The Aluminyl Anion: A New Generation of Aluminium Nucleophile

Supporting information for this article is given via a link at the end of the document. Abstract: Trivalent aluminium compounds are well known for their reactivity as Lewis acids/electrophiles, a feature that is exploited in many pharmaceutical, industrial and laboratory-based reactions. Recently, a series of isolable aluminium(I) anions (‘aluminyls’) have been reported, which offer an alternative to this textbook description: these reagents behave as aluminium nucleophiles. This minireview covers the synthesis, structure and reactivity of aluminyl species reported to date, together with their associated metal complexes. The frontier orbitals of each of these species have been investigated using a common methodology to allow for a like-for-like comparison of their electronic structure and a means of rationalising (sometimes unprecedented) patterns of reactivity. scratches the surface for what is to come from this novel class of nucleophiles. The coordination chemistry of the anions is only just beginning and small molecule activation including N 2 and CO is also yet to be reported. The use of aluminyl anions in catalysis is also an area that we are currently investigating. The field is rapidly expanding and definitely ‘one to watch’ over the next few years. been reported that remarkably show nucleophilic behaviour at aluminium. This minireview compares and contrasts the synthesis, structure, reactivity and electronic properties of these aluminyl anions whilst proposing future directions for the field.


Introduction
Aluminium is the most abundant metal in the Earth's crust, accounting for >8% of its overall mass. 1 It is found in the sedimentary rock Bauxite, which is mined on a hundred million tonne scale annually. Bauxite is readily processed via the Bayer process to give alumina (Al2O3), the raw material for the production of aluminium metal. 2 In molecular chemistry, aluminium is most commonly found in the +3 oxidation state, with the +1 state which is favoured for the heavier Group 13 metals being strongly reducing. 3 Hence, trivalent species constitute the vast majority of aluminium compounds used in chemical synthesis, such as trialkylaluminiums (typically used as alkylating reagents) and aluminium trihalides (widely used as Lewis acids). 4 Low oxidation state aluminium compounds, i.e. those where the aluminium adopts the formal oxidation states +2, +1 or 0 are known, but are considerably less common. 5

Aluminium(I) Complexes
Predating the work on aluminyl systems are the neutral aluminium compounds in the formal +1 oxidation state. The isolation of Al(I) species is challenging thermodynamically, and such species are typically highly reactive and prone to disproportionation. 6 Binary Al(I) compounds such as AlH, AlX (X = halide) and Al2O have been reported but are only stable for prolonged periods in the gas phase at low pressures. 7 The development of cryochemical methods for the handling of metastable solutions of AlX (X = Cl, Br, I) proved to be instrumental in developing the chemistry of organo-aluminium(I) systems. 7c-f Accordingly, the first example of an isolable aluminium(I) complex -tetrameric (AlCp*)4 1 (Figure 1, Cp* = C5Me5) -was reported in 1991 by Schnöckel and co-workers, 8 from the reaction between a metastable solution of AlCl and MgCp*2. An alternative synthesis using potassium metal reduction of Cp*AlCl2 was later reported by Roesky and co-workers. 9 In the solid state, 1 is centred around an Al4 tetrahedron capped by four 5 -Cp* ligands. The bulky Cp* ligands provide considerable kinetic stabilization, and contribute to its enhanced thermal tolerance compared to metastable AlX systems. In solution at elevated temperatures, 1 can undergo dissociation into monomeric AlCp* units, a factor which has been exploited in a significant part of its subsequently reported reaction chemistry. In 2000, Roesky and co-workers reported the first isolable monomeric aluminium(I) system, 2 (Figure 1), utilising a bulky βdiketiminate ligand. 10 2 features a two-coordinate aluminium(I) centre and can be regarded as an aluminium analogue of an NHC, with both a lone pair and empty p-orbital localized on the metal centre. The reactivity of these two landmark Al(I) systems (1 and 2) is vast and diverse and has recently been comprehensively reviewed by Ma, Yang, Roesky and co-workers. 11 The chemistry of 2, in particular, is undergoing a recent renaissance, being a key reagent in a number of recent publications. 12 These recent reports include the activation of small molecules (CO, C2H4) 12 c,j , reduction chemistry (including that of benzene), 12b,e,I,k,m cycloadditions 12a,d,g and various bond activations including (C-H and C-F). 12 c,f,h,l,n The vast majority of the reactions of 1 and 2 are best described oxidative additions, although some evidence of nucleophilic behaviour has been reported (especially with electropositive metals/Lewis acids) and this has been exploited in the formation of Al-metal donor/acceptor bonds. 11 Since the initial reports of 1 and 2, various other neutral aluminium(I) systems have been reported using an array of bulky ligands. These include the remarkable NHC-stabilised dialumene reported by Inoue and coworkers, 13 the cyclic alkyl amino carbene (CAAC) supported Al(I) hydride complex 14 and the monomeric tBu3 CpAl(I) 15 reported by Braunschweig and co-workers and as well as a number of other Al(I) tetramers supported by bulky ancillary ligands. 3

Anionic Group 13 Centred Nucleophiles
Before our initial report of an anionic aluminium(I) complex (an 'aluminyl' anion) 16 in 2018, analogous anions of the other group 13 elements, including boron, gallium and indium were already known in the literature. The first of these to be reported was the gallyl anion, [( tBu DAB)Ga] -3 ( tBu DAB = [( t BuNCH)2] 2-) by Schmidbaur and co-workers over 20 years ago ( Figure 2). 17 This was followed by a diaryl substituted version reported by Jones and co-workers in 2002 18a and a BIAN stabilised version in 2008 by Fedushkin, Schumann and co-workers. 18b The reactivity of these gallyl anions have been thoroughly investigated, especially with regards to their coordination chemistry. 19 However, very little has been reported with respect to small molecule activation with these systems.  Figure 2). 20 Unlike the potassium gallyl 3, which was initially isolated as an ion separated pair, 4 shows a strong electrostatic interaction between the boron lone pair and the lithium cation. Boryl system 4 exists as a dimer in the solid state (with two DME molecules bridging the two Li + cations), however a monomeric version of this complex has since been reported. 21 To date, boryl anions featuring an array of chelating ligands have been reported, 22,23 all of which exhibit strong interactions between the boron and the cation (either covalent or electrostatic); no charge-separated boryl anion has yet been reported.
The reactivity of the boryl anions has become an area of considerable interest over the last decade. They have been shown to be highly reactive reagents, which is partly due to their relatively small HOMO-LUMO energy gap. This defining feature allows them to activate a wide array of small molecules and/or substrates with strong -and -bonds. In addition, their highly directional lone pair also makes them ideal ligands in coordination chemistry. The reactivity of boryl anions has recently been covered in a comprehensive review by Weber. 22 The heaviest member of the known group 13 element(I) anions is the indyl anion. Currently, only one example of an indyl anion is known, which was reported by Coles and co-workers in 2018. 24 The anion was reported as both an ion separated pair 5 ( Figure 2) and as the more tightly bound lithium complex. Little reactivity of this indyl compound has yet been reported. However, preliminary studies with MeOTf and bulky metal halide complexes suggest nucleophilic behaviour, 24 whilst reactions with organic azides lead to oxidation of the indium. 25

Synthesis and Structure
Since the isolation of gallyl compound 3 in 1999, various research groups have targeted an aluminium analogue, in particular using similar diazabutadiene supporting ligands. In  The structure of 6 was confirmed by X-ray crystallography, which revealed the complex to exists as a dimer in the solid state. The dimer is held together by interactions between the flanking aryl groups on the NON ligands and the potassium cations. The aluminium centres are >6.6 Å apart, implying no significant interaction between them. It can also be shown by DOSY NMR experiments that the dimeric structure of 6 is retained in solution. 27 In 6, there is a (relatively weak) interaction between the O atom incorporated into the backbone of the NON ligand and the aluminium centre (2.28 Å). 27 Subsequently, two further potassium aluminyl complexes featuring bidentate diamido ligands have been reported. 28 The first of these, K2[Al( Si NON)]2 (7), was reported by Coles and coworkers in 2019, 29 and more recently K2[Al(NCCN)]2 (8) was reported by Hill, McMullin and co-workers in 2020 ( Figure 3). 30 Both of these complexes were synthesised using a similar synthetic route to 6, i.e. by the two-electron reduction of the corresponding aluminium(III) iodide precursor with potassium. Superficially, both 7 and 8 possess similar structures to 6: both are dimeric in the solid state featuring interactions between the flanking aryl rings of the bulky diamido ligands and the potassium cations. 7 features a 6-membered aluminiumcontaining heterocycle, with a N-Al-N angle of 103.89(8)°, 29 while the 7-membered ring in 8 (not surprisingly) makes the corresponding unit somewhat wider (108.84(9)°). 30 Both of these angles however are considerably smaller than that in 6 (128.11(9)°). 27 To date, only two examples of truly 'naked' aluminyl anions have been reported, i.e. aluminyl anions that show no significant interaction with the cation. The first of these is the chargeseparated complex [K(crypt-222)][Al(NON)] 9 ( Figure 3), reported by us in 2019. 31 The second example is complex 11 discussed below. Complex 9 is prepared in straightforward fashion by the addition of cryptand-222 to 6 in THF. The solid-state structure of 9 confirms that the K + cations are encapsulated within the cryptand, and that the [Al(NON)]anion engages in no close interactions with the cation (all K···Al interactions >7 Å).

Alkyl Aluminyl Systems
Other than the series of diamido supported aluminyl anions  The structures of 10 and 11 have been determined by X-ray crystallography and are monomeric in the solid state. Both feature 5-membered aluminium heterocycles, with similar C-Al-R angles (R = C, 10; N, 11); 11 being the slightly more acute at 86.8(1) compared with 90.40(5)° for that in 10. 32,33 In the solid state, aluminyl 10 shows a significant interaction between the aluminium centre and the potassium cation (Al-K 3.4549(5) Å). However, the Al-K distance is considerably longer than the sum of the covalent radii of Al and K (3.24 Å) 34 suggesting an electrostatic interaction between the Al lone pair and cation, similar to that observed in boryl lithium 4. 20 Attempts were made by the authors to synthesise a charge separated analogue by the addition of cryptand-222 or THF to 10. However, both of these reactions led to immediate decomposition of the aluminyl anion. 32 Compound 11 on the other This article is protected by copyright. All rights reserved

Author Manuscript
hand is a bone fide charge-separated species, and only the second reported example of a 'naked' aluminyl anion. The K + cation in 11 is contained within two 12-crown-4 ligands and consequently has no close interactions with the aluminyl fragment (closest Al···K distance >6.6 Å). 33

Electronic Structure
For the purposes of like-for-like comparison, the electronic structures of all of the known aluminyl anions have been assessed by quantum chemical methods using the same level of theory as part of the present review (DFT-D3, PBE0, Def-TZVP -see SI for details). The calculations have been performed on the (full) monomeric anions, neglecting any stabilising/destabilising interactions with the counter-cations, and are summarised in Figure 5. All five anions possess relatively similar HOMO-LUMO energy gaps, varying by <0.3 eV (approx. 25 kJ mol -1 ). More important in terms of reactivity is the energy separation between the occupied and vacant orbitals at the aluminium centre (i.e. the lone pair of electrons and the empty pz-orbital). In all cases, the HOMO corresponds to a lone pair on aluminium. However, in every case bar the anionic component of 10, the LUMO is a ligand-based orbital, with the empty Al pz-orbital being higher in the orbital manifold: the LUMO+3 for the three diamido systems 6-8 and the LUMO+1 in 11 (see SI for more information). That said, the energy gap between the HOMO and the LUMO+n (LUMO+n corresponding to the empty Al centred p-orbital) is also similar in all cases, varying by a maximum of just over 50 kJ mol -1 (from 3.42 eV, 330 kJ mol -1 for 10 to 4.06 eV, 392 kJ mol -1 for 7). In general terms, many of the energetic trends shown in Figure 5 can be ascribed to structural factors familiar from carbene chemistry. 35 Thus, the lower lying nature of the HOMO for diamido systems 6, 7 and 8 compared to 10 and 11 (which feature alkyl substituents) can be ascribed to the more electronegative nature of the a-substituent. That the HOMO energies for 10 and 11 are so similar despite the narrower angle at aluminium and the presence of one s electron-withdrawing nitrogen substituent in the latter is perhaps more surprising. The diamido aluminyl systems 7 and 8 are structurally very similar, with a slight difference being the wider N-Al-N angle in 8, consistent with the larger heterocycle size (108.84(9)° cf. 103.89(8)° for 7). As such, the HOMO energy is marginally higher in the case of 8 (by 14 kJ mol -1 ), consistent with higher pcharacter. Interestingly, however, the HOMO of 6, which features both a much wider N-Al-N angle (128.1(1)°) and potentially a weak additional donor/acceptor interaction from the xanthene oxygen is in a similar energy range.
In terms of the unoccupied orbitals, the differences between 10 and 11 also conform to established trends in carbene chemistry. Anion 10 is the only system where the Al pz-orbital contributes to the LUMO; the analogous orbital in 11 is the LUMO+1. The difference in energy between these two orbitals is >30 kJ mol -1 , with that of 11 being the higher of the two, consistent with -donation from the single amido substituent. For the diamido aluminyl systems 6-8, the orbital characterized by significant Al pz character lies higher in the orbital manifold (LUMO+3 in each case), with systems 7 and 8 being associated with almost identical energies. Superficially, the most surprising finding is the much lower lying nature of the corresponding orbital for xanthene-derived system 6. On the face of it, this would imply relatively little elevation on the basis of Al-O s* character, and most significantly, much reduced N-to-Al p donation from the amido substituents (cf. 7 and 8, see SI for molecular orbital images). The underlying origins for this are geometric in nature, and find precedent in the anti-Bredt carbene systems reported by Bertrand and co-workers. 36 In marked contrast to 7 and 8, the amido substituents in 6 do not lie coplanar with the AlN2 unit (  This article is protected by copyright. All rights reserved

Author Manuscript
In summary, dialkylaluminyl compound 10, not unexpectedly features the smallest energy gap between the HOMO and the Al pz-orbital at (3.42 eV, 330 kJ mol -1 ). The corresponding energy separation for alkyl amino aluminyl 11 is higher (3.72 eV, 360 kJ mol -1 ) due to elevation of the Al pz-orbital, while those for the planar diamido systems 7 and 8 are higher still. Anion 7 features the largest energy separation (4.02 eV, 392 kJ mol -1 ) with that of the related 7-membered system 8 being slightly smaller (3.90 eV, 376 kJ mol -1 ) due to the effects of ring expansion on the HOMO. 'Base-stabilized' system 6 appears in reality to be destabilized by the constraints of the xanthene ligand backbone (with a corresponding energy separation of only 3.57 eV, 345 kJ mol -1 ) due to geometrically restricted N-to-Al p donation. In terms of understanding reactivity patterns, it should be noted that only compounds 6 and 11 have so far been isolated as 'naked' aluminyl anions. For ion pairs, interaction between the Al lone pair and the K + cation would be expected to stabilise the HOMO and thereby widen the respective energy gaps.

Nucleophilic Substitution
One of the foremost points of novelty of aluminyl anions is their ability to react as aluminium-centred nucleophiles. All of the reported aluminyl anions have shown nucleophilic behaviour to some extent, with 6, 8 and 10 showing evidence for nucleophilic substitution chemistry. 27,30,32,37,38 In our initial report we showed that 6 could undergo nucleophilic substitution reactions with carbon-and metal-centred electrophiles, as well as with a proton source. 27 The reaction of 6 with either MeI or MeOTf in toluene leads to an almost instantaneous reaction to give the Al-C bonded complex (NON)AlMe 12 with elimination of corresponding potassium salt (Scheme 1). Yamashita and co-workers have reported a similar reaction between 10 and MeOTf, yielding the aluminium methyl complex 13. 32 Both of these complexes were synthesised by nucleophilic attack on the electrophilic carbon centre of MeI or MeOTf by the aluminyl. We also showed that a similar reaction can be performed using a bulky weak acid such as H2(NON). This leads to formal nucleophilic attack on a proton (i.e. protonation) to give the aluminium hydride 14 (Scheme 1). Nucleophilic substitution reactions have also been successfully employed in the synthesis of a range of Al-metal covalent bonds. 27,30,37,38 This was first demonstrated in reactions between 6 and bulky Mg/Al iodide electrophiles, 27 which were used to generate the corresponding Al-Al and Al-Mg bonded complexes 15 and 16 (Scheme 2). A small number of additional Al-metal bonded complexes have since been synthesised by this route, including those with Al-Ca (18) 30 and Al-Y (19) 38 Al-Au (20 and 21) 37 bonds (Scheme 2). It is noteworthy that all of these nucleophilic substitution reactions use electrophiles featuring weakly coordinating anions (i.e. OTf -, I -, BPh4 -), in order to promote salt elimination. Attempting these reactions with more coordinating anions (e.g. For Cl -) tends to lead to the formation of 'ate' complexes (i.e. formal oxidative addition products) with the anion bound to the Al centre (see 2.3.2). This article is protected by copyright. All rights reserved

Author Manuscript
17 on the other hand shows no reactivity towards either of these substrates. 30 In contrast, gold has an electronegativity considerably greater than that of aluminium (2.54 on the Pauling scale). Accordingly, the Al-Au bonds in 20 and 21 show reversed polarization compared with the Al-metal bonds in 16-19. 37 Reactions of 21 with the carbon centred electrophiles CO2 and diisopropylcarbodiimide yield isolable Au-C bonded products. These reactions constitute the first examples of nucleophilic attack by a stable molecular gold complex.

Oxidative addition
As previously demonstrated for neutral Al(I) systems, the two-electron oxidation of Al(I) to Al(III) is typically a highly thermodynamically favourable process. 11 The situation appears to be little different for anionic aluminyl ions. As such, various oxidative addition reactions have been reported for aluminyl systems, in some cases involving the cleavage of very strongbonds. 27,31-33 Scheme 3. Oxidative addition reactions of aluminyl anions 6, 10 and 11; tol = toluene.
In our initial work, we showed that 6 can oxidatively cleave the H-H bond of H2 and a C-H bond in benzene under mild reaction conditions. 27 These reactions generate the formal oxidative addition products 22 and 23 (Scheme 3), demonstrating that the activation of strong -bonds by aluminyl anions is achievable. Furthermore, the reaction to yield 23 represented the first example of the oxidative addition of a benzene C-H bond at a single well-defined main-group metal centre. In more recent work, we also showed that C-H activation of naphthalene was also possible. 31 Since then, Yamashita and co-workers have shown that 10 can also C-H activate benzene at room temperature (RT) to give the formal oxidative addition product 24 (Scheme 3). 32 The same group also reported reactions of 10 with C6F6 and benzyl chloride. The reaction with C6F6 yields a mixture of formal oxidative addition products 25a and 25b, showing mono-and di-activation of the C6F6, respectively. The ratio of these products was shown to vary depending on the molar ratio of C6F6 used. In contrast, the reaction of 10 with benzyl chloride did not give the expected C-Cl oxidative addition product, but a mixture of the corresponding benzylaluminate and chloroalumane. This result suggests that 10 is susceptible to a nucleophilic substitution at the chlorine atom that generates the chloroalumane and benzylpotassium. 32 Earlier this year, we showed that the activation of N-H bonds by an aluminyl anion was also possible, with the reaction between 6 and DippNH2. This reaction gave the corresponding oxidative addition product 26 (Scheme 3). 39 Kinjo and co-workers followed this report by demonstrating that the N-H activation of ammonia and Si-H activation of phenylsilane are also possible with 11. 33 These reactions led to the formal oxidative addition products 27 and 28 (Scheme 3). The group also showed reactivity with borane in the construction of an σ-aromatic AlB2 ring and cleavage of a strained C-C bond in biphenylene to give the ring opened product 29. The latter is the second example of C-C bond activation by a potassium aluminyl complex. The first reported C-C bond activation by an aluminyl anion was the remarkable ring opening of benzene reported by us in 2019. 31 The 'naked' aluminyl 9 reacts cleanly with benzene at RT to give the 7membered aluminium heterocycle 30 in high yields (Scheme 4). This reaction involves the oxidative cleavage of a C-C bond in benzene at a single aluminium centre. Remarkably, this reaction was also found to be reversible when heated. The ring opening of benzene to give 30 was calculated to be energetically favourable (-17 kJ mol -1 ), even taking into account the loss of aromaticity in benzene. 31 This has since been confirmed in a more comprehensive computational study by Fernández and coworkers reported earlier this year, 40 where the authors also investigated the effects incorporating electron withdrawing groups (EWGs) onto the ring and the C-C activation of the polycyclic aromatic systems naphthalene and anthracene. The study found that C-C activation of the polycyclic systems was unlikely, as the competing C-H activation pathway had a lower activation barrier (and more energetically favoured products). On the other hand, incorporating EWGs onto the benzene was found to favour the C-C activation both kinetically (by lowering the activation barrier) and thermodynamically. Unfortunately, this would be difficult to confirm experimentally as 9 would likely react with most of the common EWGs. Haloarenes, for example, are known to react with the related dimeric system 6 via formal C-I oxidative addition.

31
This article is protected by copyright. All rights reserved Author Manuscript corresponding dichloroaluminate. 31 This reaction ultimately shows that aluminyl anions can be used to generate functionalized acyclic products from aromatic molecules.

Cycloaddition
Various cycloaddition reactions have been reported for aluminyl anions, including those with alkenes, alkynes and polyaromatics. 29,31,41 The first of these to be reported was the reaction between 7 and cyclooctatriene (COT) by Coles and coworkers. 29 This reaction led to the two-electron reduction of the COT molecule to the essentially planar aromatic anion [COT] 2dianion, which engages in 2 -coordination to the Al centre in the product (32; Scheme 5). The addition of 18-crown-6 to this complex leads to its conversion to 33, the product of a formal (1+4) cycloaddition addition product of COT to the aluminium centre (Scheme 5). Last year, we also reported that 6 can perform cycloaddition reactions, in this case with ethylene under slightly elevated pressure (2 bar). 31 This gave the (1+2) cyclised product 34 (Scheme 5).
The most detailed study on cycloaddition reactions involving an aluminyl anion was reported earlier this year by Yamashita and co-workers. 41 In this report, the authors showed that 11 can undergo various cycloaddition reactions with naphthalene 35, anthracene 36, diphenylacetylene 37 and 1,2-stilbene 38 (both E and Z). Compound 11 reacts with the polycyclic aromatic systems naphthalene and anthracene via a (1+4) cyclisation reaction at RT, giving the aluma-norbornadiene derivatives 35 and 36, respectively (Scheme 5). Interestingly 6 also undergoes a (1+4) cycloaddition reaction with naphthalene at RT, but at 80 °C yields C-H activation products, 31 suggesting that the cycloaddition products of this type are kinetic products. Aluminyl complex 11 also undergoes (1+2) cycloaddition reactions with diphenylacetylene and (E)/(Z)-stilbenes, forming products 37 and 38 featuring AlC2 3-membered rings (Scheme 5). Interestingly, the reactions of 11 with both (E)-and (Z)-stilbenes gave exclusively the trans-cycloadduct 38. DFT calculations revealed that the cycloaddition of 11 with (Z)-stilbene proceeds via a single transition state with a carbanion character, which results in the selectivity towards the trans-cycloadduct. 41

Oxidation
Considering the highly reducing nature of the aluminyl anions, it comes as no surprise that these systems have been shown to react with a range of oxidising agents. However, by reversing the 'normal' reactivity of aluminium from electrophilic to nucleophilic, a range of previously inaccessible (in some cases highly reactive) Al(III) complexes can be accessed via this route.
In 2019, Coles and co-workers synthesised aluminium analogues of carbonyl and dioxirane moieties by oxidation with selenium. 42 The 1:1 reaction between 7 and elemental selenium (using a catalytic amount of tri-n-butylphosphine) gave the aluminium selenide complex 39 in moderate yield (Scheme 6). The complex was found to be highly insoluble in common organic solvents due to an extended network of interactions between the potassium cation and the selenide and aryl groups in the solid state. This network could be broken with the addition of cryptand-222 to sequester the potassium cations to give a soluble monomeric form, 40 (Scheme 6). Complexes 39 and 40 represent the first examples of aluminium-selenium carbonyl analogues. DFT analysis performed on 40 found a Wiberg bond index of 1.39 for the Al-Se bond, implying significant multiple bond character, even though it is highly polarized. 42 Addition of a further equivalent of selenium to 40 in THF yields complex 41, which features an AlSe2 three-membered ring (Scheme 6), which can be viewed as a heavy dioxirane analogue. The synthesis of 41 was also reported in a single step reaction by the addition of 7 to a suspension of selenium (2 equiv.) and cryptand-222 in THF. In a more recent account, 43 the group also reported the oxidation of 7 with sulphur, targeting the sulphur analogue of 39. This reaction however did not give the targeted monosufide but the 5membered AlS4 ring 42.

Author Manuscript
Later in 2019, two reports emerged almost simultaneously describing the reactivity of aluminyl anions with CO2 and N2O. These reports were by ourselves 44 and by Anker and Coles. 45 In our report, 44 we showed that 6 reacts rapidly with 1 atm. of CO2 at RT, cleanly yielding the aluminium carbonate species 43 (Scheme 7). We also reported the analogous reaction between 6 and the phenyl isocyanate, which gave the isoelectronic aluminium carbamate complex 44. Both reactions were shown to proceed via a two-step mechanism; first by initial reduction of the substrate (to give CO or PhNC) and a highly reactive aluminium oxide intermediate, which undergoes a (2+2) cycloaddition with a second equivalent of the substrate to yield the isolated products 43 and 44. Similar reactions have been reported with other low oxidation state main group systems. 46 In an attempt to isolate the oxide intermediate, the oxidation of 6 with N2O was investigated. However, the reaction between 6 and N2O (1 atm.) in toluene did not give the targeted oxide species, but the rather unexpected cishyponitrite complex 45. As with the syntheses of 43 and 44, the reaction seems to proceed via a highly reactive oxide intermediate, which remarkably undergoes a (2+3) cycloaddition with a molecule of N2O. This reaction demonstrates the first example of a cis-hyponitrite ligand being directly synthesised from N2O. The reaction between 6 and N2O in toluene was monitored by low temp. 1 H NMR, but even at -80 °C, no signs of an intermediate were observed. However, it was found that the intermediate could be trapped by the coordination of THF to the Al centre, by performing the reaction in THF at low temperature The reaction between 6 and N2O (1 atm.) <0 °C cleanly gives the THFcoordinated monoaluminoxane 46 in high yield (Scheme 7). 44 Scheme 7. Oxidation of 6 with CO2, PhNCO and N2O and subsequent reactivity of monoaluminoxane 46.
Complex 46 was found to be stable in the solid state at RT but not in solution, decomposing over the course of several hours. 46 was found to react with CO2, PhNCO and N2O at RT to give the previously isolated products 43-45, showing that the (more reactive) base-free monoaluminoxane is a viable intermediate in these reactions. Remarkably, we also found that due to the extremely polar Al-O bond, 46 could heterolytically cleave H2 at RT to give the aluminium hydride/hydroxide species 47 (Scheme 7).
Anker and Coles reported a similar study, 45 showing that aluminyl 7 also reacts with 1 atm. of CO2 to give the aluminium carbonate complex 48 (Scheme 8). However, in their case, they found that the monoaluminoxane intermediate 49 could be isolated without the coordination of a Lewis-base, directly from the reaction of 7 and N2O (1 atm.). In an analogous fashion to 46, the monoaluminoxane 49 was found to react with CO2 and N2O (in the presence of cryptand-222) in cycloaddition reactions to give 42 and the cis-hyponitrite complex 50. Addition of cryptand-222 to 49 in the absence of CO2 or N2O, leads to intramolecular C-H activation of the ligand across the Al-O bond; C-H activation has not yet been reported for 46. In a more recent account, 43 the Coles group showed that an aluminium trithiocarbonate complex 51 (Scheme 8) could also be synthesised, by the addition of a hexane solution of 7 to a solution of CS2 (2 equiv.) at RT. However, in this case the expected monosulfide intermediate could not be isolated (see Scheme 6). Interestingly though, the group found that performing the same reaction under slightly different conditions (4 equiv. 195 K), a different product could be isolated. This was the ethenetetrathiolate complex 52 (Scheme 8), which is synthesised by a C-C bond forming reaction between two CS2 units. Remarkably, the synthesis of the lighter analogue, the ethenetetraolate complex 53 was also reported, by the reaction between CO and the monoaluminoxane 49. This complex represents the first report of an ethenetetraolate ligand.  This article is protected by copyright. All rights reserved

Author Manuscript
The oxidation of 6 and 7 have also been investigated with regards to organic azides. 39,47 Two recent reports have found that under certain conditions, these reactions can give isolable aluminium imide complexes with highly polarised (and highly reactive) Al-N bonds. Earlier this year, our group reported reactions of 6 with various azides using a variety of reaction conditions (Scheme 9). 39 The RT reaction between 6 and 2 equiv. of trimethylsilyl azide gave the aluminium tetrazine complex 54, which is essentially isostructural to the cis-hyponitrite complex 45. 44 It can be proposed that the reaction to give 54 proceeds via a similar mechanism to that to give 45; i.e. firstly by reduction of one molecule of azide to give a highly reactive aluminium imide intermediate, which undergoes a (2+3) cycloaddition reaction with a further equivalent of the azide to give 54. In an attempt to trap this azide intermediate, the 1:1 reaction between 6 and trimethylsilylazide was attempted in toluene at -80 °C. This did not give the targeted imide but instead the C-H activated toluene product 55 (Scheme 9), demonstrating the highly basic nature of the aluminium imide intermediate. In an attempt to isolate a stable aluminium imide, the bulkier azide DippN3 was sought. In a 1:1 reaction between 6 and DippN3, the targeted aluminium imide 56 was isolated in high yields (Scheme 9). The compound was found to feature an extremely polar Al-N bond (Mulliken charges +0.67 Al and -0.16 N) with a Wiberg Bond Index (WBI) of 0.705. 39 Scheme 9. Oxidation of 6 with organic azides and the activation of H2 and CO by the aluminium imide 56.
The Al imide 56 was found to be extremely reactive with regards to small molecule activation, showing reactivity towards H2 and CO under mild reaction conditions. 39 The reaction between 56 and H2 led to heterolytic cleavage of the H2 molecule to give 57 (Scheme 9), whilst the reaction with CO saw the C-C coupling of two CO molecules to give an unprecedented Al bound [(DippN)CCO2] 2fragment 58.
In a report by Anker and Coles, a similar aluminium imide 59 was reported by the reaction between 7 and MesN3 (Scheme 10). 47 This complex was found to react with CO2 (1 atm.) at RT in a (2+2) cycloaddition reaction to give the carbamate complex 60. Scheme 10. Oxidation of 7 with MesN3 to give the aluminium imide 59 and subsequent reactivity with CO2.

Summary and Outlook
Although the field of aluminyl chemistry is still in its infancy, the scope of transformations reported by these novel anions is both fascinating and diverse. In this minireview, we have covered the synthesis, structures and reactivity for the currently known series of aluminyl anions. Reactions including nucleophilic substitution, oxidative addition, cycloaddition and oxidations have already been reported, with highlights including the facile C-H and C-F activation, small molecule activation and the ring opening of aromatics. DFT calculations performed on the series of known anions revealed some interesting trends with regards to their electronic structures. Firstly, anion 6 has the lowest energy LUMO+n (empty Al p-orbital) of the series in addition to one of the lowest HOMO-LUMO gaps. 6 is also one of only two anions (the second being 11) that has been isolated free from its cation, which would otherwise stabilise the HOMO. These finding are consistent with its reactivity, as the monomeric anion 6 appears to be the most reactive of the current series (the only reported anion to ring open benzene for example). The diamido anions 7 and 8 have similar electronic structures, with 8 having a slightly smaller HOMO-LUMO gap due to a more destabilised HOMO. The alkyl systems 10 and 11 feature the high energy HOMOs as expected, due to the more electropositive substituents bound to the Al centre. 10 was found to have the smallest HOMO-LUMO gap of the series, but the charge-separated anion does not appear to be stable.
We believe the chemistry highlighted in this report only scratches the surface for what is to come from this novel class of nucleophiles. The coordination chemistry of the anions is only just beginning and small molecule activation including N2 and CO is also yet to be reported. The use of aluminyl anions in catalysis is also an area that we are currently investigating. The field is rapidly expanding and definitely 'one to watch' over the next few years.