A nucleophilic gold complex

Solid-state auride salts featuring the negatively charged Au– ion are known to be stable in the presence of alkali metal counterions. While such electron-rich species might be expected to be nucleophilic (in the same manner as I–, for example), their instability in solution means that this has not been verified experimentally. Here we report a two-coordinate gold complex (NON)AlAuPtBu3 (where NON is the chelating tridentate ligand 4,5-bis(2,6-diisopropylanilido)-2,7-di-tert-butyl-9,9-dimethylxanthene) that features a strongly polarized bond, Auδ––Alδ+. This is synthesized by reaction of the potassium aluminyl compound [K{Al(NON)}]2 with tBu3PAuI. Computational studies of the complex, including quantum theory of atoms in molecules charge analysis, imply a charge at gold (−0.82) that is in line with the relative electronegativities of the two metals (Au: 2.54; Al: 1.61 on the Pauling scale). Consistently, the complex is found to act as a nucleophilic source of gold, reacting with diisopropylcarbodiimide and CO2 to give the Au–C bonded insertion products (NON)Al(X2C)AuPtBu3 (X = NiPr, 4; X = O, 5). A two-coordinate monovalent gold complex that features a highly polarized aluminium–gold covalent bond, Alδ+–Auδ−, has been synthesized using a very strongly electron-donating aluminyl ligand. In solution, the complex reacts as a nucleophilic source of gold towards heteroallenes such as carbodiimides and CO2.


<figure 1 near here>
Over the last decade, boryl anions have been utilized as highly σ-donating ligands in conjunction with numerous metals from throughout the Periodic Table 22 Figure 1) 25 . Even though boryl ligands are known to be highly electron donating, and the difference in electronegativity between gold and boron (2.54 and 2.04, respectively) 26 implies a Au(δ-)-B(δ+) polarity to these bonds, no goldcentered nucleophilic reactivity has been reported for these systems. In the current study we make use of an extremely strongly donating aluminyl ligand to synthesise a two-coordinate molecular species featuring a highly polarized Au(δ-)-Al(δ+) bond, that shows that gold can act as a nucleophile in the solution phase.

Syntheses
In recent work, we reported the preparation and characterization of the potassium aluminyl [K{Al(NON)}] 2 (1, where NON = 4,5-bis(2,6-diisopropylanilido)-2,7-di-tert-butyl-9, 9-dimethylxanthene) and demonstrated that this complex reacts as an aluminium-centred nucleophile 27 . Reactions with (NON)AlI and ( Mes Nacnac)MgI(OEt 2 ) (where Mes Nacnac = (NMesCMe) 2 CH, and Mes = 2,4,6-Me 3 C 6 H 2 ) lead to the formation of unsupported metal-metal bonds in [Al(NON)] 2 and (NON)AlMg( Mes Nacnac), respectively. We were interested in expanding this chemistry to the d-block metals, in particular gold, as systems of the type R 3 PAuX are well known catalysts/precatalysts in numerous organic reactions involving alkynes in which gold acts exclusively as an electrophile 28 . We hypothesized that by incorporating the extremely electron donating aluminyl fragment into such a system, the mode of reactivity could be reversed, i.e. making gold nucleophilic. Accordingly, 1 was found to react with two phosphine-ligated gold(I) iodide complexes, Ph 3 PAuI and t Bu 3 PAuI, to give However this procedure simply led to a 1:1 mixture of 2 and unreacted aluminyl 1. By contrast, the addition of one equivalent of the bulkier system t Bu 3 PAuI to a toluene solution of 1 resulted in a clean reaction to give the two-coordinate gold complex 3 in high crystalline yield ( Figure 2).
Presumably, the bulkier tritertbutylphosphine ligand prevents 3 from reacting with a further equivalent of t Bu 3 PAuI at room temperature, even when excess t Bu 3 PAuI is added. <Figure 2 near here> Taking the relative electronegativities of gold and aluminium into account, both of the Al-Au bond forming reactions could be regarded as bringing about a formal two-electron oxidation of the aluminium centre from +1 to +3. In the reaction to give 2, this would be accompanied by one-electron reduction of each of the two gold centres. The resulting Al(III)Au(0) 2 formalism is consistent with structural metrics -notably the Au-Au distance (vide infra). In the formation of 3 however, a similar redox formalism would necessitate two-electron reduction of the single gold centre from +1 to -1. With a view to providing a more rigorous interpretation of the Al-Au bonding in 3, and the transfer of charge accompanying bond formation, NOCV calculations were carried out. 29

Charge analysis of 2 and 3
In order to study the charge distribution in the two Al-Au bonded complexes, the effective atomic charges of all atoms in 2 and 3 were calculated using QTAIM (see Supplementary Information section 5 for further details) 37,38 . In 2, the total charge on the Al(NON) fragment is

Reactivity of 3
Given that the calculated charge on gold in compound 3 is close to -1, its possible reactivity as a gold-centred nucleophile was investigated. Accordingly, 3 was reacted with a range of unsaturated carbon-centred electrophiles. Reactions with simple aldehydes and ketones were investigated, targeting complexes of the type (NON)AlO(R) 2 CAuP t Bu 3 . However, these reactions lead to complex mixtures, presumably due to the fact that the initially-formed products contain a highly Lewis acidic aluminium centre, which reacts further with the carbonyl substrates. With this in mind, heteroallenes, such as carbodiimides and CO 2 were targeted, on the basis that the product might chelate the aluminium centre and quench its Lewis acidity.
Accordingly, one equivalent of diisopropylcarbodiimide was added to a toluene solution of 3 at room temperature, leading to clean formation of the insertion product

Data availability statement
Crystallographic data for the structures reported in this article have been deposited at the