A Neutral “Aluminocene” Sandwich Complex: η1‐ versus η5‐Coordination Modes of a Pentaarylborole with ECp* (E=Al, Ga; Cp*=C5Me5)

Abstract The pentaaryl borole (Ph*C)4BXylF [Ph*=3,5‐tBu2(C6H3); XylF=3,5‐(CF3)2(C6H3)] reacts with low‐valent Group 13 precursors AlCp* and GaCp* by two divergent routes. In the case of [AlCp*]4, the borole reacts as an oxidising agent and accepts two electrons. Structural, spectroscopic, and computational analysis of the resulting unprecedented neutral η5‐Cp*,η5‐[(Ph*C)4BXylF] complex of AlIII revealed a strong, ionic bonding interaction. The formation of the heteroleptic borole‐cyclopentadienyl “aluminocene” leads to significant changes in the 13C NMR chemical shifts within the borole unit. In the case of the less‐reductive GaCp*, borole (Ph*C)4BXylF reacts as a Lewis acid to form a dynamic adduct with a dative 2‐center‐2‐electron Ga−B bond. The Lewis adduct was also studied structurally, spectroscopically, and computationally.


Experimental Details
General Information All manipulations requiring handling under inert conditions were carried out under argon atmosphere using standard Schlenk techniques or an MBraun Glovebox with an Ar atmosphere. Benzene was obtained from an MBraun SPS and stored over molecular sieves, toluene and ether were distilled from sodium and degassed. Hexane and pentane were distilled from Na/K alloy. THF was distilled from potassium. Benzene-d6 and toluene-d8 were distilled from potassium, degassed and stored in a glove box.
NMR spectroscopy NMR spectra were recorded with either a Bruker Avance III 400 NMR spectrometer equipped with a 5 mm BBFO ATM probe head and operating at 400. 13 27 Al and Ξ = 94.094011 % for 19 F. [1] 1 H and 13 C spectra have been referenced on specific values for the respective solvent signal. The proton and carbon signals were assigned where possible via a detailed analysis of 1 H, 13 C, 1 H-1 H COSY, 1 H-1 H NOESY, 1 H-13 C HSQC, 1 H-13 C HMBC NMR spectra.
Young-type teflon-valve borosilicate NMR tubes have been used throughout the study.

Mass spectrometry
Mass spectra were recorded by the Zentrale Analytik within the Faculty of Chemistry, Göttingen applying a Liquid Injection Field Desorption Ionisation-technique on a JEOL accuTOF instrument with an inert-sample application setup under argon atmosphere. The injection capillary was washed several times with dry, distilled and inertly injected toluene before the samples were injected. Samples usually had a concentration of 1 -2 mmol/L in toluene and were prepared in a glovebox.

Crystallographic details
Crystals suitable for X-ray analysis grow from benzene solutions carefully concentrated at ambient temperature by evaporation and storage of the very concentrated liquid for a few days.
Opposed to ambient atmosphere the crystals suspended in oil rapidly lose colour and crystallinity, and crystal examination and picking was performed using an XTEMP-setup.
Crystal crop from benzene: S15 Tabulated crystallographic data for 2.

Crystallographic Details
Data Acquisition and Processing X-ray data for 1, and 2 were collected on Bruker APEX II CCD diffractometers with either Mo Kα radiation from a IµS or spinning anode source. The data were integrated using SAINT implemented in Brukers APEX3 programme suite. [6] SADABS [7] or TWINABS were used for multi-scan absorption correction. [8] Structure solution was performed with SHELXT [9] and refined using SHELXL [10] along the graphical user interphase of ShelXle. [11] In some cases DSR has been applied to treat disordered solvent molecules. [12] All hydrogen atoms were placed with a riding model. Further details on the individual data sets are tabulated in the analytical section of each compound. All structures were deposited with the CCSD.

Crystallographic and Refinement Details 1
Crystals of compound 1 were obtained from three different solvents (toluene, benzene and hexane) from concentrate solutions at ambient or low temperature (-40°C). The crystals are stable under argon atmosphere but lose their crystallinity under ambient conditions in inert oil within minutes. Crystals were therefore mounted with an XTEMP 2 device.
As the crystals of 1 from benzene were twinned, the two reciprocal lattices were sorted using RLATT from within the Bruker Apex 3 2018.7-2 GUI. All three datasets were integrated using SAINT 8.38A.
All three structures showed disorder within the solvent molecules in solvent accessible voids and within the majority (the entire borole sub unit) of the structure itself, which in consequence results in very poor intensity of reflections with a resolution higher than about 1.2 Å. To fit the solvent molecules with as little parameters as possible, the solvent molecules within the moieties were fitted using the SQUEEZE model, as implemented in PLATON. [13] The disorder of the Ph* and Xyl F groups were treated differently. The Ph* moiety was modelled using a modified mesityl group as included in the DSR programme with all the non tBu-methyl group positions being refined as a rigid-body. [12] The positions of the bound methyl groups were refined freely (see figure on the right).
Within the five membered borole unit, Cɑ and Cβ positions were restrained to have similar 1,2 and 1,3 distances. The resulting target symmetry of the restrains would be equivalent to a mirror plane through the boron atom and the opposing carbon-carbon bond. All C-C distances from the borole ring to the outer substituents were refined to be equivalent as well. All tert-butyl groups were restrained to have similar 1,2 and 1,3 distances. Equivalent restraints were applied to the trifluoromethyl groups.
Atomic displacement parameters of atoms within the disordered borole moiety were refined to be have similar Uij components to their neighbours (SIMU). Additionally, rigid body restraints for the atomic displacement parameters were applied to these atoms (RIGU).
With the very similar electron density pattern of a (C-Ph*) vs a (B-Xyl F ) moiety, the quasi five-fold symmetry, as well as the disorder, pose the question, whether there are additional orientations. All putative combinations of boron positions for the two disorders were evaluated, with the reported structures showing a significantly lower R-value than the alternatives. The difference can, in large part, be attributed to the fit of the CF3 groups. The model should therefore represent the two main positions of the borole moiety. However, due to the nature of the disorder and the limited resolution, additional minor occupation where the ring overlaps but is rotated differently, cannot be ruled out.
Despite the considerable efforts the resulting data-to-parameter ratio was still low for all three structures. This is an inherent result from the structure itself, as already mentioned before. However, the derived features are similar between all three structures and consistent with all other experimental and especially theoretical results.
Representation of the used rigid group. The positions of the red atoms were refined as rigid group, the black methyl groups were refined freely.

S24
Tabulated values for the key structural features of the "Aluminocene" 1 from various data sets. Please note, that there are differences of the Al-B distances in Disorder 1 and Disorder 2. This may indicate that the exact assignment/modelling of (C-Ph*) vs.(B-Xyl F )-units may be incomplete. Depictions of the disordered borole subunit within the molecule 1. Part 1 (Blue), Part 2 (orange). The second fragment is a borole unit rotated by ca. 36° with an inversion of the paddlewheel tilt of the aryl groups. This major disorder, along with further disorder within the t-Bu groups causes the low resolution of the obtainable data.

Refinement Details 2
The structure contains one molecule of lattice benzene, which is disordered modelled using SIMU, RIGU and SAME commands. Two tert.-butyl groups and a CF3 group are disordered and each modelled over two positions using SIMU, RIGU and SAME commands.

Structure Optimisation, Frequency Calculation and Thermochemical Approximations
For thermochemical approximations, structures were optimised with Gaussian09.D01 [14] applying the BP86 functional [15] and Grimmes D3 dispersion correction [16] with def2-SVP [17] basis sets on all elements. Frequency calculations were performed on these structures and absence of imaginary frequencies confirmed true local minima on the potential energy surface.
Thermochemical corrections stem from these calculations. Single point energies were calculated on these structures using a def2-TZVP basis set on all atoms.
[a] Thermochemical corrections stem from BP86-D3-def-SVP optimisation and frequency calculations. S26 Summary GIAO-NMR computations Computational examination was performed using ORCA (version 4.1.). [18] For numerical accuracy, a gridsize of "5" and a final step gridsize of "6" is applied. GIAO-NMR spectroscopic properties were calculated as implemented as the default in ORCA4.1 applying RIJK-PBE0 [19] functional on structures previously optimised using the RI-BP86-D3BJ-def2TZVP/J model chemistry. [15,17,20] Input structures were based on X-ray structures of 2 and A. For NMR calculations of the reference set of small molecules, def2-TZVPP basis sets were chosen for B, Al and Ga and def2-TZVP for all other elements.
For the rather large molecules 1 and 2, def2-TZVPP basis sets were chosen for B, Al and Ga, while a def2-TZVP basis was chosen for the core carbon atoms (namely borole Cα and Cβ positions, the ipso-CxylF atom as well as the inner cyclopentadienyl carbon atoms). A def2-SVP basis set was applied for all other atoms.   Frontier Orbital Depictions Selected canonical frontier orbitals from BP86 calculations (vide supra) are shown. All drawn at an isosurface value of 0.04 a.u. using the programme ChemCraft for visualisation. [24] All hydrogen atoms are omitted for the sake of clarity.

LUMO HOMO
Topology Analyses Topology analyses and Bader Charge-analyses [25] were carried out using the Multiwfn programme [26] or AIMAll [27] on the RI-BP86-D3BJ-def2TZVP wave function files obtained from ORCA. To shed further light onto the structure analysis of the aluminium sandwich complex further analyses were carried out. The results from topology analyses did not differ between wavefunctions obtained from BP86 or PBE0 functional calculations and no qualitative change between def-SVP basis sets and def2-TZVPP basis sets were observed. In all cases same CP and bonding path were found giving the same molecular graphs. We further investigated the parent all hydro substituted η 5 ,η 5 -(C4BH5),(C5H5) Al complex. Structures have been optimised using both BP86 and PBE0 functional and def2-TZVPP basis sets. No imaginary frequencies were found confirming minimum structures. The geometries obtained are summarised in the following Figure  Some features of the QTAIM analyses for both calculations are depicted below. The isodensity surfaces show that electron density around the boron atom is significantly reduced when compared to the densities at Cα but also Cβ.