Reduction of organic azides by indyl-anions. Isolation and reactivity studies of indium–nitrogen multiple bonds

Stepwise reaction of an indyl-anion with organic azides initially forms the indium imide, which undergoes (2 + 3)-cycloaddition to generate the indium tetrazenide.


Introduction
Low valent compounds of the group 13 elements aluminium, gallium and indium exhibit a wide range of chemical reactivity. 1 When present in the +1 oxidation-state, 2 the electron-conguration of the metallic element implies the presence of a lone-pair of electrons in an ns orbital, prompting comparisons with neutral group 14 carbenoid species. 3 Consequently, a rich area of coordination chemistry has developed, particularly focussed on the Ga(I) compounds. 4 In addition to the ability of these compounds to behave as ligands, the lighter homologues are potent reducing agents, readily giving up two electrons to attain a more stable +3 oxidation-state. This reactivity has been harnessed in a wide range of chemical reactions, 5 many of which are unique to this class of compound. 6 The most common members of this class of compound are represented by the general formula M(X), where the charge on the metal is balanced by a mono-anionic ligand, [X] À . These ligands are typically bulky, a requirement to limit (or prevent) aggregation and protect the metal from unwanted redox chemistry. This concept is best illustrated in the context of this work by the series M(BDI Ar ) (BDI ¼ b-diketiminate, [HC(CMeNAr)] À , Ar ¼ 2,6-i Pr 2 C 6 H 3 ), for which mono-metallic Al, 5f Ga 5d and In 5a compounds are known.
A recent development in the chemistry of mono-valent aluminium and indium is to employ a dianionic ligand [X 2 ] 2À to support M(I) metal centres, generating an overall negative charge on the metal-containing species, [M(X 2 )] À . Whilst this class of compound has been well studied for gallium, 4a,4b,7 the corresponding aluminyl-8 and indyl- 9 anions have only been recently isolated (Fig. 1), and hence the chemistry of these compounds is in its infancy. 10 Initial studies of [Al(X 2 )] À and [In(X 2 )] À systems indicate signicant lone-pair character at the metal (from DFT calculations), with preliminary reactivity consistent with an Al(I) or In(I) nucleophile. We report in this contribution an investigation of the reducing potential of a new potassium indyl compound towards organic azides. This class of substrate was selected to target synthetically challenging indium imide species.
Monomeric group 13 metal imides [M(X)(NR)] n (X ¼ ancillary ligand; M ¼ Al, Ga or In; R ¼ organic fragment; n ¼ 1) are of academic and practical interest in several research areas. They formally contain metal-nitrogen multiple-bonds, 11 and are implicated as intermediates in the formation of electronically important AlN, GaN and InN materials. 12 Isolation of these Fig. 1 The first reported aluminyl-and indyl-anions (Ar ¼ 2,6-i Pr 2 C 6 H 3 ). compounds remains, however, synthetically challenging and only ve examples have been crystallographically characterized since the rst structural report in 2001 (III-VI, Fig. 2). 13 Furthermore, these unusual compounds are restricted to a single example of an indium imide (VIb). 13d Structural characterization showed that the In-N bond distance in VIb (1.928(3)Å) was signicantly shorter than the range observed for monomeric In amides (2.05-2.09Å), and that the C-In-N-C core adopted a trans-bent geometry. These data were consistent with In-N multiple bond character.
The isolation of III-VI was achieved through kinetic stabilization of the M-N imide bonds using sterically demanding ligands that prevent formation of ring-and cage-structures containing m 2 -and m 3 -NR ligands. 14 A major limitation of this approach is that the bulk required to protect the imide bond from intermolecular aggregation renders it inaccessible to potential substrates, preventing any coherent study of its reactivity. It is encouraging to note, however, that in the absence of external substrates, intramolecular activation of ligand substituents can occur. This strongly suggests that once formed, the M-N imide functional group is highly reactive. 15 A general synthetic methodology to group 13 imides is the reduction of organic azides by a monovalent M(I) metal complexes (Scheme 1). 13b-d, 16 These reactions proceed with elimination of N 2 and oxidation of the metal M(III), which occurs with a concurrent increase in the coordination number of the metal. It is of note that, if insufficient steric protection is provided during synthesis, in situ addition of unreacted azide in the reaction mixture to the transient 'M]NR' bonds can occur (Scheme 1). 17 For Al and Ga, this has enabled the isolation of metallotetrazenes VII (also referred to as metal-containing tetrazoles), 17,18 which are rationalized as the product of a (2 + 3)cycloaddition. This chemistry has not been extended to indium.
In this contribution we report a new potassium indyl salt and its controlled (stepwise) reactivity with organic azides. The initial products are characterised as a new class of anionic indium(III) imide, shown crystallographically and computationally to contain In-N imide multiple bonds. Reaction of isolated examples with additional azide proceeds via a (2 + 3)cycloaddition pathway to generate tetrazenido-indium salts, containing the rst structurally characterized examples of the InN 4 -heterocycle.

Synthesis of a new potassium indyl salt
The NON Ar -ligand (NON Ar ¼ [O(SiMe 2 NAr) 2 ] 2À ; Ar ¼ 2,6-i Pr 2 C 6 H 3 ) stabilizes anionic indyl species as the indyllithium complex In(NON Ar )(Li{THF} 3 ), or in the ion-separated salt 9 Reactivity studies of these species have been hampered by their inherent instability, prompting us to examine an alternative source of the indyl anion. A modied procedure was therefore developed that avoids lithium reagents, and does not require the use of expensive crypt-222 reagents to stabilize the salt.
The reaction of 1 with InCl 3 affords a new indium-containing complex, 2. The NMR spectra show a symmetrical environment for the ligand backbone with a single peak for the SiMe 2 groups. Although this is consistent with the three-coordinate species 'In(NON Ar )Cl', elemental analysis was inconsistent with this formula and the compounds was therefore analysed by singlecrystal X-ray diffraction ( Fig. 3 and Table 1).
The asymmetric unit of 2 contains the four-coordinate indium anion [In(NON Ar )Cl 2 ] À (Fig. 3a). The charge is balanced by a potassium atom that forms p-aryl interactions with an Ar-groups of the diamide ligand, and has a close-contact with a chloride ligand. The crystal structure shows a 1-D polymer [2] n with additional interactions between the K-atom and aryl-/chloride groups from neighbouring molecules (Fig. 3b).
Reduction of 2 with two equivalents of potassium yields the new indyl compound K[In(NON Ar )] (3) as a hexane soluble, yellow solid. The 1 H NMR spectrum of 3 displays a single resonance for the SiMe 2 substituents, indicative of a symmetrical environment for the NON Ar -ligand. There are no resonances attributable to In-H hydride ligands (Fig. S10 †). 20 Scheme 1 Synthesis of group 13 metal-imides from organic azides and proposed conversion to metallotetrazenes (VII).

Reactivity of the indyl anion with organic azides
Our initial attempt to isolate an imide from 3 was made using an equimolar amount of the sterically demanding 2,6-bis(diphenylmethyl)-4-t Bu-phenyl azide (Ar ‡ N 3 , Scheme 2). The reagents were combined at À78 C, allowed to warm to room temperature and stir for 1 h. The 1 H NMR spectrum of colourless crystals 4 obtained on workup showed a loss symmetry for the NON-backbone (d H 0.58 and 0.45, 6H, SiMe 2 ) and a reduction in the intensity of the CHPh 2 resonance (d H 5.56, 1H). A new peak at 3.24 ppm (with no corresponding carbon resonance in HSQC experiments, Fig. S15 †) is assigned to an NH functionality.
The structure of 4 was determined by X-ray diffraction and shows a k 2 -C,N-N(H){C 6 H 2 (CPh 2 )(CHPh 2 )-t Bu-2,6,4} ligand chelating to a four-coordinate, anionic indium(III) centre ( Fig. 5 and Table 2). The potassium counterion is located between two aryl-substituents of the alkyl-amido ligand (C/K distances 3.148(2)-3.581(2)Å). The In-N3 distance (2.1855(14)Å) is consistent with a single bond to an amide nitrogen, and the location of electron density assigned to H1x in the difference map further supports this conclusion. Similar intramolecular activation has been observed at Al 21 and Sn 22 amido derivatives of the Ar ‡ group, although in these instances the mechanism leading to the products are not known. We propose that formation of 4 occurs via intramolecular addition of a methine  To mitigate complications from ligand activation, the reaction was repeated with the sterically less intrusive 2,4,6-trimethylphenyl azide (mesityl azide, MesN 3 ) under the conditions described above (Scheme 2). Concentration of the resulting solution and storage at À30 C gave deep orange crystals (5). The 1 H NMR spectrum is consistent with a symmetrical NONbackbone (d H 0.33, 12H, SiMe 2 ) and a freely rotating Mes group (d H 1.57, 6H, 2,6-C 6 H 2 Me 2 ).

Computational analysis of In-N imide bond
Optimisation and subsequent analysis using density functional theory conrmed the multiple-bond character of the In-N imide unit in 5 and the isolated anion [In(NON Ar )(NMes)] À from compound 6 ([6] À ) (see ESI †). This is clearly demonstrated from the increased Wiberg bond orders for this group (5, 0.59; [6] À 0.71) that are substantially higher than the In-N bonds to the NON Ar -ligand (5, 0.22/0.29; [6] À , 0.25).
To explore the nature of this bond in more detail, plausible resonance structures analogous to those examined for VIb, 13d were submitted for NBO calculations (A-D, Scheme 3). The quality criterion used to compare results calculated for the different resonance structures is the percentage of non-Lewis (n-L) components, where a lower non-Lewis percentage indicates a better representation. Resonance form C did not yield a viable solution by this method. However, structures A (triple bond), B (double bond) and D (single bond) all have a low n-L contribution to their overall NBO solution (Table S2 †). The best localisations were achieved for multiple-bonded A and B (n-L ¼ 1.965% and 1.990% respectively), while D was only slightly less well localised (n-L ¼ 2.094%). Although it is difficult to extract a precise numerical value for the multiplicity of the In-N imide bond from these computational data, the results conrm a strong multiple-bond component in accordance with crystallographic results, and observed reactivity (vide infra).
Quantum Theory of Atoms In Molecules (QTAIM) analysis of 5 and [6] À has been performed. The bond critical point between the In and N imide bonds have a low ellipiticity (3) of 0.079 and 0.072 for 5 and [6] À , respectively, inconsistent with a conventional In]N double where a larger value (>0.25) is predicted. These data suggest a non-elliptical cross-section of electron density in the In-N bond vector. This is consistent with a model proposed by Power and co-workers in related gallium imides related to VIa, 13b in which an organogallium(I) species interacts with a singlet nitrene, with incomplete donation of electron pairs (represented by dashed lines in Fig. 8).

Reactivity of indium imides with organic azides
As the imido-mesityl substituents are considerably less bulky than the terphenyl groups in IV-VI, we wished to determine whether the In-N imide bond in 5 was available for controlled reactivity studies. Inspired by the proposed formation of metallotetrazenes from group 13 metal imides (Scheme 1), we investigated the reactivity of isolated samples of 5 with organic azides RN 3 (R ¼ Mes, SiMe 3 ).
Addition of a solution of RN 3 to an orange solution of 5 at room temperature resulted in decolorization over an approximate 5 minute period (Scheme 4). NMR spectra show changes corresponding to the addition of mesityl (7) or SiMe 3 (8) groups, consistent with their incorporation in the product. In agreement with these data, the composition of the products as the rst examples of indium tetrazenido compounds was conrmed by X-ray crystallography ( Fig. 9 and 10, Table 4). These results demonstrate that in this instance, the potassium atoms in 5 do not adversely affect the reactivity of the In-N imide bond. We therefore propose that 5 behaves chemically as 'In]NMes', and Scheme 3 Possible resonance structures for indium imide anions.  that the formation of the unsymmetrical tetrazene 8 strongly endorses the previously assumed (2 + 3)-cycloaddition pathway of transient imides.
In all cases the anion comprises two approximately orthogonal rings fused at a four-coordinate indium centre. The metallotetrazene rings are essentially planar, with nitrogennitrogen bond lengths indicating double-bond character between atoms in the 3-and 4-positions of the heterocycle (see VII, Scheme 1). These parameters are consistent with neutral aluminum- 17,18 and gallium-18b derivatives, although we note that compounds 7 and 8 represent the rst structurally characterized indium compounds containing the tetrazenide ligand, and are unique examples where the MN 4 -heterocycle is a component of an anionic species.

Conclusions
This work describes the rst detailed reactivity study of an indylanion. We conrm that the negative charge associated with the indium centre does not adversely affect their ability to act as a reducing agent towards organic azides. The reactions proceed cleanly with elimination of dinitrogen and oxidation of the indium(I) to In(III). The isolated compounds have been structurally veried as a new class of anionic indium imide, shown computationally to contain In-N imide multiple bonds. Furthermore, we demonstrate that the reduced size of the imidesubstituent in this work compared with previous examples allows access to the In-N imide bond, demonstrated by the reaction with additional equivalents of azide. The products from this (2 + 3)-cycloaddition are the rst time that this reaction has been extended to indium, and crystallographic analysis conrms a planar InN 4 -heterocycle as a component of the anion.

Conflicts of interest
There are no conicts to declare.