Organolithium aggregation as a blueprint to construct polynuclear lithium nickelate clusters

By exploiting the high aggregation of aliphatic lithium acetylides, here we report the synthesis and structural analysis of polynuclear lithium nickelate clusters in which up to 10 equivalents of organolithium can co-complex per Ni(0) centre. Exposure of the Ni(0)-ate clusters to dry air provides an alternative route to homoleptic Ni(ii)-ates.

All manipulations were carried out under an inert atmosphere of argon using standard Schlenk line 1,2 or glovebox techniques (MBraun UNILab Pro ECO, <0.1 ppm H2O and O2). Due to the extreme air, moisture, and often temperature sensitivity of described compounds, rigorously inert conditions must be maintained to allow for the isolation of crystalline and pure samples. All manipulations, except for the preparation of organolithium starting materials, must avoid the use of Teflon-coated stir bars and Teflon cannulae, and glass-coated stir bars should be used. Specific experimental details can be found below. THF was dried and distilled from Na/benzophenone and stored over 4 Å molecular sieves, then further dried and vacuum distilled over NaK2.8 or a sodium mirror. Hexane, pentane, Et2O, toluene and benzene were pre-dried using a MBraun MBSPS 5, then further dried and vacuum distilled over NaK2.8 or a sodium mirror, and stored over 4 Å molecular sieves. (Me3Si)2O was degassed, dried and distilled over CaH2 and stored over 4 Å molecular sieves. THF-d8,Tol-d8 and C6D6 were dried and vacuum distilled over NaK2.8 and stored over 4 Å molecular sieves in a glovebox prior to use. Me3Si-C≡C-Li was prepared as previously reported. 3 Ni(COD)2 was purchased from commercial sources (Sigma Aldrich or Strem). All other reagents were used as supplied without further purification.
NMR spectra were recorded on Bruker Avance III HD 300 MHz or 400 MHz spectrometers at 300 K unless otherwise specified. 1 H NMR spectra were referenced internally to the corresponding residual protio solvent peaks. CHN elemental microanalyses were performed on a Flash 2000 Organic Elemental Analyser (Thermo Scientific). Samples were prepared and crimped in tin capsules in an argon filled glovebox. Analyses were performed in triplicate, and reference standards (e.g. nicotinamide) were measured prior to use as controls.
Single crystals suitable for X-ray diffraction were grown from Et2O and pentane at -30 °C. Ni(COD)2 (28 mg, 0.1 mmol) and t Bu-C≡C-Li (79 mg, 0.9 mmol) were combined in Et2O (2.5 mL) and stirred at room temperature for 4 hours. The dark green solution was evaporated to dryness and the residues were extracted into pentane (1.5 mL), filtered through a glass wool plug and stored at -30 °C.
After 1 week, the dark green crystals were separated from the supernatant and dried under argon. Yield -53 mg (62%).
Crystalline samples were plagued with green [Li9Ni(C≡C-t Bu)9]2 and could therefore not be isolated in analytically pure form.
N.B. It was not possible to confidently assign signals in the 1 H, 7 Li or 13 C{ 1 H} NMR spectrum. See Spectra S10-11 for 1 H and 7 Li NMR spectra.
N.B. It was not possible to confidently assign signals in the 1 H, 7 Li or 13 C{ 1 H} NMR spectrum. See Spectra S12-13 for 1 H and 7 Li NMR spectra.

Synthesis of Li2(Et2O)nNi(C≡C-t Bu)4 (7)
Ni(COD)2 (55 mg, 0.2 mmol) and t Bu-C≡C-Li (172 mg, 2.0 mmol) were combined in Et2O (5 mL) and stirred at room temperature for 2 hours. The dark brown solution was exposed to dry air through the attachment of a CaCl2 filled drying tube and stirred at room temperature for 1 hour resulting in a colour change to red then pale yellow. The reaction mixture was evaporated to dryness then extracted into hexane (1 mL) and Et2O (0.5 mL), filtered through a glass wool plug, and stored at -30 °C. After 48 hours, colourless crystals of 7 were separated from the supernatant, washed with cold pentane (2 × 0.5 mL) and dried under argon. Yield -44 mg (55%).
The rational synthesis of compound 7 directly from Ni(II) precursors was also attempted, but no product could be reliably isolated.

Synthesis of Li2(Et2O)2Ni(C≡C-SiMe3)4 (8)
Ni(COD)2 (55 mg, 0.2 mmol) and Me3Si-C≡C-Li (166 mg, 2.0 mmol) were combined in Et2O (5 mL) and stirred at room temperature for 2 hours. The bright orange solution was exposed to dry air through the attachment of a CaCl2 filled drying tube and stirred at room temperature for 1 hour resulting in a colour change to red then pale yellow. The reaction mixture was evaporated to dryness then extracted into (Me3Si)2O (1 mL) and Et2O (0.2 mL), filtered through a glass wool plug, and stored at -30 °C. After 1 week, colourless crystals of 8 were separated from the supernatant, washed with cold pentane (0.5 mL), and dried under argon. Yield -42 mg (34%). N.B. It was not possible to obtain suitable elemental analysis data for 8 due to a persistent red microcrystalline impurity that contaminated isolated samples.

Oxidative Homocoupling Experiments
Ni(COD)2 (14 mg, 0.05 mmol) and t Bu-C≡C-Li (86 mg, 1.0 mmol) were combined in Et2O (5 mL) and stirred at room temperature for 2 hours. The dark brown solution was exposed to dry air through the attachment of a CaCl2 filled drying tube and stirred at room temperature for 2 hours resulting in a colour change to pale brown. The reaction was quenched with MeOH (1 mL) and hexamethylbenzene (27 mg, 0.17 mmol) was added as an internal standard. An aliquot was taken, evaporated to dryness and redissolved in CDCl3 for NMR spectroscopic analysis, which indicated a spectroscopic yield of 57% for 1,4-di-tert-butyl-1,3-diyne ( Figure S1-2).

S9
Attempts to assess the quantity of Li2(solv)nNi(C≡C-t Bu)4 or residual t Bu-C≡C-Li by quenching the reaction with other electrophiles such as Me3SiCl were inconclusive. Similarly, attempted oxidative homocoupling reaction with Me3Si-C≡C-Li were inconclusive since the formed Me3Si-C≡C-C≡C-SiMe3 product was observed to react with free Me3Si-C≡C-Li to give numerous unidentified side products.

DOSY NMR Spectroscopy
Estimated molecular weights (MW) were calculated from the diffusion coefficients established from the 1 H DOSY NMR spectrum using Stalke's external calibration curve (ECC) 5-7 method and using the residual proton signal of the deuterated solvent or 1,2,3,4-tetraphenylnaphthalene as internal standards unless otherwise specified. It should be emphasised that the current ECC method is not yet optimised for aggregates that have molecular weights above >600 g mol -1 due to the lack of suitable reference compounds. Nevertheless, the 1 H DOSY NMR spectra support that the lithium acetylides form large aggregates in the absence of bulk THF and that the lithium nickelate clusters are retained in non-donor solvents (toluene) whilst they dissociate in donor solvents (THF). Attempts to assess the solution-state aggregation of t Bu-C≡C-Li in Et2O solution using an internal calibration curve were unsuccessful due to overlap of the t Bu signal with the protio solvent signal (both Et2O or MTBE), and lack of suitable internal standards (i.e. high molecular weight and soluble in Et2O).

Variable Temperature NMR Spectroscopy
The lithium nickelate clusters [Li9Ni(C≡C-t Bu)9]2 (2) and Li10(Et2O)3Ni(C≡C-SiMe3)10 (3) were further analysed by variable temperature 1 H and 7 Li NMR spectroscopy ( Figure S20-23). Whilst the broad signals observed at room temperature split into multiple sharp signals upon cooling -80 °C, it was not possible to confidently assign any of these signals to the distinct chemical environments observed in the solid-state structures.     (Tables S1-4). In all cases, crystals immersed in an inert parabar oil were mounted at low temperatures and transferred into the nitrogen stream (100 or 173 K).
Perfluorinated oils should be avoided for the lithium nickelates.
All measurements were made on a RIGAKU Synergy S area-detector diffractometer using mirror optics monochromated Cu Kα radiation ( = 1.54184 Å) or on a RIGAKU XtaLAB Synergy R, HyPix-Arc 100 areadetector diffractometer using mirror optics monochromated Mo Kα radiation ( = 0.71073 Å). Data reduction was performed using the CrysAlisPro program. 11 The intensities were corrected for Lorentz and polarization effects, and an absorption correction based on the Gaussian method using SCALE3 ABSPACK in CrysAlisPro was applied. The structure was solved by direct methods or intrinsic phasing using SHELXT, 12 which revealed the positions of all non-hydrogen atoms of the compounds. All nonhydrogen atoms were refined anisotropically. H-atoms were assigned in geometrically calculated positions and refined using a riding model where each H-atom was assigned a fixed isotropic displacement parameter with a value equal to 1.2Ueq of its parent atom (1.5Ueq for methyl groups). Refinement of the structure was carried out on F 2 using full-matrix least-squares procedures, which minimized the function Σw(Fo 2 -Fc 2 ) 2 . The weighting scheme was based on counting statistics and included a factor to downweight the intense reflections. All calculations were performed using the SHELXL-2014/7 13 program in OLEX2. 14 For [Li10(Et2O)4(C≡C-t Bu)10] (1), a disorder model was used for parts of the structure where the occupancies of each disorder component was refined through the use of a free variable. The sum of equivalent components was constrained to 1, i.e. 100%. The structure has been checked for void areas, however none could be found. The low density connects well with the dynamic disorder behavior in the structure. Twinning can be detected at post refinement steps, however the inclusion of the twin law, did not improve the refinement.
For [Li9Ni(C≡C-t Bu)9]2 (2), Disorder model for parts of the structure where the occupancies of each disorder component was refined through the use of a free variable. The sum of equivalent components was constrained to 1, i.e. 100%. Areas containing disorder solvents were found where a satisfactory solvent model could not be achieved, therefore, a solvent mask was use The structure shows signs of twinning however a satisfactory twin law could not be found, leading to high final residual densities and R statistics.
For Li10(Et2O)3Ni(C≡C-SiMe3)10 (3), a disorder model was used for parts of the structure where the occupancies of each disorder component was refined through the use of a free variable. The sum of equivalent components was constrained to 1, i.e. 100%. Due to density warnings, a solvent mask was used to locate voids but these did not contain any electron density inside. The structure was refined as an inversion twin due to uncertainty with the Flack parameter.

S26
For Li10(Et2O)3Ni(C≡C-t Bu)10 (4), a disorder model was used for parts of the structure where the occupancies of each disorder component was refined through the use of a free variable. The sum of equivalent components was constrained to 1, i.e. 100%. Areas containing disorder solvents were found where a satisfactory solvent model could not be achieved, therefore, a solvent mask was used to include the contribution of electron density found in void areas into the calculated structure factor.