Persilylation of ferrocene: the ultimate discipline in sterically overcrowded metal complexes

Abstract

We report the preparation and structural characterization of the first persilylated metallocene via the metalation of decabromoferrocene. Although Grignard conditions turned out to be insufficient due to the steric and electronic effects of silyl groups causing a decreased nucleophilicity of the metalated intermediates, stepwise lithium–halogen exchange yields complex mixtures of polysilylated compounds FeC₁₀DMS_nH_10−n (n = 10, 9, 8) including the targeted decasilylated ferrocene. These mixtures were successfully separated allowing a systematic study of silylation effects on ferrocene by XRD, CV, NMR and UV/vis spectroscopy supported by DFT calculations. The findings were used to develop a high-yielding and simple preparation method to generate a tenfold substituted overcrowded ferrocene, FeC₁₀DMS₈Me₂.

---

Introduction

Given the outstanding variety of possible cyclopentadienyl (Cp) complexes reported so far, cyclopentadienyl ligands with dimethylsilyl (DMS) or trimethylsilyl (TMS) groups have found widespread synthetic applications. As they are superior in steric demand to the corresponding carbon containing analogues, iso-propyl and tert-butyl,^1,2 they allow the isolation of reactive f-block^3–7 and main group complexes^8–12 and improve hydrocarbon solubility.¹³ Moreover, the introduction of silyl groups allows the generation of sensitive (redox) species due to hyperconjugation effects, allowing the isolation of highly reduced organic compounds^14–18 or low-valent actinide and lanthanide compounds, like {[C₅H_5−n(SiMe₃)_n]₃M}¹⁻ (n = 1, 2), containing metal centers with a formal oxidation state of +II as demonstrated by the Evans group.^19–24 Although higher degrees of silylation are expected to result in an increased kinetic stability of low-valent species due to steric demand and changed electronic properties of the Cp ligand, the steric overload required and the further changed electronic properties of the Cp ligands pose a formidable synthetic challenge.

So far, only one transition metal complex with a pentakis(dimethylsilyl) substituted Cp ligand is known in the literature. Sünkel and Hofmann prepared [Mn(CO)₃(C₅DMS₅)] by iterative lithiation and silylation of [Mn(CO)₃(C₅Br₅)] (Scheme 1, top).²⁵ Although the preparation of [C₅DMS₅]⁻ by the reaction of C₅DMS₆ with MeLi has been reported,²⁶ it has never been used for follow-up chemistry. Here, the corresponding C₅DMS₆ was prepared by the reaction of C₅Br₆ with Mg and DMSCl. Claims in the Russian literature regarding the preparation of C₅TMS₆ ²⁷ were not reproducible. In contrast, the synthesis of [C₅TMS₃H₂]⁻ is well established and the ligand is easily coordinated in reactions with metal halides.^28–31 However, the synthetic challenge becomes more demanding for metallocenes due to the repulsion of the sterically crowded Cp ligands in the “sandwich” structure. Okuda achieved high silylation degrees by the reaction of iron(II) halides with tris(1,2,4-TMS)cyclopentadienyllithium to yield 1,1′,2,2′,4,4′-hexakis(trimethylsilyl)ferrocene. As a consequence of the steric demand, a distorted structure of the silylated Cp ligands is observed.³² Butler et al. reported an alternative approach for hexakis(trimethylsilyl)ferrocene including a lithium–halogen exchange of 1,1′,2,2′,3,3′-hexabromoferrocene followed by silylation with TMSCl.³³ Recently, some of us reported an improved synthetic protocol for decabromoferrocene³⁴ by the bromination of [FeC₁₀(Hg(O₂CC₃H₇)₁₀].³⁵ The reaction pathway emerging from perbrominated ferrocene is supposed to be a promising route to persilylation. In the context of ferrocene as a highly versatile model compound in organometallic chemistry,³⁶ we were interested in synthesizing the first persilylated ferrocene to investigate the resulting steric and electronic effects.

Scheme 1 Overview of synthetically prepared persilylated compounds. Top: persilylation of [Mn(CO)₃(C₅Br₅)] by iterative lithium–halide exchange and reaction with DMSCl. Middle: silylation of [FeC₁₀Br₁₀] with DMSCl via lithium–halogen exchange and Grignard reaction. Bottom: the corresponding mass spectra of the conversions.

Results and discussion

Unsubstituted ferrocene

One major challenge in converting [FeC₁₀Br₁₀] effectively into the subsequent silylated ferrocene was to avoid trapping of protons by basic intermediates. Therefore, the reaction itself and especially the used solvents and silanes as well as the glassware had to be made as dry as possible (ESI, S1†). The persilylation of [FeC₁₀Br₁₀] was carried out by a direct reaction with elemental magnesium or by bromine–lithium exchange followed by subsequent trapping with a suitable DMS electrophile (Scheme 1, bottom). Herein, the degree of silylation was followed by mass spectrometry.

All reactions including Grignard formation were carried out using DMSCl or DMSOTf in tetrahydrofuran (THF) by varying the conditions like temperature and reaction time (ESI, Table S1†). For all reactions, mass spectrometry revealed that the Grignard formation was complete as no brominated species were observed. However, the silylation step seems to proceed very slowly at high degrees of silylation, as the formation of decasilylferrocene (m/z = 766.2530 for [FeC₁₀DMS₁₀]⁺) was only observed in traces. Instead, mixtures of [FeC₁₀DMS₇H₃] (m/z = 592.1711 for [FeC₁₀DMS₇H₃]⁺) and [FeC₁₀DMS₈H₂] (m/z = 650.1946 for [FeC₁₀DMS₈H₂]⁺) as main products were obtained (Scheme 1, bottom) in the presence of DMSCl indicating that the nucleophilicity of the Grignard species {FeC₁₀DMS_n(MgBr)_10−n} (n = 8, 9, 10) is decreased with every DMS group attached to it. Therefore, additional reactions were carried out with DMSOTf as a more electrophilic silylation reagent. In the reactions including DMSOTf, however, silylation of the attached DMS groups takes place rather than the formation of persilylferrocene, causing oligomerization of the attached silyl groups as well as polymerization of the solvent.

As the generated silylated Grignard intermediates seem to be too unreactive for further silylation, to give the target [FeC₁₀DMS₁₀], organolithium compounds were used instead. Herein, lithium is introduced by lithium–halogen exchange with ^tBuLi and FeC₁₀Br₁₀ followed by the addition of DMSCl. The introduction of lithium in the ortho-position to bromine can cause a subsequent elimination of lithium bromide, which is known for several aryls followed by aryne formation and decomposition.^37,38 Therefore, a lithiation agent and an electrophile were added at −100 °C. The stability of the lithiated species varies depending on the solvents. While in n-pentane the compounds only decompose slowly at room temperature, which can be explained by the low solubility in alkanes, in THF and Et₂O decomposition was observed above ca. −80 °C. The silylation itself was mainly observed between ca. −80 °C and room temperature in THF and in n-pentane between −50 °C to room temperature indicating that the silylation proceeds faster in THF. However, in Et₂O the lithiated intermediates decompose before the silylation process starts. The mass spectra of the reactions in THF and n-pentane still revealed incomplete substitution as after the first lithiation attempt with up to 80 equivalents of ^tBuLi, in both cases brominated species were still visible at m/z = 672.0876 for [FeC₁₀DMS₇BrH₂]⁺, m/z = 730.1196 for [FeC₁₀DMS₈BrH]⁺, m/z = 750.0010 for [FeC₁₀DMS₇Br₂H]⁺, m/z = 788.1330 for [FeC₁₀DMS₉Br]⁺ and m/z = 808.0328 for [FeC₁₀DMS₈Br₂]⁺ (ESI, Fig. S57†). Here, the compounds [FeC₁₀DMS₈Br₂] (1) and [FeC₁₀DMS₉Br] (2) were isolated and characterized (see the ESI†). Due to the presence of brominated species in the mass spectra, we exclude the formation of [FeC₁₀Li₁₀] as the main intermediate of the reaction. Improved conversion is achieved by multiple metalation cycles. Here, the choice of the solvent becomes more important after multiple lithiation and silylation iterations. While after four iterations in n-pentane brominated species are still visible in the mass spectra, in THF three iterations are sufficient to transform all brominated compounds. However, the formation of side products at m/z = 606.1510 (FeC₂₅H₅₄Si₇), m/z = 680.2075 (FeC₂₈H₆₄Si₈) and m/z_found = 738.2315 (FeC₃₀H₇₀Si₉) is observed (Fig. 1, bottom). As n-pentane still offers sufficient reaction speed it was chosen to be the most suitable solvent for the first lithiation and later changed to THF to reduce side reactions. Nevertheless, via this approach we were able to isolate the desired [FeC₁₀(DMS)₁₀] (6) in milligram amounts.

Fig. 1 Possible regioisomers of [FeC₁₀DMS₈H₂] and [FeC₁₀DMS₇H₃].

The persilylated derivative is only poorly soluble in most organic solvents but can consequently be easily separated by crystallization from THF or n-pentane to afford compound 6 as a purple solid. Further purification of the remaining polysilylated metallocene mixtures was accomplished by reverse phase HPLC, using methanol/water as the eluent. Starting from the crude mixture not only were the hepta-, octa-, nona- and decakis(dimethylsilyl)ferrocene separated, but also their corresponding regioisomers in milligram scale. Considering the heptakis- and octakis(dimethylsilyl)ferrocene, in principle four or three possible regioisomers, respectively, exist. However, only four of the seven possible compounds were isolated. The distribution of the observed substitutional pattern reveals that silylation occurs equally at both Cp ligands rather than a consecutive silylation of only one Cp ligand, as metallocenes 3c, 3d and 4c with a tetrakis(dimethylsilyl)cyclopentadienyl ligand were dominantly formed (Fig. 1, purple). This indicates that silylation might have been hampered due to strong steric hindrance.

Single crystals of all isolated compounds were obtained by recrystallization from MeOH/H₂O or THF mixtures and analyzed via single-crystal XRD. The steric influence of the polysilylated cyclopentadienyl ligands becomes clear from several features: for the sterically overcrowded metallocenes a staggered conformation is preferred. In the solid state structures extraordinarily large Fe–C bond lengths up to 2.118(1) Å in comparison with that of unsubstituted ferrocene (Fe–C: 2.051(2)–2.063(2) Å (ref. 39)) are observed, indicating repulsion of the parallel cyclopentadienyl ligands. Furthermore, a bent structure for the cyclopentadienyl ligands is observed, as almost all Si atoms are found out of the Cp plane (Table 1, Si–Cp^plane). This effect is the most pronounced in the [FeC₁₀DMS₁₀] (6) and [FeC₁₀DMS₉H] (5) structures which exhibit even larger angulations in 1,1′,2,2′,4,4′-hexakis(trimethylsilyl)ferrocene (12.8° (ref. 40)). Moreover, the DMS groups on the persilylated ligand prefer to order in a highly symmetric circular manner. As a result, two sets of methyl groups are found pointing towards or away from the iron center while the silicon–hydrogen bond is in the same plane as the Cp ligand. In unsymmetrically silylated ferrocenes (e.g.3a, 4a), the repulsive interaction is compensated by tilting of the respective ferrocene (Table 1, Cp–Cp′^tilt), whereupon the parallel orientation of the DMS substituents on the less substituted Cp ligands is repealed (Fig. 2).

Fig. 2 Molecular structures in solid state of compounds [FeC₁₀DMS₈H₂] (4a) and [FeC₁₀DMS₁₀] (6). Ellipsoids are drawn with a 50% probability level. Hydrogen atoms and disorder were omitted for clarity. Dashed lines display the plane of the five-membered Cp ring. Color code: purple – silicon, orange – iron, grey – carbon.

Table 1 Selected structural parameters

Compound	Distances/Å	Angles/°
	Fe–C	Cp–Cp′^tilt	Si–Cp^plane
a Multiple metallocene molecules are found within the asymmetric unit.
FeC10DMS10 (6)	2.000(8)–2.118(2)	0.0(1)	15.1(2)–13.2(1)
FeC10DMS9H (5)^a	2.044(17)–2.115(3)	3.0(4), 3.5(1)	31.7(3)–9.5(3)
FeC10DMS8H2 (4a)	2.055(5)–2.111(6)	8.0(2)	20.4(2)–5.4(2)
FeC10DMS8H2 (4b)^a	2.051(3)–2.115(3)	2.5(1), 3.8(1)	10.1(1)–3.6(1)
FeC10DMS7H3 (3a)	2.056(2)–2.097(2)	5.0(1)	10.4(1)–1.3(1)
FeC10DMS7H3 (3b)	2.057(2)–2.099(2)	3.1(1)	11.4(1)–3.3(1)

---

The repulsive interactions cause signs of rotational hindrance regarding the DMS groups in the ¹H and ¹³C{¹H} NMR spectra. Considering rotatable silyl groups, the compound [FeC₁₀DMS₁₀] (6) should show one signal for the magnetically equivalent DMS methyl protons in the ¹H and ¹³C{¹H} NMR spectra, respectively. However, two signals at δ = 0.76 and 0.09 ppm with an integral ratio of 1 : 1 in the ¹H spectrum and at δ = 2.41 and −0.18 ppm in the ¹³C{¹H} spectrum are observed instead, as the pairs of methyl groups do not resolve together and are consequently chemically inequivalent. A similar behavior is observed for compounds 5 and 4b with [C₅(DMS)₄H]⁻ ligands. The metallocenes with threefold silylated cyclopentadienyl ligands exhibit lower rotational barriers for the silyl groups next to the unsubstituted position. Hence, heptasilyl isomers 3a and 3b do not show the expected four methyl signals in their ¹³C{¹H} NMR spectra, but instead six signals, while the octasilyl compound 4a exhibits four differentiated ¹³C{¹H} signals instead of three expected in the absence of rotational hindrance.

Substitution of the cyclopentadienyl protons typically results in significant electronic and electrochemical changes in metallocenes as HOMO and LUMO energies are shifted. In classical acceptor-substituted compounds, e.g. haloferrocenes, the substituents mainly stabilize the HOMO while weakly destabilizing the LUMO,⁴¹ resulting in high redox potentials for the reversible redox couple Fe^+II/+III.⁴² In silylated ferrocenes, e.g. [Fe(C₅(TMS)H₄)₂] (0.01 V vs. FcH), similar redox potentials to those of unsubstituted ferrocene are observed.⁴³ The electrochemical properties of the polysilylated metallocenes 3 to 6 were determined by cyclic voltammetry in THF and referenced against the FcH/FcH⁺ redox couple. All compounds exhibit a reversible one-electron oxidation with half-wave potentials between E_1/2 = 0.161 V and 0.172 V, indicating that the silyl groups do not affect the HOMO energies significantly.

In contrast to this, the UV/VIS spectra of isolated silylferrocenes 3 to 6 indicate decreasing HOMO–LUMO energy gaps. Those spectra were recorded in n-hexane (Fig. 3). Unsubstituted ferrocene exhibits an absorption band at λ = 441 nm in the visible regime corresponding to the unresolved ¹A₁ → ¹E₁ and ¹A₁ → ¹E₂ spin-allowed d–d transitions, explaining the orange color of this compound.^41,44 Incorporation of dimethylsilyl groups causes a bathochromic shift yielding a color transition from red to purple from heptakis- to decakis(dimethylsilyl)ferrocene. This shift has been observed in previously reported polysilylated ferrocenes^2,45,46 as well. However, in the spectra of silylferrocenes 3 to 6 this band is accompanied by a shoulder which is most pronounced in the spectrum of [FeC₁₀DMS₁₀] (6).

Fig. 3 Normalized UV/VIS spectra of compounds [FeC₁₀DMS₁₀] (6), [FeC₁₀DMS₉H] (5), [FeC₁₀DMS₈H₂] (4b), [FeC₁₀DMS₇H₃] (3b) and ferrocene in n-hexane.

To explain these findings, density functional theory (DFT) calculations were performed, comparing unsubstituted ferrocene, [FeCp₂], with the persilylated derivative (6) (ωB97X-D/def2-TZVPP(D)//r²SCAN-3c level, see the ESI† for details). For [FeCp₂], the expected D_5h symmetric minimum is found, with the usual D_5d transition state for the rotation of the Cp ligands (at a free energy of about 5 kJ mol⁻¹, consistent with the well-known free rotation in solution⁴⁷). Time-dependent DFT calculations of the lowest excitation energies give a band at λ = 461 nm, for a ¹A₁ → ¹E₁ transition (in D₅ notation) and two additional bands at λ = 351 nm and 512 nm, arising from the second ¹A₁ → ¹E₁ and the formally forbidden ¹A₁ → ¹E₂ transitions, respectively (ESI, Fig. S78†). The overall spectrum agrees very well with the experimentally determined spectrum. It is likely that the closeness of the two lowest excitation energies and the similar intensities (when also considering contributions from the D_5d transition state) lead to a broadening of the peak, without a clearly discernible shoulder.

Replacing the Cp–hydrogen atoms with DMS groups so that the fivefold axis of the Cp ligand is retained results in two possible conformers (of D₅ and S₁₀ symmetry, respectively) which differ in the relative orientation of the Si–H bonds on the two rings. The electronic energy difference between the two conformers vanishes, consistent with the experimentally observed disorder in the crystal structure of 6 (ESI, Fig. S79†). Based on our calculations, these two conformers lead to qualitatively different spectra (ESI, Fig. S78†). A low-energy shoulder is found for the S₁₀ structure but not for the D₅ structure. However, the D₅ conformer exhibits much larger intensities, as the ¹A₁ → ¹E₁ transition becomes dipole-allowed. We suspect that the shoulder observed experimentally (Fig. 3) may be due to a significant contribution to these low-energy regions of the spectrum from further areas on the conformational energy surface that are accessed dynamically at experimental temperatures. Contributions from different, energetically similar structures may be expected also for the hepta-, octa- and nonasilylated compounds, explaining their comparable band shapes.

Most notably, the TDDFT calculations clearly confirm the overall bathochromic shift of the spectrum upon silylation. NPA charges (ESI, Table S78†) become more positive from [FeCp₂] to 6. Most of this substituent effect seems to be structural, caused by the substantially longer Fe–C distances in 6. Computations on [FeCp₂] with the carbon positions of 6 do already provide most of the changes in charges and band positions, while finally adding the DMS substituents has a relatively small effect.

Closer analysis of the MOs and of the character of the electronic excitations shows that the degenerate HOMO in 6 is genuinely nonbonding (with d_xy and d_x²−y² character) and thus also changes very little (by less than 0.15 eV) upon expanding the Fe–C distances and upon finally adding the DMS substituents. This is consistent with the small effect of the silyl substituents on the redox potentials (see above). The low-lying excitations are out of this HOMO and out of another high-lying occupied MO with predominant metal d_z² character. This MO, which is very slightly σ-antibonding, drops in energy by about 0.25 eV upon bond expansion but goes up by 0.17 eV when adding the DMS substituents, thus staying also almost unchanged overall compared to [FeCp₂]. The changes in the band gap and low-lying excitation energies from [FeCp₂] to 6 are thus largely due to changes in the virtual MO energies. In particular, the (twofold degenerate) LUMO comes down by 0.47 eV upon bond lengthening and by another 0.20 eV upon adding the DMS substituents. The LUMO is clearly π-antibonding and becomes less antibonding for the silyl-substituted ferrocenes, thus explaining the bathochromic shifts.

1,1-Substituted ferrocenes

Although we were able to isolate the persilylated derivative [FeC₁₀DMS₁₀] (6) the demonstrated synthetic approach is inconvenient for further synthetic applications in terms of selectivity and experimental effort. In order to generate highly overcrowded metallocenes in a selective manner we modified the substitutional pattern to 1,1′-disubstituted ferrocenes. Here the presence of methyl groups should in part compensate the electron-withdrawing and steric effects of the silyl substituents. The polybrominated compound [FeC₁₀Br₈Me₂] (7) was synthesized by eightfold mercuration of 1,1′-dimethylferrocene followed by bromination with KBr₃, analogous to the synthetic protocol for decabromoferrocene.³⁴ The silylation of the corresponding metallocene was carried out with elemental magnesium and DMSCl in anhydrous THF. In contrast to the analogous reaction of decabromoferrocene the perfunctionalized derivative was obtained selectively in high yields (88%). Due to the high selectivity of the reaction step the purification of the substrates is accomplished easily by recrystallization instead of complex chromatographic methods. The obtained compound [FeC₁₀DMS₈Me₂] (8) was analyzed by NMR, UV/VIS and single-crystal XRD. The polysilylated metallocenes show similar features as [FeC₁₀DMS₁₀] (6). In the ¹H and ¹³C{¹H} NMR spectra the signals which were assigned to the methyl groups attached to silicon split into a set of four signals indicating again a hindered rotation along the Si–C^Cp axis. In the UV/VIS spectra the observed absorption maximum in the visible region at λ_max = 490 nm is again red-shifted, indicating decreased HOMO–LUMO energy gaps. The observed wavelengths are similar to those of [FeC₁₀DMS₈H₂] (4b: λ_max = 495 nm) with tetrasilylated Cp ligands. Single crystals of compound 8 were obtained upon crystallization from MeOH/THF mixtures. The molecular structure shows again elongated metal–Cp bonds of 2.082(4)–2.099(4) Å compared to those of unsubstituted ferrocene (Fe–C: 2.051(2)–2.063(2) Å 39). The Cp rings deviate from a parallel arrangement by only 4.6(2)° due to the lower steric demand of the methyl groups. The silyl groups are found out of the Cp plane (5.3–20.9°) in spite of the reduced steric demand induced by the methyl groups.

Conclusions

The focus of this work was to find appropriate methods for the generation of highly silylated ferrocenes to investigate the effects of steric overload and charge stabilization by the silyl groups.⁴⁸ Reactions were carried out by metalation of bromoferrocenes [FeC₁₀Br₁₀] and [FeC₁₀Br₈Me₂] followed by quenching with DMSCl. We observed that the introduction of silyl groups inhibits further silylation yielding only a mixture of polysilylated compounds [FeC₁₀DMS_nH_10−n] (n = 10, 9, 8) containing the targeted decasilylated ferrocene in low yields. The reactivity of the polymetalated intermediates and the selectivity of the products drastically improved when electron donating groups were applied and the corresponding octasilylated metallocenes [FeC₁₀DMS₈Me₂] were isolated purely in good yields despite the sterically demanding substituents.

Mixtures of polysilylated metallocenes [FeC₁₀DMS_nH_10−n] (n = 10, 9, 8, 7) were successfully separated by HPLC allowing a systematic study on silylation effects on ferrocene by XRD, CV, NMR and UV/VIS spectroscopy. Here, the steric overcrowding is reflected by significant rotational hindrance of the tetra- and pentasilylated cyclopentadienyl ligands regarding the DMS groups and greater ordering. Furthermore, the cyclopentadienyl ligands are found to exhibit distortions from planarity within the molecular structure.⁴⁹ The maximum UV/VIS absorbances are systematically red-shifted with further silylation indicating decreased energy gaps between occupied and virtual MOs. DFT calculations show that this is largely due to the lengthening of the Fe–C bonds caused by the steric crowding induced by the silyl substituents. The occupied frontier MOs are largely nonbonding in nature and thus not affected much by the substitution, in agreement with small changes in the redox potentials. But the metal–ligand antibonding character of the lowest-lying virtual MOs is reduced compared to that of the parent ferrocene, giving rise to the observed red shifts.

The approach of exploiting sterically demanding and electron donating groups to force high degrees of silylation allows the isolation of sterically overcrowded cyclopentadienyl compounds on multigram scale. Even a subsequent methylation to obtain the trimethylsilyl substituted analogues should be feasible to obtain chemically more inert and increasingly overloaded compounds. In the future, the syntheses of other main group, transition metal or lanthanide or actinide complexes with persilylated cyclopentadienyl ligands could be of great interest.

Author contributions

S. M. R. designed the project and wrote the manuscript. Furthermore, she performed the synthesis, recorded cyclic voltammetry and UV/VIS data as well as prepared, measured and solved the single-crystal XRD data. R. S. and P. S. R. performed experiments and analyzed the data. Both authors were supervised by S. M. R. M. R. and M. K. performed the quantum chemical calculations and revised the manuscript. C. F. designed the HPLC experiments. M. M. supervised the project and revised the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This study was funded by Deutsche Forschungsgemeinschaft (DFG) – project number: 387284271 – SFB 1349. We would like to acknowledge the assistance of the Core Facility BioSupraMol supported by the DFG.

References

1. J. Okuda, Chem. Ber., 1989, 122, 1075–1077.
2. J. Okuda and E. Herdtweck, J. Organomet. Chem., 1989, 373, 99–105.
3. M. P. Plesniak, X. Just-Baringo, F. Ortu, D. P. Mills and D. J. Procter, Chem. Commun., 2016, 52, 13503–13506.
4. D. Cui, O. Tardif and Z. Hou, J. Am. Chem. Soc., 2004, 126, 1312–1313.
5. F. Ortu, J. Lui, M. Burton, M. Fowler, A. Formanuik, M.-E. Boulon, N. F. Chilton and D. P. Mills, Inorg. Chem., 2017, 56, 2496–2505.
6. D. Cui, M. Nishiura and Z. Hou, Angew. Chem., Int. Ed., 2005, 44, 959–962.
7. W.-X. Zhang, Z. Wang, M. Nishiura, Z. Xi and Z. Hou, J. Am. Chem. Soc., 2011, 133, 5712–5715.
8. L. M. Engelhardt, P. C. Junk, C. L. Raston and A. H. White, J. Chem. Soc., Chem. Commun., 1988, 1500–1501.
9. A. H. Cowley, P. Jutzi, F. X. Kohl, J. G. Lasch, N. C. Norman and E. Schlüter, Angew. Chem., 1984, 96, 603–604.
10. M. J. Harvey, T. P. Hanusa and M. Pink, J. Chem. Soc., Dalton Trans., 2001, 1128–1130.
11. P. Jutzi, E. Schlüter, M. B. Hursthouse, A. M. Arif and R. L. Short, J. Organomet. Chem., 1986, 299, 285–295.
12. M. Westerhausen, S. Schneiderbauer, N. Makropoulos, M. Warchhold, H. Nöth, H. Piotrowski and K. Karaghiosoff, Organometallics, 2002, 21, 4335–4341.
13. C. H. Winter, X. X. Zhou and M. J. Heeg, Inorg. Chem., 1992, 31, 1808–1815.
14. A. Sekiguchi, T. Nakanishi, C. Kabuto and H. Sakurai, J. Am. Chem. Soc., 1989, 111, 3748–3750.
15. A. Sekiguchi, M. Ichinohe, C. Kabuto and H. Sakurai, Organometallics, 1995, 14, 1092–1094.
16. A. Sekiguchi, T. Matsuo, K. Ebata and H. Sakurai, Chem. Lett., 1996, 25, 1133–1134.
17. A. Sekiguchi, T. Matsuo and H. Watanabe, J. Am. Chem. Soc., 2000, 122, 5652–5653.
18. A. Sekiguchi, K. Ebata, C. Kabuto and H. Sakurai, J. Am. Chem. Soc., 1991, 113, 1465–1466.
19. J. C. Wedal, J. M. Barlow, J. W. Ziller, J. Y. Yang and W. J. Evans, Chem. Sci., 2021, 12, 8501–8511.
20. M. R. Trinh, J. C. Wedal and W. J. Evans, Dalton Trans., 2021, 50, 14384–14389.
21. R. R. Langeslay, M. E. Fieser, J. W. Ziller, F. Furche and W. J. Evans, Chem. Sci., 2015, 6, 517–521.
22. R. R. Langeslay, M. E. Fieser, J. W. Ziller, F. Furche and W. J. Evans, J. Am. Chem. Soc., 2016, 138, 4036–4045.
23. M. R. MacDonald, J. E. Bates, J. W. Ziller, F. Furche and W. J. Evans, J. Am. Chem. Soc., 2013, 135, 9857–9868.
24. M. E. Fieser, M. R. MacDonald, B. T. Krull, J. E. Bates, J. W. Ziller, F. Furche and W. J. Evans, J. Am. Chem. Soc., 2015, 137, 369–382.
25. K. Sünkel and J. Hofmann, Organometallics, 1992, 11, 3923–3925.
26. A. Sekiguchi, Y. Sugai, K. Ebata, C. Kabuto and H. Sakurai, J. Am. Chem. Soc., 1993, 115, 1144–1146.
27. M. S. Miftakhov, G. A. Tolstikov and S. I. Lomakina, Zh. Obshch. Khim., 1976, 46, 2754–2754.
28. P. Jutzi, E. Schlüter, C. Krüger and S. Pohl, Angew. Chem., Int. Ed. Engl., 1983, 22, 944–944.
29. M. A. Edelman, P. B. Hitchcock, M. F. Lappert, D.-S. Liu and S. Tian, J. Organomet. Chem., 1998, 550, 397–808.
30. H. Chen, P. Jutzi, W. Leffers, M. M. Olmstead and P. P. Power, Organometallics, 1991, 10, 1282–1286.
31. C. P. Morley and P. Jutzi, Organometallics, 1987, 6, 1084–1090.
32. J. Okuda and E. Herdtweck, Chem. Ber., 1988, 121, 1899–1905.
33. I. R. Butler, D. M. Evans, P. N. Horton, S. J. Coles and P. J. Murphy, Organometallics, 2021, 40, 600–605.
34. S. M. Rupf, I. S. Dimitrova, G. Schröder and M. Malischewski, Organometallics, 2022, 41, 1261–1267.
35. S. M. Rupf, G. Schröder, R. Sievers and M. Malischewski, Chem. – Eur. J., 2021, 27, 5125–5129.
36. P. Štěpnička, Dalton Trans., 2022, 51, 8085–8102.
37. C. H. Winter and K. N. Seneviratne, Comments Inorg. Chem., 1996, 19, 1–23.
38. J. R. Baran Jr., C. Hendrickson, D. A. Laude Jr. and R. J. Lagow, J. Org. Chem., 1992, 57, 3759–3760.
39. P. Seiler and J. D. Dunitz, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1982, 38, 1741–1745.
40. J. Okuda, J. Organomet. Chem., 1989, 376, C1–C4.
41. M. S. Inkpen, S. Du, M. Hildebrand, A. J. P. White, N. M. Harrison, T. Albrecht and N. J. Long, Organometallics, 2015, 34, 5461–5469.
42. D. A. Khobragade, S. G. Mahamulkar, L. Pospíšil, I. Cíšarová, L. Rulísek and U. Jahn, Chem. – Eur. J., 2012, 18, 12267–12277.
43. A. Abri and A. abbaszad Rafe, J. Chin. Chem. Soc., 2015, 62, 273–279.
44. Y. S. Sohn, D. N. Hendrickson and H. B. Gray, J. Am. Chem. Soc., 1971, 93, 3603–3612.
45. O. J. Curnow, G. M. Fern, S. Klaib, U. Böhme, H. Lang and R. Holze, J. Electroanal. Chem., 2005, 585, 167–171.
46. J. Okuda, R. W. Albach and E. Herdtweck, Polyhedron, 1991, 10, 1741–1748.
47. M. Rosenblum and R. B. Woodward, J. Am. Chem. Soc., 1958, 80, 5443–5449.
48. A previous version of the manuscript has been deposited on a preprint server: S. M. Rupf, R. Sievers, P. S. Riemann, M. Reimann, K. Kaupp, C. Fasting and M. Malischewski, ChemRxiv, 2023,  DOI:10.26434/chemrxiv-2023-538rg.
49. Deposition numbers CCDC 2248053–2248061 contain the supplementary crystallographic data for this paper.†.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2248053 (1), 2248054 (2), 2248060 (3a), 2248059 (3b), 2248057 (4a), 2248058 (4b), 2248056 (5) and 2248055 (6). For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3dt01133j

---