Reversible Dissociation of a Dialumene

Abstract Dialumenes are neutral AlI compounds with Al=Al multiple bonds. We report the isolation of an amidophosphine‐supported dialumene. Our X‐ray crystallographic, spectroscopic, and computational DFT analyses reveal a long and extreme trans‐bent Al=Al bond with a low dissociation energy and bond order. In solution, the dialumene can dissociate into monomeric AlI species. Reactivity studies reveal two modes of reaction: as dialumene or as aluminyl monomers.


General Considerations
All manipulations were carried out under an argon atmosphere using standard Schlenk or glovebox techniques unless stated. Reactions were carried out in glass Schlenk tubes. Due to the high sensitivity of the compounds synthesised, it was necessary to silanize the glassware prior to use. Glassware was washed with a 5% solution of dichlorodimethylsilane in toluene then allowed to dry. The glassware was then washed with methanol and dried in an oven at 110 °C for at least 16 hours prior to use.
Solvents were obtained from an inert solvent purification system and stored over 4 Å molecular sieves. C6D6, d8-toluene and d8-THF were dried over a potassium mirror then vacuum distilled and stored over 4 Å molecular sieves. Ambient temperature NMR spectra were recorded on Bruker PRO 500 MHz, AVA 400, 500 or 600MHz spectrometers. 1 H and 13 C spectra were referenced to residual solvent signals. 31 P NMR spectra were referenced to an external standard of 85% H3PO4 in H2O. Mass spectra were acquired using Thermo Mat 900 XP hi resolution double focussing sector mass spectrometer from solid samples. Elemental Analysis was performed by Elemental Microanalysis Ltd. UV-vis spectra were recorded on a Varian Cary 50 Scan UV-vis spectrometer and Shimadzu UV-1800 Spectrometer with VICI-DBS Single Cell Peltier accessory for variable temperature spectra.
Aluminium(II) hydride dimer VI and aluminium(II) iodide dimer 2 were synthesised according to a modified literature procedure. 1 Na/K was synthesised by mixing freshly cut Na and K (0.5 g each) until a uniform liquid was achieved. All other reagents were purchased from commercial suppliers and used without further purification.
Key for NMR spectroscopic assignment of ligand resonances:

Dihydrodialane VI
This compound was synthesised according to a modified literature procedure. 1 A stirred solution of NMes/P t Bu2 substituted aluminium dihydride (1.00 g, 2.5 mmol) and [ Mes NacNacMg]2 (0.99g, 1.4 mmol, 0.55 eq) in benzene (125 mL) was heated to 65 °C for 16 hours. The resulting solution was cooled to 50 °C, filtered to a flask at room temperature, concentrated to 100 mL and left to crystallise at room temperature. The mixture was filtered, and the crystals washed with benzene and dried to afford the product as a white crystalline solid with identical characterisation to that previously reported. The most common contaminant was the magnesium by-product [ Mes NacNacMg(μ-H)]2, which crystallises as large yellow block crystals. The product VI must be completely clean of this contaminant for further reactivity to Diiododialane 2. Typical yields of 400-500 mg (40-50%).

Diiododialane 2
This compound was synthesised according to a modified literature procedure. The preparation was conducted in parallel in two 250 mL round bottomed Schlenk flasks which were then combined during work-up.
Two silanized 250 mL Schlenk flasks were each charged with dihydrodialane VI (280.0 mg, 0.351 mmol) and toluene (130 mL). The mixtures were stirred vigorously and heated with a heat gun until all solid had dissolved. The solutions were stirred at room temperature for 30 minutes and then stirred vigorously whilst a solution of I2 in toluene was added dropwise (4.4 mL of 20 mg/mL solution, 0.351 mmol, 1 eq), accompanied by an immediate colour change to yellow. The solutions were stirred for 30 minutes and then concentrated to 30 mL each. The resultant yellow solutions were filtered into a 100 mL Schlenk flask and the volatiles removed in vacuo to afford a pale yellow solid (typical yields of 550-600 mg, 75-80%). Pure material can be afforded by crystallising 2 from toluene at 4 °C, however the initial solid was of sufficient purity for use in the synthesis of dialumene 1.

Synthesis of Dialumene 1
This synthesis was carried out in two portions, which were then combined for work-up. When reactions were attempted on a larger scale, initiation of the reduction was slow and increased decomposition of the product was observed.
In an Ar filled glovebox, two silanized 100 mL Schlenk flasks were each charged with Na/K (15.9 mg, 0.547 mmol, 2 eq). A solution of Al(II) iodide dimer 2 (287.2 mg, 0.274 mmol, 1 eq.) in THF (50 mL) was added to each flask quickly and the mixtures were stirred vigorously for 5 hours. After approximately 1 hour, the solutions turned intense purple (the exact initiation period depends on purity of starting materials and stirring speed). After 5 hours, the 31 P{ 1 H} NMR spectrum shows the solution contains approximately 40% product by integration. Although starting material is still present in the reaction mixture, higher reaction times result in the formation of higher amounts of by-products and no increase in yield of 1.
The volatiles were removed from both flasks, then the product was extracted into toluene (total 130 mL) and filtered into a silanized 250 mL flask. The filtrate was concentrated to 40 mL and cooled to -30 °C overnight. The solid was obtained was separated by filtration and the product was afforded as a purple crystalline solid (133 mg, 0.167 mmol, 31%). 1 was stored in the glovebox freezer at -30 °C. Single crystals suitable for X-ray diffraction were grown from a saturated toluene solution at -30 °C.

Other attempted reduction conditions to form 1
A number of other conditions were attempted for the reduction of 2 to dialumene 1. From our observations, THF solvent is important (reactions in toluene or benzene were slow or did not occur). Similarly, closely-stoichiometric quantities of reductant are essential to avoid decomposition by removal of the ligand. Short reaction times are also neccessary to limit decomposition of 1 in solution. Thus, reductions of 2 with KC8 were too slow. Reactions with dropwise addition of 2 equivalents of K(naphthalenide) or Li(naphthalenide) at room temperature afforded instant conversion to 1 in a similar conversion to Na/K, but had the added complication of naphthalene contaminant in the resulting product. We therefore chose to optimise the reaction with Na/K. Extraction of the product following reduction was attempted with pentane, but the higher solubility of 1 in toluene meant work-up in toluene was easier and faster (affording lower decomposition of 1 and higher yields).

Solubility and Stability of 1
Dialumene 1 is poorly soluble in a range of solvents. Solubility is highest in THF, benzene or toluene, but is limited to < 6 mg/mL. The solubility of 1 in hexane is lower.
1 is somewhat unstable in solution, depending on the solvent. After 6 hours in THF, benzene, or hexane, solution we found that 3%, 10%, or 20 % (respectively) of 1 had decomposed. The principal decomposition product is the dihydrodialane VI.
In the solid state, dialumene 1 decomposes slowly (days) at room temperature, and is thus best stored at −30 ºC.

Synthesis of Dialuminacyclobutane 4
A J Young's ampoule was charged with dialumene 1 (100.0 mg, 0.126 mmol) and toluene (25 mL). The solution was degassed three times by freeze/pump/thaw and the flask was then refilled with ethene at a pressure of 1 atm. The flask was shaken, and then the solution was vigorously stirred. After 20 minutes, a colour change from dark purple to pale yellow was observed. The solution was stirred for a further 30 minutes at room temperature. The volatiles were removed in vacuo and the product extracted into pentane, filtered, concentrated, and cooled to -30 °C. The resultant solid was redissolved, filtered, and recrystallised to afford 4 as colourless crystals (20.0 mg, 0.0243 mmol, 19%). The high solubility of the product precluded higher yields. Crystals suitable for X-ray diffraction were grown from a saturated pentane solution at room temperature.
Reactions carried out on an NMR scale (3 mg of dialumene 1 in 0.5 mL C6D6) in shaken tubes were complete in 5 minutes.
4 exists as three diastereomers (X, Y and Z). Overlapping resonances precluded the determination of the ratio of these isomers and their individual characterisation.

Synthesis of Dialuminacyclobutene 5
To a stirred solution of dialumene 1 (96.8 mg, 0.122 mmol, 1eq) in toluene (30 mL), a solution of diphenylacetylene (21.7 mg, 0.122 mmol, 1 eq) in 5 mL toluene was added. The resultant solution was stirred for 1 hour to afford a bright yellow solution. The volatiles were removed in vacuo and the product extracted into pentane (30 mL), concentrated, and cooled to -30 °C overnight. The first crop of solid was contaminated with the products of decomposition, so was recrystallised in pentane at -30 °C to afford 5 as a yellow crystalline solid (67.3 mg, 0.069 mmol, 57%). 5 exists as three diastereomers (41% X, 12% Y and 47% Z). Due to overlapping resonances, it was not possible to distinguish/assign individual diastereomers in the 1 H and 13 C spectra.

Synthesis of aluminacyclopropene 6
To a stirred solution of dialumene 1 (90.3 mg, 0.114 mmol, 1eq) in toluene (30 mL), a solution of diphenylacetylene (38.7 mg, 0.227 mmol, 2 eq) in 5 mL toluene was added. The resultant solution was stirred for 3 hours to afford a bright yellow solution. The volatiles were stored at -30 °C overnight. The first crop of solid was contaminated with the products of decomposition, so was recrystallised in toluene at -30 °C to afford 6 as a yellow crystalline solid (46.7 mg, 0.082 mmol, 36%). High solubility of the product in toluene precluded a higher yield.

Diastereomers of dialumene 1
Dialumene 1 contains 6 stereogenic centres: two in each norbornene unit of the ligand backbone and one at each Al atom due to the trans-bending of the Al-Al bond. These stereogenic centres result in 6 distinct possible diastereomers. This is analogous to the N,Pstabilised Al(II) hydride dimers we previously reported. 1 A full stereochemical explanation of the 6 diastereomers was provided in that previous publication.
Diastereomers A, B and C all contain trans-P atoms with respect to the Al-Al bond. These diastereomers are found in the X-ray crystal structure. Diastereomers D, E and F all contain cis-P atoms with respect to the Al-Al bond. We computed the relative energies of diastereomers A-F for dialumene 1 and found A, B and C to be very similar in energy (within error of the computational methods used). Diastereomers A-C are more stable than diastereomers D, E and F.

Diastereomers of dialuminacyclobutane 4 and dialuminacyclobutene 5
Reaction of diastereomers A, B and C of dialumene 1 in a [2+2] cycloaddition with ethylene or diphenylacetylene results in the formation of three diastereomeric products. Below is an illustration of the possible products from reaction with ethylene. The analysis applies equally to 5. Reaction of A or B with ethylene results in the formation of one diastereomeric product irrespective of addition to the 'top' or 'bottom' face of the Al=Al bond. The products from reaction of diastereomers A and B are enantiomeric, and thus have identical 31 P{ 1 H} NMR chemical shifts. The products have two 31 P resonances because the P atoms are inequivalent (due to the influence of the norbornene ring, which is in opposite orientations at each Al/P centre). In the product from reaction with ethylene (4), these are two mutually coupled doublets, but in the product from reaction with diphenylacetylene (5), these are two singlets, likely due to a change in the P-Al-Al-P dihedral angle and therefore lower value for 3 JP-P.
When ethylene adds to the third dialumene diastereomer, 1-C, addition to the top or bottom face of the Al=Al bond generates two distinct products. This is due to the relative orientation of the norbornene ligand backbone units with respect to the new C-C bridge. Both products are meso-compounds and so the P atoms are in equivalent environments, resulting in each compound appearing as a singlet in the 31 P{ 1 H} NMR spectrum of 4. Figure S3: experimental UV-vis of 1 (hexane solution) and those predicted using TD-DFT (SMD-B3LYP-D3/6-311G(2d,2p)). λmax = 567.0 nm (ε = 10362 L mol -1 cm -1 ).

UV-Vis spectroscopy of dialumene 1
A series of variable temperature UV-vis spectra of 1 in hexane or toluene were recorded from 5 to 65 °C. We observed no change in λmax over this temperature range compared to the spectra at room temperature.
Comparing the experimental UV-Vis spectrum of 1 (λmax = 567.0 nm) to that predicted by TD-DFT reveals good agreement, albeit with some blue-shift of the lowest energy absorption (δ = −47 nm, λmax = 520nm, Figure S4). TD-DFT predicts an absorption for the aluminyl monomer 3 at 410nm. We could observe no such absorption in solutions of 1 over the temperature range 5 to 65 °C, indicating that 3 is not present in substantial (observable) concentration. (the predicted absorption of 3 at 284nm arises from a HOMO to LUMO transition (Al lone pair to Al p orbital) and a HOMO-1 to LUMO+1 transition (associated with the ligand backbone), but experimentally would be obscured by absorptions (visible in the experimental spectrum) from arene π-π* transitions).

Low temperature NMR spectroscopy for dialumene 1
Variable temperature 1 H and 31 P{ 1 H} NMR spectroscopy studies were performed on d8toluene solutions of dialumene 1 over the temperature range of 188 -300 K. Due to the low solubility and stability of 1, decomposition was observed in both sets of spectra over the course of the experiment.
Other than the expected variation of chemical shift with temperature, 1 H spectra over the range 188 -300 K are very similar. Across all temperatures, the 1 H NMR spectrum contains two full sets of ligand peaks, indicating two distinct ligand environments. Figure S4: 31 P{ 1 H} NMR spectra for dialumene 1 over temperature range of 203 to 300 K. In the 1 H NMR data, two full sets of ligand resonances are observed. These are assigned to the time averaged environments of isomers A/B and C. From 1 H NMR spectra taken at low temperatures (188-243 K), it was possible to extract thermodynamic data on the interconversion of A/B and C ( Figure S5, Table S1).

Computational Methods
All electronic structure calculations were carried out using the Gaussian 16 (Revision B.01) program. 2 Initial coordinates of all compounds were extracted from the experimental singlecrystal X-ray structures. Geometries were optimised at the M062X-D3/def2SVP 3-5 level of theory without symmetry constraints, and minima were confirmed by the absence of imaginary eigenvalues in the hessian matrix, by way of a frequency calculation performed at the same level of theory.
Comparison of metrical parameters for the dialumenes I, II, and 1 indicates that optimisation at the M062X-D3/def2SVP level of theory gives excellent agreement with their experimentally determined geometries (see following Tables for summary of  The NBO program (version 6.0) was used to perform Natural Bond Orbital analyses on the optimised structures at the B3LYP-D3/6-311G(2d,2p) level of theory. 10 The topology of the electron density was analysed using QTAIM (quantum theory of atoms in molecules), as implemented in the AIMALL package. 11 Time dependent (TD-)DFT calculations were carried out at the SMD-B3LYP-D3/6-311G(2d,2p) level of theory, employing the Tamm-Dancoff Approximation (TDA) and employing hexane as the solvent (ε=1.8819). Natural Transition Orbital calculations were then carried out on selected excited states of the TD-DFT calculations, using the same level of theory, to provide an intuitive picture of the involved states. 12 NMR shielding tensors were calculated at the SMD-B3LYP-D3/6-311G(2d,2p) level of theory, using the Gauge-Independent Atomic Orbital (GIAO) 13 method and employing benzene as the solvent (ε=2.2706).
Analysis of the topology of the Electron Localisation Function (ELF) 14 , η(r), was carried out using the Multiwfn program 15 . ELF basin analysis was carried out with a grid step size of 0.10 Bohr and by integrating the electron density over the basins, the average population of each basin was obtained. The sum of the synaptic valence basin populations corresponding to the Al-Al bond, ∑iVi(Al,Al) was used as an estimate of the total basin population of the Al-Al bond.

ELF Analysis
Al 1 -Al 2 1.11 1.32 a 1.33 a -3.76 Al 1 -P 2.14 ----  Figure S24: Molecular graph for dialumene II. Figure S25: Laplacian ∇ 2 ρ(r) of the electron density in the plane containing Al1 and Al2 for dialumene II. Table S19: Selected properties of the Bond Critical Points of dialumene II. "(r) is the electron density (e/a0 3 ), ∇ 2 "(r) is the Laplacian of the electron density (e/a0 5 ), #n are eigenvalues of the Hessian of "(r) (e/a0 5 ), $ is the bond ellipticity defined as (#1/#2)-1, V(r), G(r) and H(r) represent the potential, kinetic and total energy density, respectively (Hartree/a0 3 ), |V(r)|/G(r) is the ratio of potential to kinetic energy and %(A,B) is the delocalisation index. a The presence of a non-nuclear attractor (NNA) and associated basin at the midpoint of the Al-Al bond complicates the analysis of the QTAIM parameters. In order to calculate the bond delocalisation index between the two Al atoms, one defines an effective delocalisation index by dividing the NNA basin equally between the two Al basins. Thus, if the two surrounding aluminum atoms are indicated by 1 and 2 and with the definitions of the localisation (#) and delocalisation indices (%) one finds: %(Al1,Al2)eff = %(Al1,Al2) + 0.5•%(Al1,NNA) + 0. 5•%(A2,NNA) + 0.5•#(NNA).   Figure S27: Laplacian ∇ 2 ρ(r) of the electron density in the plane containing Al1 and Al2 for 1.    Plots of the Laplacian along bond paths containing the Al centres are shown in Figure S28. Along the Al-P bond path, a distinct minimum in the 1D plot of the Laplacian can be identified corresponding to the valence shell charge concentration (VSCC) of the P atom. Located in the same atomic basin is a second, shallower region corresponding to the VSCC of the Al centre. A similar situation is also found for the Al-N bond, and in both cases this pattern is diagnostic of dative character of the Al-N and Al-P bonds. In contrast, a continuous region of charge concentration (∇ 2 !bcp < 0) is symmetrically arranged around the bcp near the midpoint of the Al-Al bond.  Figure S31: Laplacian ∇ 2 ρ(r) of the electron density in the plane containing Al1 and Al2 for dihydrodialane VI. Table S21: Selected properties of the Bond Critical Points of dihydrodialane VI. "(r) is the electron density (e/a0 3 ), ∇ 2 "(r) is the Laplacian of the electron density (e/a0 5 ), #n are eigenvalues of the Hessian of "(r) (e/a0 5 ), $ is the bond ellipticity defined as (#1/#2)-1, V(r), G(r) and H(r) represent the potential, kinetic and total energy density, respectively (Hartree/a0 3 ), |V(r)|/G(r) is the ratio of potential to kinetic energy and %(A,B) is the delocalisation index. a The presence of a non-nuclear attractor (NNA) and associated basin at the midpoint of the Al-Al bond complicates the analysis of the QTAIM parameters. In order to calculate the bond delocalisation index between the two Al atoms, one defines an effective delocalisation index by dividing the NNA basin equally between the two Al basins. Thus, if the two surrounding aluminum atoms are indicated by 1 and 2 and with the definitions of the localisation (#) and delocalisation indices (%) one finds: %(Al1,Al2)eff = %(Al1,Al2) + 0.5•%(Al1,NNA) + 0. 5•%(A2,NNA) + 0.5•#(NNA).   A dark purple block-shaped crystal with dimensions 0.17×0.11×0.05 mm 3 was mounted on a suitable support. Data were collected using a SuperNova, Dual, Cu at home/near, Atlas diffractometer operating at T = 120.01(10) K.

X-Ray Crystallography
Data were measured using w scans using CuKa radiation. The total number of runs and images was based on the strategy calculation from the program CrysAlisPro (Rigaku, V1.171.41.99a, 2021). The maximum resolution that was achieved was Q = 76.511° (0.83 Å).
The diffraction pattern was indexed The total number of runs and images was based on the strategy calculation from the program CrysAlisPro (Rigaku, V1.171.41.99a, 2021) and the unit cell was refined using CrysAlisPro (Rigaku, V1.171.41.99a, 2021) on 18089 reflections, 156% of the observed reflections.
The structure was solved and the space group P21/n (# 14) determined by the ShelXT (Sheldrick, 2015) structure solution program using Intrinsic Phasing methods and refined by full matrix least squares on F 2 using version 2018/3 of ShelXL (Sheldrick, 2015). All nonhydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model. Hydrogen atom positions were calculated geometrically and refined using the riding model.

_refine_special_details:
The structure was modelled as a non-merohedral twin with twin law [-1 0 0 / 0 -1 0 / 0 0 1] (BASF = 0.3212(12)). A disordered toluene solvent molecule using the FragmentDB function of Olex2 and fitting the idealised molecules onto peaks observed in a difference map. RIGU and SADI restraints were used to control the toluene. The value of Z' is 0.5. This means that only half of the formula unit is present in the asymmetric unit, with the other half consisting of symmetry equivalent atoms.

Dialuminacyclobutane 4 Crystal Data and Experimental
Experimental. Single colourless slab-shaped crystals of 4 were obtained by recrystallisation from a concentrated pentane solution at room temperature. A suitable crystal 0.60×0.30×0.20 mm 3 was selected and mounted on a suitable support on an Bruker APEX-II CCD diffractometer. The crystal was kept at a steady T = 100(2) K during data collection. The structure was solved with the ShelXT 2018/2 (Sheldrick, 2018) structure solution program using the intrinsic phasing methods solution method and by using Olex2  A colourless slab-shaped crystal with dimensions 0.60×0.30×0.20 mm 3 was mounted on a suitable support. Data were collected using an Bruker APEX-II CCD diffractometer operating at T = 100(2) K.
Data were measured using f and w scans using MoKa radiation. The maximum resolution that was achieved was Q = 30.552° (0.83 Å).
The diffraction pattern was indexed and the unit cell was refined using SAINT (Bruker, V8.40A, after 2013) on 9905 reflections, 6% of the observed reflections. There is a single molecule in the asymmetric unit, which is represented by the reported sum formula. In other words: Z is 4 and Z' is 1.

Dialuminacyclobutene 5 Crystal Data and Experimental
Experimental. Single yellow block-shaped crystals of 5 were obtained by recrystallisation from pentane at -30 °C. A suitable crystal 0.34×0.22×0.18 mm 3 was selected and mounted on a suitable support on an Xcalibur, Eos diffractometer. The crystal was kept at a steady T = 120.01(10) K during data collection. The structure was solved with the ShelXT (Sheldrick, 2015) structure solution program using the Intrinsic Phasing methods solution method and by using Olex2  There is a single molecule in the asymmetric unit, which is represented by the reported sum formula. In other words: Z is 4 and Z' is 1.

_refine_special_details:
The structure was modelled as a pseudomerohedral twin with (BASF = 0.3212 (12)). Two regions of disordered solvent were identified and modelled. One consists of two toluene molecules alternating position over a symmetry element, which were refined using an isotropic model. The second was modelled as a disordered mix of pentane (64%) and toluene (36%). In all cases the disordered molecules were modelled using the FragmentDB function of Olex2 and fitting the idealised molecules onto peaks observed in a difference map.

Aluminacyclopropene 6
Crystal Data and Experimental _refine_special_details: Twin law [-1 0 0 / 0 -1 0 / 1 0 1] was used to account for the metric symmetry mimicking a hexagonal setting. The refined twin scale factor (0.0234(2)) is small but makes a visible difference to the displacement ellipsoids and the refinement residuals.
Positional disorder in the model was identified from peaks in a difference Fourier map and refined with appropriate geometric and displacement ellipsoid restraints.