ (left) and 6 (right).

The dimer (μ-Cl)₂[Au(Rf)₂]₂ (6) could also be crystallized and characterized by X-ray diffraction (Fig. 3 right).

The sequence of reaction from 1 in Scheme 1 cannot be reproduced for solv = MeCN because it complexes the Ag⁺ cations and prevents quite efficiently the precipitation of AgCl, but solutions of cis-[Au(Rf)₂Cl(NCMe)] (7) are reversibly formed easily from 6 (more details in ESI†). The stability of the different species in front of Rf–Rf homocoupling follows the order 2 ≫ 7 > 4 ≫ 6, according to the following observations: (i) complex 2 requires heating in toluene above 373 K to observe reductive elimination; (ii) solutions of 6 in toluene with similar concentrations of MeCN or OEt₂ (toluene : solv = 80 : 20 molar ratio) are stable for days at room temperature, but the solutions with OEt₂ decompose much faster at 333 K (see details in ESI†); (iii) in the absence of any coordinating solvent or ligand the solutions of 6 in CDCl₃ produce fast Rf–Rf coupling at 293 K. This rate sequence supports a very plausible sequence of the coordinating ability of the fourth ligand: CDCl₃ ≪ bridging Cl < OEt₂/OH₂ < MeCN < terminal Cl⁻. The RE rates observed strongly suggest coupling via dissociation to a tricoordinate intermediate. This interpretation is also in line with the extremely fast Ar–CF₃ coupling from [Au(Ar)(CF₃)I(PR₃)] in CH₂Cl₂ by treatment with AgSbF₆,¹⁸ and the high-energy barrier for Ar–Ar′ coupling in four-coordinated [Au(Ar)(Ar′)Cl(PR₃)].¹⁹ Dissociative coupling mechanisms are common in isoelectronic and isostructural Pd^II complexes.²⁰

Stoichiometric abstraction of one halide from (NBu₄)(cis-[Au(Rf)₂Cl₂]) (2) in CH₂Cl₂/Et₂O gives selective access to fairly stable solutions of cis-[Au(Rf)₂Cl(solv)] (4), preventing the formation of 3 and 5. In non-coordinating CDCl₃, 6 is formed but it has only limited stability at room temperature. Other Au/Ag proportions, not illustrative for the discussion, are commented on in the ESI.† In order to understand the effect of changing these proportions, it is important to be aware that, as most of the initial AgClO₄ is not dissolved, its reaction with 1 is heterogeneous. Hence, the change of solid AgClO₄ proportion is not directly reflected in a proportional change of Ag⁺ concentration.

The previous results allow for some mechanistic analysis of the processes sketched there. First of all, a concerted mechanism for the oxidative addition of Cl₂ to [Au^IRf₂]⁻ can be excluded: it should produce cis-[Au^III(Rf)₂Cl₂]⁻ (2), which is the most stable isomer, and 1 would never be observed. Consequently, an electrophilic attack to gold by Cl–Cl, followed by Cl⁻ release and coordination must operate (Scheme 2).

Scheme 2 The most plausible mechanism for the oxidative addition of Cl₂ to the nucleophilic anionic gold(I) complex.

The thermodynamically driven 1 to 2 isomerization must follow the mechanism proposed much earlier by Kochi and Hoffmann from studies combining experiments and calculations.²¹ This isomerization consists of ligand dissociation (Cl⁻ dissociation in this case), followed by topomerization from trans-R₂ to the thermodynamically favourable cis-R₂ coordination, and ligand (Cl⁻) recoordination (Scheme 3).

Scheme 3 Isomerization via ligand dissociation plus topomerization.

If the first easy leaving chloride is precipitated with a silver salt of a non-coordinating anion, AgClO₄, the remaining chloride coordinates additionally to the empty coordination position created, to form a bridged dimer 6 in non-coordinating chloroform. As mentioned, these chloro bridges are not very strong. In fact, they are easily split by coordinating molecules such as Et₂O or OH₂, forming cis-[Au(Rf)₂Cl(solv)] (4) (Scheme 1).

At this point it is clear why the presence of a coordinating solvent can be determining for the different fate of 1. In CH₂Cl₂, CHCl₃ or toluene, and in the absence of AgClO₄ the relatively high dissociation energy of a Cl⁻, required for topomerization, makes the isomerization of 1 to 2 very slow at room temperature: heating in toluene is required for fast isomerization. When the presence of OEt₂ helps for progressive partial solubilization of AgClO₄, irreversible precipitation of AgCl produces a non-observed intermediate trans-[Au(Rf)₂Cl(solv)] (I1). Since the dissociation of the OEt₂ ligand is easier than for Cl⁻, the topomerization at room temperature to the cis isomer 4 becomes much faster (Scheme 4).

Scheme 4 Ag⁺-fuelled mechanism of the ligand-disproportionative Rf/Cl rearrangement from trans-[Au(Rf)₂Cl(solv)] in the presence of a coordinating solvent (OEt₂).

The coexistence of I1 and 4 opens a new 1 : 1 reaction pathway that involves disproportionative Cl-for-Rf exchange affording compounds 3 and 5 (Scheme 4). For this reason, they are in equimolecular proportion in the mixture (Fig. 1). The rate competition between the formation of 4, that of I1, and their consumption through the Rf/Cl mixed bridged dimer I2 (see Fig. S8, ESI†) determines the final proportion of 4vs.3 and 5 in the products. It is worth remarking that the three aryl products 3, 4 and 5 are not in equilibrium, since the reaction of 6 with Et₂O (solv) only produces 4 (Scheme 1).

Density functional theory (DFT) calculations at the wb97x-D level were carried out in order to establish the thermodynamics of the reactions in Scheme 4 (see computational details in the ESI†). The results confirm our guess that trans-[Au(Rf)₂Cl(OEt₂)] (I1) is the highest energy form.¹⁵ It is however accessible because the formation of insoluble AgCl compensates the thermodynamic balance. Its cis isomer 4 is 17.3 kcal mol⁻¹ more stable. The rearrangement of trans-[Au(Rf)₂Cl(OEt₂)] (2 equivalents) into [Au(Rf)₃(OEt₂)] (3) + trans-[Au(Rf)Cl₂(OEt₂)] (5) (more stable than its cis isomer) viaI2 is favourable by 27.5 kcal mol⁻¹, and the splitting of I2 with 2OEt₂ is also favourable by 11.5 kcal mol⁻¹. These values support that isomerization and Rf/Cl rearrangements on OEt₂ complexes are irreversible processes with low activation barriers that make the proposed intermediates unobservable. Only the fact that in the end a considerable amount of 4 survives needs to be explained by kinetic reasons: its formation is faster than its consumption rate to produce I2, and this rate difference produces neat 4 observed when the process is finished. Most probably, the rate of consumption of 4 is limited by the rate of production of I1, which depends on the solubilization of AgClO₄ in OEt₂. This is supported by the observation that adding preformed 4 does not alter the rate of formation of 3 and 5 significantly.

In summary, the initially striking variety of complexes produced from a single precursor finds a fully satisfactory explanation. This ligand-disproportionative rearrangement, arising simply depending on the choice of solvent or other often disregarded details, is a warning on possible complexities in catalysis under oxidative conditions where complexes of this kind are formed (e.g., in Ar–Ar coupling via gold(I)/gold(III) cycles), or when using gold(III) complexes for materials. For synthetic applications, (NBu₄)(cis-[Au(Ar)₂Cl₂]) (Ar = Pf, Rf) is a stable and storable most convenient starting precursor of cis-Au(Ar)₂ species.

We thank the Spanish MINECO (Projects CTQ2017-89217-P and CTQ2016-80913-P) for their financial support. M. N. P-D. is grateful to the Spanish MECD for a FPU grant.

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. (a) A. Luquin, E. Cerrada and M. Laguna, in Gold Chemistry: Applications and Future Directions in the Life Sciences, ed. Mohr, F., Wiley-VCH, Weinheim, 2009, ch. 3, pp 93–181; (b) S. Coco and P. Espinet, Gold Chemistry, ch. 8, pp 357–396.
2. (a) V. R. Bojan, J. M. L. de-Luzuriaga, E. Manso, M. Monge and M. E. Olmos, Organometallics, 2011, 30, 4486–4489; (b) M. Bachmann, O. Blacque and K. Venkatesan, Chem. – Eur. J., 2017, 23, 9451–9456; (c) T. von Arx, A. Szentkuti, T. N. Zehnder, O. Blacque and K. Venkatesan, J. Mater. Chem. C, 2017, 5, 3765–3769.
3. W. J. Wolf, M. S. Winston and F. D. Toste, Nat. Chem., 2014, 6, 159–164.
4. A. Tamaki, S. A. Magennis and J. K. Kochi, J. Am. Chem. Soc., 1974, 96, 6140–6148.
5. (a) L. Rocchigiani and M. Bochmann, Chem. Rev., 2020 DOI:10.1021/acs.chemrev.0c00552; (b) A. Nijamudheen and A. Datta, Chem. – Eur. J., 2020, 26, 1442–1487.
6. J. Vicente, M. D. Bermúdez and J. Escribano, Organometallics, 1991, 10, 3380–3384.
7. (a) A. Genoux, J. A. González, E. Merino and C. Nevado, Angew. Chem., Int. Ed., 2020, 59, 17881–17886; (b) L. Rocchigiani, J. Fernandez-Cestau, P. H. M. Budzelaar and M. Bochmann, Chem. – Eur. J., 2018, 24, 8893–8903.
8. R. Usón, A. Laguna, J. García and M. Laguna, Inorg. Chim. Acta, 1979, 37, 201–207.
9. R. Usón, A. Laguna, M. Laguna and M. Abad, J. Organomet. Chem., 1983, 249, 437–443.
10. P. Espinet, A. C. Albéniz, J. A. Casares and J. M. Martínez-Ilarduya, Coord. Chem. Rev., 2008, 252, 2180–2208.
11. For the rearrangement leading to [Au(Pf)₂(tht)₂][Au(Pf)₄] as a by-product, from mixtures of [Au^III(Pf)₃(tht)] and [Au^I(Pf)(tht)] see: J. Coetzee, W. F. Gabrielli, K. Coetzee, O. Schuster, S. D. Nogai, S. Cronje and H. G. Raubenheimer, Angew. Chem., Int. Ed., 2007, 46, 2497–2500.
12. For the formation of [Au^III(Pf)₃(NCMe)] during the oxidation of NBu₄[Au(Pf)₂] with [NO]PF₆ (not observed with its C₆Cl₅ analogue) see: A. T. Overton, J. M. L. de-Luzuriaga, M. E. Olmos and A. A. Mohamed, Organometallics, 2012, 31, 3460–3462.
13. Oxidation of [Au(Pf)(PPh₃)] with selectfluor or PhI(OAc)₂ led to the rearrangement products [Au(PPh₃)₂][Au(Pf)₄] and [Au(Pf)₃(PPh₃)] respectively: M. Hofer and C. Nevado, Eur. J. Inorg. Chem., 2012, 1338–1341.
14. E. J. Fernández, A. Laguna, J. M. L. de-Luzuriaga, M. Monge, M. Montiel, M. E. Olmos, J. Pérez, R. C. Puelles and J. C. Sáenz, Dalton Trans., 2005, 1162–1164.
15. The well-known antisymbiotic effect or transphobia, states that the trans arrangement of two strong σ-donor groups (in our case the Rf groups), is a destabilizing contribution compared to their cis arrangement: (a) R. G. Pearson, Inorg. Chem., 1973, 12, 712–713; (b) J. Vicente, A. Arcas, D. Bautista and P. G. Jones, Organometallics, 1997, 16, 2127–2138.
16. For neutral [AuMe₂(OTf)(OH₂)] see: S. Komiya, J. C. Huffman and J. K. Kochi, Inorg. Chem., 1977, 16, 2138–2140.
17. For examples of cationic gold aquo complexes see: (a) J. Serra, P. Font, E. D. Sosa Carrizo, S. Mallet-Ladeira, S. Massou, T. Parella, K. Miqueu, A. Amgoune, X. Ribas and D. Bourissou, Chem. Sci., 2018, 9, 3932–3940; (b) M. S. Messina, J. M. Stauber, M. A. Waddington, A. L. Rheingold, H. D. Maynard and A. M. Spokoyny, J. Am. Chem. Soc., 2018, 140, 7065–7069.
18. M. S. Winston, W. J. Wolf and F. D. Toste, J. Am. Chem. Soc., 2014, 136, 7777–7782.
19. K. Kang, S. Lui, T. Xu, D. Wang, X. Leng, R. Bai, Y. Lan and Q. Shen, Organometallics, 2017, 36, 4727–4740.
20. M. Pérez-Rodríguez, A. A. C. Braga, M. García-Melchor, M. H. Pérez-Temprano, J. A. Casares, G. Ujaque, A. R. de Lera, R. Álvarez, F. Maseras and P. Espinet, J. Am. Chem. Soc., 2009, 131, 3650–3657.
21. S. Komiya, T. A. Albright, R. Hoffmann and J. K. Kochi, J. Am. Chem. Soc., 1976, 98, 7255–7265.

---

Footnote

† Electronic supplementary information (ESI) available: Synthesis and full characterisation of the complexes, X-ray and computational details; 27 pages. CCDC 2025754–2025757 and 2043412. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0cc06450e

---