Synthesis of a low-valent Al4+ cluster cation salt

Low-valent aluminium compounds are very reactive main-group species and have therefore been widely investigated. Since the isolation of a stable molecular Al(I) compound in 1991, [(AlCp*)4] (Cp* = [C5Me5]–), a variety of highly reactive neutral or anionic low-valent aluminium complexes have been developed. By contrast, their cationic counterparts have remained difficult to access. Here, we report the synthesis of [Al(AlCp*)3]+[Al(ORF)4]– (RF = C(CF3)3) through a simple metathesis reaction between [(AlCp*)4] and Li[Al(ORF)4]. Unexpectedly, the [Al(AlCp*)3]+ salt forms a dimer in the solid state and concentrated solutions. Addition of Lewis bases results in monomerization and coordination to the unique formal Al+ atom, giving [(L)xAl(AlCp*)3]+ salts where L is hexaphenylcarbodiphosphorane (x = 1), tetramethylethylenediamine (x = 1) or 4-dimethylaminopyridine (x = 3). The Al+–AlCp* bonds in the resulting [(L)xAl(AlCp*)3]+ cluster cations can be finely tuned between very strong (with no ligand L) to very weak and approaching isolated [Al(L)3]+ ions (when L is dimethylaminopyridine). Although neutral and anionic low-valent aluminium complexes are widespread, their cationic counterparts have remained rare. Now, a salt of [Al(AlCp*)3]+ featuring a formal low-valent Al+ cation has been isolated that dimerizes in concentrated solutions and the solid state, and also forms Al4 clusters on coordinating with Lewis bases.

One example of a cationic low-valent aluminium compound has been reported-the salt [Al 5 Br 6 ·(thf) 6 ] + [Al 5 Br 8 ·(thf) 4 ] − (VII; thf = tetrahydrofuran) 35,36 . However, VII formed by serendipity in a poorly understood reaction from a starting material that is only available from a very specialized apparatus under specific conditions, and therefore cannot be prepared with a rational direct synthesis. Yet, cationic low-valent aluminium compounds are attractive owing to a potential transition-metal-like ambiphilic reactivity 37 . Examples for a similar ambiphilicity have been reported for cationic low-valent gallium and indium compounds [38][39][40][41] , whose chemistry is more developed. Such cationic low-valent group 13 complexes profit from a more pronounced electrophilicity compared to their neutral analogues. Moreover, the stabilization of the lower oxidation state induced by a positive charge is a promising strategy to control reductive eliminations at low-valent main-group elements, with the first examples being reported for cationic complexes of Ga and Ge (refs. [42][43][44]. Here, we present the synthesis and full characterization of the cationic, low-valent aluminium complex salt [Al(AlCp*) 3 4 ]with R F = C(CF 3 ) 3 ), accessible via a simple metathesis procedure, and initial investigations on its reactions with strong donor ligands.
Characterization. Single-crystal X-ray diffraction analysis revealed a trigonal pyramidal geometry at the unique aluminium atom (Fig.  2 47 ), which he defined as an ideal model for a 2e2c metal-metal bond. Yet, in the second unit, the Al 4 pyramids are tilted against one another, resulting in a longer Al−Al distance of 3.005(2) Å. Related trans-bent structures are observed for main-group donor-acceptor dimers 3 . With both geometries observed in the same crystal, the bonding modes between the two 'naked' aluminium atoms need to be energetically comparable.
Solid-state UV-visible spectroscopy revealed the gap between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of the dark-purple crystalline solid 1 including the dimer dications as 1.74-1.92 eV (645-725 nm), depending on the method of determination . Intriguingly, solutions of 1 in 1,2-DFB and 1,3-DFB show distinct colour changes from yellow to deep purple upon concentration or cooling of the reaction mixture. In the UV-visible spectrum of 1, the colour change is accompanied by the appearance of a broad UV-visible absorption band at wavelength λ = 564 nm (Fig. 2f). The experimental spectrum of the purple solution is in line with the time-dependent density functional theory (DFT) computed spectrum of the dimer with the short Al-Al bond as observed in the molecular structure of 1 (calculated band position λ calc. = 581 nm, using the bp86 functional with d3bj dispersion correction and the def2-svp basis set (bp86-d3bj/def2-svp); Supplementary Fig. 62). By contrast, the computation of the UV-visible spectrum of the monomer does not include a band above 500 nm. Hence, an equilibrium between a purple [Al 8 (Cp*) 6 ] 2+ [pf] 2 − dimer and its corresponding yellow monomer is postulated, with the dimer being favoured at lower temperatures and higher concentrations.
On the basis of the experimental data, an excess of the monomer at standard temperature and pressure conditions is suggested to yield an equilibrium constant for the dimerization in Comparison with the DFT-computed NMR shifts (δ calc. ) as well as those of known AlCp* clusters 9 (δ exp. ) shows that the resonance at δ exp. = -40 ppm (δ calc. = -44 ppm) belongs to the aluminium atoms of the AlCp* moieties, and the −275 ppm signal to the unique aluminium atom (calculated NMR chemical shifts for the monomer δ calc.,monomer = -329 ppm and the dimer δ calc.,dimer = -300 ppm). Similarly high-field shifted 27 Al NMR resonances have been observed only for the large metalloid clusters Al 50 Cp* 12 (δ = -272 ppm) 13 and SiAl 14 Cp* 6 (δ = -273 ppm) 48 . While no decomposition of solid 1 stored in a glove box at room temperature was observed after months, the concentrated, violet solution of 1 in 1,3-DFB (125 mg ml -1 ) turns yellow at room temperature after 12 h accompanied by the formation of metallic aluminium. In the 27 Al NMR spectrum of the decomposition products, the resonance of literature-known 49 4 ] as a by-product (Fig. 3a). Notably, less concentrated solutions of 1 (50 mg ml -1 ) took five days before a fading of the purple solution to yellow accompanied with formation of elemental aluminium was completed. In addition, [(AlCp*) 4 ] was absent in the 27 Al NMR spectrum of the decomposition products (Supplementary  51 ) are readily accessible. By contrast, the formation of the respective Al complex could not be achieved to date. Rather, compound 2 can be regarded as an inverse-sandwich complex, in which the lone pairs at the reactive low-valent aluminium atoms are masked by reformation of the tetrameric cluster. Overall, the dimer dication is sensitive towards disproportionation into Al(0) and  (Fig. 4b).
Hence, the covalent bonding between the Al Cp* atoms is reconstituted in the complexes. Notably, the Al−Al bond length-change upon complexation is even more pronounced in the complex Here, the average Al L -Al Cp* bond length of 2.802(1) Å is even larger than the average Al Cp* -Al Cp* bond length of 2.670(1) Å. Hence, with coordination of dmap as a strong donor ligand, an inversion of the relative Al-Al bonds in the non-symmetric Al 4 complexes compared to 1 + could be achieved, and apparently 5 + is already very close to an isolated [:Al(L) 3 ] + complex ion. These structural changes of the cationic Al 4 clusters are also reflected in the 27 Al NMR spectra.
For 3 + and 4 + , the resonances of the aluminium atoms of the AlCp* moieties are shifted high-field compared to 1, that is, shifted to δ exp. = −65 ppm (δ calc. = -66 ppm) and to δ exp. = -59 ppm (δ calc. = -59 ppm), respectively. However, only for the tmeda complex, an NMR signal of the unique Al atom can be observed at δ exp. = 48 ppm (δ calc. = 9 ppm). Upon dissolution of 5 in 1,2-DFB, the orange solution shows an NMR signal at δ exp. = −83 ppm, fitting to the calculated value of δ calc. = -86 ppm. Yet, already after a few minutes, a 27 Al NMR resonance at δ = -79 ppm, indicative of the formation of free [(AlCp*) 4 ], was observed. After two weeks at room temperature, the colour of the solution changed to yellow and only the 27 Al NMR resonance of [(AlCp*) 4 ] was detected.
Al (1) Al (2) Al (3) Al ( 1). a, In concentrated solutions, 1 is present mostly as a dimer and undergoes a rapid and clean disproportionation into Al(0) and Al(III) (top). By contrast, in diluted solutions 1 is present mostly as a monomer, and its slow decomposition yields the dimer 2 at room temperature (r.t.; bottom). b, Molecular structure of 2 isolated from a less concentrated solution of 1 in 1,3-DFB after five days. Al atoms shown in blue; hydrogen atoms and [pf]anion omitted for clarity. Owing to a super-structure and extensive disorder, the data were not sufficient to allow for discussion of bond lengths and connectivity in 2 + (Supplementary Section 2.4 for discussion). thermal displacement of the ellipsoids was set at 50% probability. [Al(AlCp*) 3 ] + unit, the 3s 2 -like lone-pair orbital at the formally cationic aluminium atom resides as HOMO-2 (that is, the third highest occupied molecular orbital) at an energy of -8.55 eV (Fig. 5a). The HOMO and HOMO-1 (-8.50 eV) are energetically very close, are degenerate and present an interaction of the lone-pair orbitals of the AlCp* moieties with the 3p x and 3p y orbitals at the unique Al atom (Supplementary Fig. 92). The LUMO (-4.32 eV) of 1 + has strong 3p z character, whereas LUMO+1 and LUMO+2 are dominated by the 3p x and 3p y orbitals of the unique Al + atom. Upon dimerization to the dication [Al 2 (AlCp*) 6 ] 2+ , interactions of the HOMO-2 with the LUMO between two cations result in a weak σ-bonding interaction as indicated by the small calculated Al + -Al + Wiberg bond index of 0.64 as well as by low values for electron density (ρ(r) = 0.25 electrons Å -3 ) and its ellipticity (ε = 0.01) at the Al + -Al + bond critical point from quantum theory of atoms in molecules analysis (QTAIM analysis; Supplementary Fig. 62).
This result is further supported by an energy decomposition analysis combined with natural orbitals for chemical valence (EDA-NOCV; Fig. 5c) as promoted by Frenking and coworkers in recent years 57 . The plots of the deformation density values Δρ (1) and Δρ (2) in Fig. 5c (associated with contributing energy terms ΔE Orb(1) / ΔE Orb(2) ) represent the major orbital interaction energy (E Orb ) terms (69% of ΔE Orb ) and display a charge flow from the HOMO−2 to the LUMO of the fragments (Fig. 5a). Here, Δρ (1) shows the predominant σ-bond between the unique Al + atoms, whereas Δρ (2) displays a minor σ*-bonding component. Yet, the anti-bonding character of the latter is diminished by admixture of the LUMO (p x at Al + ), resulting in a strengthening of the interaction in the respective Al 4 clusters (more detail is in Supplementary Section 3.2). Although the Al + -Al + bonding interaction between the [Al(AlCp*) 3 ] + fragments is very labile, it is sufficient to overcome the Coulomb repulsion between the cationic fragments. Furthermore, the dimerization demonstrates the amphiphilicity of the low-valent aluminium cation.
the 3p x and 3p y orbitals at the unique aluminium atom (64% in total, ΔE Orb(1) + ΔE Orb (2) ). The minor interactions represent synergistic σ-donation and σ-back-donation between the Al + and (AlCp*) 3 unit (ΔE Orb(3) + ΔE Orb(4) = −48.57 kcal mol -1 (28%); Supplementary Fig. 84). Hence and in agreement with our deduction from the structure and chemistry, monomeric 1 + is best described as a cationic aluminium atom, which is coordinated by three strongly electron-donating AlCp* substituents. This assignment agrees with the absence of bond critical points between the Al Cp* atoms in the QTAIM analysis of 1 + (Supplementary Fig. 78). Yet, the situation changes upon complexation of the unique aluminium atom. Here, the orbital interaction terms associated with the delocalization of the lone pair of the unique Al atom into the (AlCp*) 3 fragment increase notably in the order 3 + < 4 + < 5 + , while the energy terms associated with the electron donation from the (AlCp*) 3 unit into the 3p orbitals of the [(L) x Al + ] moiety ((L) x = (tmeda) 1 , (cdp) 1 , (dmap) 3 )) decrease in the same sequence ( Supplementary Figs. 86-91). In 5 + , the energy term of the delocalization of the lone-pair orbital at (dmap) 3 Al + into the (AlCp*) 3 fragment even becomes the dominant contribution to the total orbital interaction energies (ΔE Orb(1) = −61.02 kcal mol -1 or 52%; Fig. 5d). The π-bonding interactions combine to ΔE Orb(2) + ΔE Orb(3) = −37.90 kcal mol -1 (32%; Supplementary Fig. 90). Consequently, the role of the AlCp* units can be altered decisively upon complexation of the unique Al atom in 1: Whereas the AlCp* units mainly act as Lewis bases in 1 + , the AlCp* units become increasingly Lewis acidic in the isolated clusters and finally lead to the observed shortening and inversion of the Al-Al bond lengths in the molecular structure of 5 + .
These results are supported by the computed QTAIM charges at the unique Al atoms. For 1 + , the QTAIM charge at the unique Al atom of +0.58 reflects the charge delocalization onto the strongly Lewis basic AlCp* units. In comparison to the QTAIM analysis of [(AlCp*) 4 ] (q Cp* = -0.84, q Al = +0.84), the negative partial charges at the Cp* substituents decrease in 1 + (q Cp* = -0.65, q AlCp* = +0.79). Nevertheless, a substantial positive charge remains at the unique Al atom. Although for 3 + and 4 + only a marginal increase of the QTAIM charges to 0.63 and 0.68 compared to 1 + is calculated, the dmap-ligated Al atom in 5 + bears a highly positive charge of q Al = 1.17 and can be denoted as a cationic, low-valent aluminium atom. Consequently, 5 + may act as a source of a cationic, monomeric Al + and will be studied in future research.

Conclusion
In summary, we report the synthesis of a cationic, low-valent aluminium compound [Al(AlCp*) 3 ] + [pf] -(1) accessible on the gram scale via a simple metathesis route. Indicating an ambiphilic reactivity, crystallographic as well as UV-visible spectrometric and computational studies reveal the dimeric structure [Al 2 (AlCp*) 6 ] 2+ ([pf] -) 2 of 1 in the solid state as well as in solution at high concentration and low temperature. At lower concentrations and higher temperatures, the monomeric structure prevails. Addition of Lewis bases to solutions of 1 uses the monomeric cation as acceptor and allows for the synthesis of cluster cations of type [(L) x Al(AlCp*) 3 ] + [pf] -((L) x = (tmeda) 1 , (cdp) 1 , (dmap) 3 ). Their solid-state structures, the NMR shifts and computational studies demonstrate that the dominating bonding interactions can be fine-tuned and even inverted by the donor strength of the added ligands, as maximized in the highly non-symmetric dmap-substituted cluster. Consequently, this salt may potentially be used as a synthon for an [:Al(L) 3 ] + salt, which due to its cationic nature potentially may be able to conduct reversible oxidative addition and reductive elimination chemistry with small molecules.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41557-022-01000-4. methods General methods and instrumentation. All manipulations of air-and/or water-sensitive compounds were performed under an inert argon atmosphere using standard Schlenk or glove box techniques. All glassware used in reactions was stored overnight in an oven at 180 °C and thoroughly flame dried prior to usage. Cp*H (Sigma-Aldrich), potassium hexamethyldisalazide (Sigma-Aldrich, >95%), dmap (Sigma-Aldrich, >98%) and tmeda (Sigma-Aldrich, >99%) were used as received. Li[pf] (ref. 58 ), cdp (refs. 59,60 ) and [(AlCp*) 4 ] (ref. 45 ) were prepared using procedures known in the literature. The 1,3-DFB and 1,2-DFB (Fluorochem) were refluxed over CaH 2 , fractionally distilled and stored over activated 3 Å molecular sieves. The n-pentane was collected from the solvent-purification system and used as received. NMR samples were prepared inside an inert atmosphere glove box in NMR tubes equipped with a gas-tight J. Young valve. The 1 H, 7 Li, 13 C, 19 F, 27 Al and 31 P NMR spectra were acquired either on a Bruker Biospin Avance II 400 MHz WB or a Bruker Avance III HD 300 MHz spectrometer. The 1 H and 13 C NMR spectra are reported relative to tetramethylsilane and were calibrated to residual solvent resonances. Underlined atoms represent the atoms/groups causing the respective NMR signal. A small "s" corresponds to singlet, "m c " to a centered multiplet. Data analysis was performed using Bruker TOPSPIN 3.5 software. The broad resonance at δ = 70 ppm observed in 27 Al NMR spectra corresponds to a background from Al nuclei in the probe head. The mass-spectrometric experiments were performed with a Thermo-Fischer LTQ XL linear ion-trap mass spectrometer equipped with an electrospray-ionization source (Capillary temperature T cap = 100 °C; source voltage, 3.5 kV; tube lens voltage, 95 V). Mass spectra were obtained by electrospray ionization from a millimolar solution in 1,2-DFB. In the CID (charged-induced dissociation) experiments, helium served as the collision gas. The supplied collision energy was adjusted by varying the normalized collision energy between 0 and 25. Fourier transform infrared spectra (FTIR) were recorded inside a glove box with a Bruker ALPHA equipped with QuickSnap Eco-ATR (ATR, attenuated total reflectance) module and ZnSe crystal. The spectra were measured at room temperature in the range of 4,000-550 cm -1 with 64 scans and a resolution of 2 cm -1 . The data were processed with the Bruker OPUS 7.5 software package and, if not stated otherwise, a baseline correction with three iterations was performed. Fourier transform Raman spectra were recorded with a VERTEX 70 with Bruker RAM II Modul (1,064 nm exciting line of a Nd-YAG laser) and liquid-nitrogen-cooled Ge detector. The samples were flame-sealed in soda-lime glass Pasteur pipettes and were measured at room temperature in the range of 4,000-80 cm -1 with up to 10,000 scans and a resolution of 4 cm -1 . The data were processed with the Bruker OPUS 7.5 software package and, if not stated otherwise, a baseline correction with five iterations was performed. All infrared and Raman spectra were normalized to 1, and the intensities are reported as follows: ≥0.8 = very strong (vs), ≥0.6 = strong (s), ≥0.4 = medium (m), ≥0.2 = weak (w), <0.2 = very weak (vw). The graphical representations were created with OriginPro 2021. Solution UV-visible spectra were recorded on a Varian Cary 50 UV-visible spectrophotometer in quartz cuvettes (thickness, 1 mm) in 1,2-DFB solution. Solid-state UV-visible spectra were recorded on a Thermo Scientific Evolution 600 UV-visible spectrophotometer. Single-crystal X-ray diffraction data were collected using either a Bruker SMART APEXII QUAZAR detector with fixed-Chi D8 goniometer and Incoatec Mo microsource or a Bruker D8 VENTURE with PHOTONIII detector, fixed-Chi D8 goniometer and Incoatec Mo/Cu microsource. Crystals were selected under perfluoropolyether oil, mounted on 0.1-to 0.3-mm-diameter CryoLoops and quench-cooled using an Oxford Cryostream 800 open flow N 2 cooling device. Data were collected at 100 K using monochromated Cu Kα or Mo Kα radiation (λ = 1.5418 or 0.71073 Å, respectively). Data processing was done with SHELXS/ XL and refined by least squares on weighted F 2 values for all reflections; disordering of fragments was done with the help of the implemented DSR tool (disordered structure refinement). Graphical representations were prepared using Olex2 v.1.2. Finalization of gathered data was done using the Finalcif tool. Powder diffractograms were recorded with the sample sealed with perfluoropolyalkylether oil (AB128330, abcr) in a 0.3-mm-thick capillary (Hilgenberg; wall thickness, 0.01 mm) at 100 K in the 2θ range 2.0-40.0° with a STOE STADI P powder diffractometer with Mo Kα1 radiation (λ = 0.709300 Å) equipped with a Ge(111) monochromator and Mythen 1 K detector. Data acquiring, data processing and the calculation of powder diffractograms from single-crystal data were performed using STOE WinXPOW package. (1.93 mg, 1.98 mmol, 0.25 equiv.) were weighed into a Schlenk flask. The 1,2-DFB (15 ml) was added at −40 °C to yield a dark-purple solution. The reaction mixture was stirred for 3 h at −40 °C and then warmed to room temperature. The solvent was evaporated under reduced pressure, and 1,3-DFB (12 ml) was added. The resulting reaction mixture was stirred for 15 min, and precipitation of a white solid was observed. The mixture was filter-canulated and the deep-purple solution was allowed to crystallize in a freezer at -40 °C. The title compound (967 mg, 0.653 mmol, 33%) was isolated as dark-purple crystals. The solution was shaken and after 10 s the solvent was evaporated in high vacuum. The obtained orange powder was washed with toluene (3 × 1 ml) and dried under vacuum. The title compound (28 mg, 15 μmol, 45%) was isolated as an orange powder. Since 5 decomposes in solution, bulk purity was verified by powder X-ray diffraction (see Supplementary Information) and vibrational spectroscopy. Crystals of 5 with a quality fit for single-crystal X-ray diffraction could be grown by layering a solution of 5 in 1,2-DFB with n-heptane at -30 °C.