Trapping and Reactivity of a Molecular Aluminium Oxide Ion

: Aluminium oxides constitute an important class of inorganic compound that are widely exploited in the chemical industry as catalysts and catalyst supports. Due to the tendency for such systems to aggregate via Al-O-Al bridges, the synthesis of well-defined, soluble, molecular models for these materials is challenging. Here we show that reactions of the potassium aluminyl complex K 2 [( NON )Al] 2 ( NON = 4,5-bis(2,6-diiso-propylanilido)-2,7-di-tert-butyl-9,9-dimethylxanthene) with CO 2 , PhNCO and N 2 O all proceed via a common aluminium oxide intermediate. This highly reactive species can be trapped by coordination of a THF molecule as the anionic oxide complex [( NON )AlO(THF)] - , which features discrete Al-O bonds and dimerizes in the solid state via weak O … K interactions. This species reacts with a range of small molecules including N 2 O to give a hyponitrite ([N 2 O 2 ] 2- ) complex and H 2 , the latter offering an unequivocal example of heterolytic E-H bond cleavage across a main group M-O bond. [23] Investigation of the broader scope of E-H activation by 5 and related systems will be reported in due course.

Aluminium and oxygen are both among the most abundant elements in the Earth's crust, together forming over 50% of its elemental composition. [1] As such, bonding between the two elements occurs naturally on a huge scale: many minerals, ores and gemstones contain aluminium oxygen bonds. [2] Technologically, aluminium oxides and the related alumoxanes (e.g. methylalumoxane, MAO, [MeAlO]n), have attracted significant interest due to their ability to act as catalysts and/or co-catalysts in a range of reactions, including the polymerisation of aldehydes, [3] epoxides [4] and olefins. [5] Binary aluminium oxides are also active catalysts in a number of industrial transformations, including the Claus process, which converts H2S into elemental sulphur, [6] while alumina is widely exploited as a heterogeneous catalyst support.
Synthesising well-defined molecular, hydrocarbon soluble models of these compounds is therefore of considerable interest, as these systems may provide insight into the patterns of reactivity associated with Al-O bonds. However, the synthesis of systems of this type is challenging: aluminium-oxygen linkages are among the strongest element-element single bonds (ca. 502-585 kJ mol -1 ), [7] and as such, molecular aluminium oxide units have a strong thermodynamic incentive to oligomerise to form insoluble materials. [8] By employing sterically demanding ancillary ligands, however, a number of molecular complexes featuring bridging Al-O-Al motifs have been synthesised. Notable examples, shown in Figure 1, include { t Bu2Al(py)}2(µ-O) (I, py = pyridine), [9] {(Nacnac)Al}2(MeAl)(µ-O)3 (II, Nacnac = [(DippNCMe)2CH], Dipp = 2,6-diisopropylphenyl) [10] and [(Mes*)Al(µ-O)]4, (III, Mes* = 2,4,6-t Bu3C6H2). [11] In addition, Roesky and co-workers have reported that the mono-alumoxine IV can be trapped by coordination of B(C6F5)3 to the aluminiumbound oxygen atom. [12]  Reports of molecular aluminium oxide ions in which the oxide bears a formal negative charge (Al-O -) are even rarer still, presumably because of the stronger electrostatic drive to form Al-O-M bridges. Only one complex of this type has been crystallographically characterised, [(Nacnac)Al(Me)OLi]3 (V, Figure 1), which exists as a tightly-bound trimer in the solid state with each oxide bonded to two lithium cations; [13] no reactivity studies of this complex have been reported to date.
Our study was driven by a desire to investigate the reactivity of the strongly nucleophilic (and reducing) aluminyl system 1 towards small unsaturated molecules. Accordingly, a benzene solution of 1 was exposed to one atmosphere of CO2 at room temperature, resulting in a rapid colour change from yellow/orange to colourless and the formation of a single new product. This species was identified as the aluminium carbonate complex K2[(NON)Al(CO3)]2 2 on the basis of spectroscopic, analytical and crystallographic measurements (Scheme 1 and ESI). A similar reaction occurs between 1 and two equivalents (or excess) of phenyl isocyanate, yielding the isoelectronic aluminium carbamate complex K2[(NON)Al(O2CNPh)]2 3, the solid-state structure of which was also confirmed crystallographically (see ESI). The reactions of 1 with CO2 and phenyl isocyanate both require two equivalents of the substrate, and by analogy with related (isoelectronic) silylene systems, [16] are hypothesized to occur via initial reduction of the substrate, to give CO/PhNC and a common highly reactive aluminium oxide intermediate of the general formula [K{(NON)AlO}]n. Experimental evidence for this proposal comes from the 1 H NMR spectrum of the reaction mixture that yields 3, which confirms the presence of one equivalent of phenyl isocyanide. In addition, an in situ 13 C{ 1 H} NMR spectrum of the reaction mixture generating 2 also confirms the presence of CO.
From this aluminium oxide intermediate, the reaction proceeds via assimilation of a second equivalent of the substrate to give the isolated products 2 and 3. Similar mechanisms have been reported for other low-oxidation state main group complexes with CO2, such as [(Nacnac)Mg]2 reported by Jones [17] and the acyclic silyene [(NHI)Si-Si(SiMe3)3] (NHI = Nheterocyclic iminato ligand) reported by Inoue. [16g] In an attempt to observe this oxide intermediate, the reactions between 1 and CO2/PhNCO were monitored by variable temperature (VT) 1 H NMR in d8-toluene (-80 to 25 °C). However, even at -80 °C, both reactions were complete in seconds, and no evidence of any intermediate species could be obtained from these measurements. Furthermore, the reaction of 1 with one equivalent of PhNCO (per aluminium centre) gives a 1:1 mixture of 3 and unreacted starting material 1, suggesting that the oxide intermediate is even more reactive towards the substrate than potassium aluminyl complex 1.
With this in mind, a more direct route to the oxide was sought: a benzene solution of 1 was exposed to N2O (1 atm.) at room temperature. In situ 1 H NMR monitoring reveals the clean formation of a single new product, which can be identified by Xray crystallography not as the target oxide, but remarkably as the cis-hyponitrite complex K2[(NON)Al(N2O2)]2 (4) (Scheme 1). In a similar fashion to the reactions to give 2 and 3, it appears that 1 reacts with two equivalents of N2O to give 4, presumably also via the same highly reactive aluminium oxide intermediate. The formation of the cis-hyponitrite ligand has been previously reported in d-block chemistry, via the radical coupling of two metal-bound NO ligands. [18] Metal cis-hyponitrite complexes have also been proposed to be transient intermediates in catalytic NO reductions, [19] however the synthesis of the ligand directly from N2O is to the best of our knowledge, unprecedented. Structurally, 4 is dimeric in the solid state (Figure 2  atmosphere for several weeks, however in solution (THF or benzene) it decomposes over the course of 12 h at room temperature. Over a shorter time period, X-ray quality crystals of 5 could be grown from a concentrated benzene solution. As with compounds 2-4, 5 exists as a dimer in the solid state (Figure 2), held together by potassium counter-ions which engage in a combination of interactions with the flanking aryl rings and with the aluminium-bound oxide ligands. The two (non symmetryrelated) Al-O bond lengths are identical within error (1.6772 (12) and 1.6754(12) Å) and are among the shortest Al-O bond distances reported to date, [20] being considerably shorter than those found in the lithium aluminium oxide trimer [(Nacnac)Al(Me)OLi]3 (V; 1.698(1) Å). [13] This To investigate the lability of the coordinated THF in 5, exchange with d8-THF was monitored in solution by VT 1 H NMR. At temperatures below -10 °C, the coordinated THF in 5 does not appear to exchange with d8-THF (as signalled by the unchanging intensity of the signals associated with coordinated protio-THF). However, at 0 °C and above, scrambling of coordinated THF with d8-THF is observed; distinct 1 H resonances are observed at 0 °C for coordinated (δH = 3.84 ppm for the OCH2 protons) and 'free' protio-THF (δH = 3.62 ppm), while at 10 °C only the signal at 3.62 ppm is seen. This suggests that at temperatures above 0 °C, the coordinated THF molecule is chemically labile. Accordingly, while 5 does not react with N2O below -10 °C (in THF), it is rapidly converted to the cishyponitrite complex 4 above this temperature. In addition, 5 also reacts rapidly with CO2 and PhNCO at room temperature to give 2 and 3, respectively (Scheme 3). A defining feature of transition metal complexes bearing highly polarised bonds is their ability to heterolytically cleave kinetically inert E-H bonds under mild conditions; d-block metal imide systems, for example, have been shown to cleave H-H and C-H bonds. [21,22] As the extent of the polarisation of the Al-O bond in 5 appears to be comparable to that of M=N bonds in early transition metal imides, we were interested to investigate the reactivity of 5 towards H2. Accordingly, a solution of 5 in benzene reacts with H2 at room temperature and pressure over the course of 16 h to give the aluminium hydride hydroxide complex K2[(NON)Al(H)OH]2 6 in high (87%) isolated yield (Scheme 3). This chemistry represents the first structurally authenticated example of H2 activation across an aluminium oxygen bond.
The inability of 5 (or its more reactive THF-free form) to from Al-O-Al bridges presumably leads to significant unquenched Lewis basic character (at O), which in conjunction with the Lewis acidic aluminium centre offers some parallels with frustrated Lewis pair (FLP) chemistry in the activation of H2. [23] Investigation of the broader scope of E-H activation by 5 and related systems will be reported in due course.