s-Block metal complexes of superbulky (tBu3Si)2N−: a new weakly coordinating anion?

Sterically hindered amide anions have found widespread application as deprotonation agents or as ligands to stabilize metals in unusual coordination geometries or oxidation states. The use of bulky amides has also been advantageous in catalyst design. Herein we present s-block metal chemistry with one of the bulkiest known amide ligands: (tBu3Si)2N− (abbreviated: tBuN−). The parent amine (tBuNH), introduced earlier by Wiberg, is extremely resistant to deprotonation (even with nBuLi/KOtBu superbases) but can be deprotonated slowly with a blue Cs+/e− electride formed by addition of Cs0 to THF. (tBuN)Cs crystallized as a separated ion-pair, even without cocrystallized solvent. As salt-metathesis reactions with (tBuN)Cs are sluggish and incomplete, it has only limited use as an amide transfer reagent. However, ball-milling with LiI led to quantitative formation of (tBuN)Li and CsI. Structural characterization shows that (tBuN)Li is a monomeric contact ion-pair with a relatively short N–Li bond, an unusual T-shaped coordination geometry around N and extremely short Li⋯Me anagostic interactions. Crystal structures are compared with Li and Cs complexes of less bulky amide ligands (iPr3Si)2N− (iPrN−) and (Me3Si)2N− (MeN−). DFT calculations show trends in the geometries and electron distributions of amide ligands of increasing steric bulk (MeN− < iPrN− < tBuN−) and confirm that tBuN− is a rare example of a halogen-free weakly coordinating anion.


Introduction
According to the formal IUPAC denition, 1 metal amide complexes should not be described as organometallic compounds.However, due to their crucial role and close relationship to organometallic chemistry, they are oen considered as such. 2 Especially the group 1 metal amides have found widespread applications as amide ligand transfer reagents to access metal amide complexes across the periodic table.Given the considerably higher electronegativity of N compared to C, amide anions are somewhat less Brønsted basic than corresponding carbanions.Despite this fundamental difference, carefully chosen, sterically hindered amide bases like lithium diisopropylamide (LDA) or lithium 2,2,6,6-tetramethylpiperidide (LiTMP) found fame as powerful non-nucleophilic deprotonation reagents. 3,4Amide anions are also markedly different from carbanions by the presence of two lone-pairs of electrons at N, which sets them apart as great bridging ligands.
Their rich coordination chemistry has led to stunning examples of their unique deprotonation power 5,6 and self-assembled aggregates in which multiply deprotonated substrates are embedded in a ring of metal cations that act as an inverse crown ether 7 (e.g.I in Scheme 1).
Within the large range of amines, the silyl-substituted amine HN(SiMe 3 ) 2 (1,1,1,3,3,3-hexamethyldisilazane) is arguably the most common source for synthesis of metal complexes. 8,9The corresponding (Me 3 Si) 2 N − ligand, oen abbreviated as HMDS or N 00 , is attractive while its Me 3 Si-substituents offer steric protection of the metal center and stabilize the neighbouring negative charge on N by polarization and negative hyperconjugation; cf. the pK a values for HN(SiMe 3 ) 2 (25.8) and HNiPr 2 (35.7). 10,11There are, however, also drawbacks of this ligand which are exemplied by N-Si bond cleavage 12,13 or Me-Si deprotonation. 14n order to achieve greater stability and improve steric protection, a large range of bulkier silyl-substituted amides have been designed. 15Such bulky monodentate ligands achieved stabilization of low-oxidation-state Zn I and Mg I centers (e.g. in II). 16,17They also found application in the synthesis of nearly linear lanthanide metal complexes which have been studied extensively for their magnetic properties. 18Our interest in bulky amide ligands is related to their ability to lower the aggregation state of alkaline-earth (Ae) metal complexes 19,20 and reduce the nuclearity of Ae metal hydride clusters. 21For this reason, bulky Ae metal amide complexes like Ae[N(SiiPr 3 ) 2 ] (III) are much more reactive in hydrogenation catalysis 21 than Ae[N(SiMe 3 ) 2 ] 2 catalysts 22,23 and under controlled reaction conditions even found application as catalysts for Hydrogen-Isotope-Exchange (HIE) in aromatic substrates. 24As the activities of these catalysts increase with the size of the amide ligand, we are interested in s-block metal complexes with even bulkier amide ligands.It is, however, questionable what the limitations to the bulk of the substituents are.Herein, we report on the unusual coordination chemistry of the extremely bulky (tBu 3 Si) 2 N − anion (abbreviated: tBu N − ), for which there is hitherto a complete lack of knowledge, and show comparisons with the smaller (iPr 3 Si) 2 N − ( iPr N − ) and (Me 3 Si) 2 N − ( Me N − ) anions.

Syntheses
The parent amine (tBu 3 Si) 2 NH ( tBu NH) was obtained by thermal decomposition of (tBu 3 Si)N]N-NH(SitBu 3 ) following a method reported by Wiberg (Scheme 2). 25 A slightly modied procedure gave the ligand precursor in crystalline purity (yield: 71%).Although the Si-N-Si angle in tBu NH is reported to be 167(2)°, 25 a later renement 26 and our structure determination show a linear structure with N on an inversion center.There are, however, large N displacement factors in two dimensions (U 1 0.072, U 2 0.018, U 3 0.054) which is the plane perpendicular on the Si-N-Si axis (Fig. 1a).This shows that, similar to [(Ar) 5 -Cp] 2 Ae sandwich complexes, 27 the molecule slightly deviates from linearity and the central N atom is disordered over a ring of positions.Due to disorder the exact location of the N-H hydrogen atom could not be determined.As the N-H functionality is fully embedded in the bulk of two very large tBu 3 Sisubstituents (Fig. 1b), the deprotonation of tBu NH turned out to be extremely challenging.
Lithiation with nBuLi in boiling hexane, or with nBuLi/ TMEDA at somewhat lower temperatures to avoid TMEDA deprotonation, 28 did not give any conversion.Treating tBu NH with KH in boiling THF or under microwave conditions at 180 °C did not show reaction.In contrast, the somewhat smaller amine iPr NH could be smoothly deprotonated even in toluene. 18ddition of 18-crown-6 did give deprotonation and a small batch of crystals with composition [K + $(18-crown-6)(THF) 2 ] [ tBu N − ] (1) was isolated in very low yields (crystal structure: Fig. S37 and S38 †).Repeated attempts to improve this synthesis led to the conclusion that this procedure is irreproducible.As tBu NH did not even react with the superbase mixture nBuLi/ KOtBu, this amine is particularly resistant towards deprotonation.
Analysis of the thermodynamics of the reaction by DFT calculation (B3PW91/def2tzvp including GD3BJ dispersion corrections) showed that silyl-substituents have a strong stabilizing effect on the amide anion and that deprotonation of silylsubstituted amines becomes more exergonic with increasing bulk (Scheme 3).The reluctance of tBu NH to be deprotonated is therefore exclusively due to kinetic problems related to the very poor accessibility of the N-H proton.We reasoned that application of a metal electride 29 may solve this problem.Addition of tBu NH to a dark-blue solution of K 0 in NH 3 led to rapid decolorization, however, we were only able to isolate highly insoluble KNH 2 in the form of a grey powder.The latter is likely formed by reaction of intermediate ( tBu N)K with NH 3 which, although contrathermodynamic (see Scheme 3), can be explained by the insolubility and precipitation of KNH 2 .Reaction of tBu NH with a blue K 0 /18-crown-6 electride solution 29a led to crown ether decomposition, which is a known side-reaction for such reagents. 30However, the reaction with a blue Cs + /e − electride formed by addition of Cs 0 to THF gave full conversion of the amine.The reaction is extremely slow and needs slight heating at 40 °C for four days.Under these conditions not only ( tBu N)Cs (2) but also several solvent decomposition side-products are formed.As THF is prone to C-H activation in the OCH 2 group and subsequent ring opening, 31 we changed the solvent to the more robust tetrahydropyran (THP).Reaction of a Cs 0 /THP electride solution with tBu NH resulted in considerably less sideproduct formation and gave solvent-free ( tBu N)Cs in 84% yield.Although formally a deprotonation, this procedure should be described as a redox reaction in which one equivalent of Cs 0 reduces tBu NH to tBu N − and 0.5 equivalent of H 2 (the latter could be detected by 1 H NMR monitoring).We used excess of Cs 0 to accelerate conversion and reduce the amount of side-products.This synthetic method shows that electrides, which nowadays can also be obtained simply by ball-milling, 29c may have strong potential as a reagent for the formal deprotonation of challenging substrates.
The amide complex ( tBu N)Cs (1) with a large heavy alkali metal cation could potentially be used in syntheses of Ae amide complexes by salt-metathesis.However, reactions between ( tBu N)Cs and AeI 2 to give ( tBu N) 2 Ae and CsI were found to be problematic.These ligand exchange reactions in either THF or toluene are very slow, irreproducible and incomplete.This led to complex reaction mixtures from which in one case we were able to isolate some crystals of the ion-pair [ISr + $(THF) 6 ][ tBu N − ] (3) for which we could determine the structure (Fig. S43  Interestingly, ball-milling ( tBu N)Cs and LiI and subsequent extraction with hexane cleanly led to formation of ( tBu N)Li ( 6) which was isolated in the form of colorless crystals in 55% yield.The driving force for this exchange reaction is the formation of CsI.Note that ( tBu N)Li could not be obtained directly from the amine and a Li base.Addition of either 12-crown-4 or PMDTA to a THF solution of ( tBu N)Li led to crystallization of the free tBu N − anion with non-coordinating Li + $(12-crown-4) 2 (7, 71% yield) or Li + $(PMDTA)(THF) (8) cations (crystal structures: Fig. S49-S52 †); PMDTA = N,N 0 ,N 0 ,N 00 ,N 00 -pentamethyl-diethylenetriamine.
Scheme 4 Comparison of polar solvent-free Li and Cs amide structures with ligands of increasing bulk: Me N < iPr N < tBu N.
shaped coordination geometry.The Li + cation in ( tBu N)Li is embedded in a surrounding of tBu groups with extremely short anagostic interactions.The shortest Li/CH 3 distances in the non-disordered molecules range from 2.303(8) to 2.338(7) Å (average: 2.325 Å).This is considerably shorter than the shortest reported anagostic Li/CH 3 interactions for low-coordinate Li complexes by Snaith (2.415(7)-2.823(7)Å; average: 2.607 Å) or Scherer (2.658(5) Å). 33,34 The shortest Li/H distances in a structure determination of ( tBu N)Li with freely rened H atoms vary from 1.68-2.00Å but without neutron diffraction data, these values are not accurate.
The very short Li/CH 3 distances in ( tBu N)Li indicate strong secondary bonding interactions.Since metal/H 3 C-X bonds become stronger with decreasing electronegativity of X, i.e. with increasing H 3 C d− -X d+ bond polarity, 35 it is surprising that the Li/H 3 C-C bonds in ( tBu N)Li (average: 2.325 Å) are so much shorter than the average Li/H 3 C d− -Si d+ anagostic interactions in trimeric 36,37 or tetrameric 38 ( Me N)Li of 2.888 Å and 2.955 Å, respectively.The N-Li bond in ( tBu N)Li can be easily cleaved by addition of 12-crown-4 which resulted in crystallization of the ion-pair [Li + $(12-crown4) 2 ][ tBu N − ] (7) (see Fig. S49 and S50 †).
The amide complex with the larger Cs + cation, ( tBu N)Cs, is hardly soluble in hexane.Despite numerous attempts, it was impossible to obtain crystals from strictly nonpolar solvents.However, it does dissolve in polar but weakly coordinating chlorobenzene 39 from which it crystallized solvent-free.The crystal structure (Fig. 2b) shows a nearly linear tBu N − anion (Si-N-Si 177.6(1)°) and very short Si-N bonds (1.652(2) Å).In contrast to ( tBu N)Li, there is no N-metal bonding (N/Cs 5.520(2) Å).The Cs + cation resides in a cavity spanned by four tBu N − anions in which there are at most weak anagostic Cs/ CH 3 interactions (3.497(3)-3.562(3)Å).The Cs + atom is disordered over two positions separated by circa 0.45 Å.This is likely due to the extremely weak electrostatic bonding interaction between tBu N − , one of the largest amide anions, and Cs + , the largest stable metal cation.Addition of toluene led to strong Cs + /toluene interactions and crystallization of monomeric ( tBu N)Cs$(toluene) 3 in the monoclinic space group P2 1 /c with one molecule in the asymmetric unit (Fig. 2c).The Cs + cation is bound to the three toluene solvent molecules in a h 6 -fashion with Cs-ring centroid distances ranging from 3.251 to 3.321 Å.The Cs coordination sphere is completed by two anagostic Cs/CH 3 interactions of 3.423(2) and 3.605(2) Å.There is hardly structural information on organometallic Cs compounds.The few reported Cs/ H 3 C d− -Si d+ anagostic interactions in ( Me N)Cs complexes, which should be stronger than Cs/H 3 C-C bonds, 35 are much longer (range: 3.623(4)-3.879(5)Å). 39,40 This again shows the importance for secondary bonding in metal complexes with the tBu N ligand.The Si-N-Si backbone in the anion tBu N − is close to being linear (177.6(1)°).
Although there are many structures of alkali metal complexes with bulky bis(silyl) amide ligands, 42,43 a comparison is oen difficult due to use of different coordinating solvents or different silyl substituents.In order to compare Li and Cs amide structures with ligands of increasing steric bulk, Me N < iPr N < tBu N, we therefore also synthesized ( iPr N)Li (9, 85% yield) and ( iPr N)Cs (10, 68% yield).Both could be crystallized either from apolar hexanes or slightly polar aromatic solvents like toluene.
There is a clear trend in the crystal structures of solvent-free ( Me N)Li, ( iPr N)Li and ( tBu N)Li (Scheme 4).][38] The bulkier ( iPr N)Li crystallizes as a monomer with strong anagostic interactions to neighbours resulting in a onedimensional polymer.Addition of toluene results in formation of a solvated monomer.The bulkiest ( tBu N)Li forms a discrete monomer in which Li is saturated only by intramolecular anagostic interactions.Although the N-Li bond in ( tBu N)Li is easily broken by addition of 12-crown-4, crystallization of ( Me N)Li with this Li + specic crown ether gave monomeric ( Me N)Li$(12-crown-4) with a short N-Li contact of 1.965(4) Å. 32 Although alkali metal complexes with large cations usually tend to form extensive coordination polymers, the structure of solvent-free ( Me N)Cs is only dimeric, while addition of toluene results in a linear array of dimers bridged by Cs/h 6 -toluene interactions (Scheme 4). 45In contrast, ( iPr N)Cs crystallizes from hexanes as a monomer with a short Cs-N distance and an extended network of intra-and intermolecular anagostic interactions.The complex with the bulkiest amide ligand, ( tBu N)Cs, crystallized as an ion-pair in which the amide ligand functions as a weakly coordinating anion (WCA) through longer anagostic Cs/C interactions.From toluene the complex crystallized as ( tBu N)Cs$(toluene) 3 in which there is also no Cs-N contact.
The superbulky amide anion tBu N − is therefore an odd example of a halogen-free WCA.Nearly all WCAs are heavily uorinated or halogenated 46,47 and there are only few examples of halogen-free WCAs. 48,49The latter are especially desirable for their great stability as electrolytes in metal batteries. 50,517][58] A rst example of unsupported metal/O(SiMe 3 ) 2 bonding was only recently structurally characterized. 59It has been previously discussed that the main reason for the poor electron pair donating abilities of R 3 SiOSiR 3 is negative hyperconjugation, i.e. delocalization of free electron pairs at the central O into the s*(Si-R) bond, leading to wide Si-O-Si angles, short Si-O bonds and long Si-R bonds. 54However, this theory has been abandoned and its unusual geometry can also be explained with the ionic character of the Si-O bond which increases upon widening the Si-O-Si angle. 52,55Despite the strong similarities between isolobal (R 3 Si) 2 O and (R 3 Si) 2 N − species, the bonding and electronic structure of bis(silyl) amide ligands has not been described in detail.Herein we provide DFT calculations on monomeric Li and Cs complexes and free anions of increasing bulk: Me N − < iPr N − < tBu N − (Scheme 5).
The structures have been optimized at the B3PW91/def2tzvp level of theory.As it was found that using Grimme's third dispersion correction with Becke-Johnson damping (GD3BJ) gave a better match with the experimental data, we only show results with dispersion correction.The coordination geometry of Li in ( tBu N)Li deserves some special attention.Also in the calculated structure the Li + cation is fully embedded in the ligand and bound by a short N-Li bond and anagostic interactions.This results in an unusually large value for the buried volume (V bur ). 60For monomeric structures the following values have been calculated: ( Me N)Li 45.4%, ( iPr N)Li 72.7% and ( tBu N)Li 86.6% (H atoms have been included, Table S2 and Fig. S32-S34 †).
Comparison of the optimized structures for ( Me N)Li, ( iPr N)Li and ( tBu N)Li shows that the N-Li bonds slightly elongate and the Si-N-Si angles considerably widen when the ligand bulk is increased.A similar trend can be recognized for the corresponding Cs amide complexes with the difference that the Cs-N bond in ( tBu N)Cs is completely cleaved, even in a calculated gas phase structure.The difference between The Si-N bond distances remain surprisingly constant when increasing the bulk of the silyl substituents.
As Li and Cs are extremes in the alkali metal series, we also calculated the structures of ( tBu N)M (M = Na, K, Rb; Fig. S24 †).The gradual increase in calculated N-M distances is larger than the increase in ionic radii (Table 1).The difference between these values, (N-M) -(ionic radius), steadily increases from Li to Rb and at Cs becomes extremely large.The calculated N-metal distances in ( tBu N)M compare well with those in crystal structures of monomeric ( Me N)M complexes in which metals have been solvated with multi-dentate ligands (Table 1).These data show that although the tBu N − anion becomes gradually less coordinating from Li + to Rb + , it is truly weakly coordinating only for Cs + .However, it should be considered that these are gas phase calculations in which charge separation is notoriously difficult.Even weak donor ligands like aromatic solvents may induce N-M bond dissociation already for smaller metal cations.
Comparison of the free amide anions show a similar widening of Si-N-Si angles.Optimization of the Me N − anion without considering dispersion gave a linear minimum with a Si-N-Si angle of 179.8°.2][63] The value for the iPr N − anion (158.1°) also corresponds with experiment (152.8(1)°). 64The calculated Si-N-Si angle in tBu N − is truly linear (180.0) and ts the angle in the crystal structure of ( tBu N)Cs ( 177 3 upon becoming more linear. 65The invariance in Si-N bond lengths in Me N − , iPr N − and tBu N − is likely due to two opposing effects that counterbalance each other: (a) increasing bulk results in Si-N-Si widening and therefore Si-N bond shortening, (b) increasing bulk results in Si-N bond lengthening due to increased repulsion of the silyl substituents.In contrast to the invariance of the Si-N bond lengths, the Si-C bonds become longer with widening the Si-N-Si angle.This effect is amplied by increased repulsion of the silyl substituents.Lengthening of the Si-C bonds is in agreement with decreasing Wiberg bond indices (WBIs) (Scheme 5).
Although the Si-N bonds in the series are of similar length, the WBIs are reduced going from Me N − (0.92) to tBu N − (0.78) due to an increase of the ionic character in the Si-N bond.The increase of Si-N bond ionicity in the row Me N − < iPr N − < tBu N − is evident from the charges calculated by Natural-Population-Analysis (Scheme 5). 66Bulky substituents result in Si-N-Si widening and an increase of negative charge on N from −1.73 ( Me N − ) to −1.96 ( tBu N − ) with a concomitant increase of positive charge from +1.83 to +2.15 on Si.The ionicity of the Si-N bonds is also evident from atoms-in-molecules analysis. 67Covalent bonds typically have large electron densities r(r) and negative values for the Laplacian V 2 r(r) in the bond-critical-point (bcp).The Si-N bonds in the series have low electron densities (0.14-0.15 a.u.) and positive Laplacians (0.77-0.80 a.u.), typically observed for ionic bonding (Scheme 5).
It seems counterintuitive that the amide anion with the highest charge on N ( tBu N − ) shows the poorest donor ability.Although this partially may be explained by steric hindrance, poor coordination properties have also been described for linear H 3 Si-O-SiH 3 in which sterics do not play any role. 55A simple electronic explanation can be found in differences in the spatial arrangement of electrons at the donor site.Whereas the electron density at the central O in bent ether ligands is directional and has the form of a cashew nut, the electron pairs in a linear ether are in a circular donut shape. 55Despite the high electron density at O in the latter, there is an unfavorable nondirectional distribution of the charge density.Similar arguments explain the weakly coordinating behavior of the nearly linear tBu N − anion.
The weakly coordinating behavior of tBu N − is also nicely demonstrated by comparison of its electrostatic potential isosurface with those of iPr N − and Me N − (Fig. 3).The negatively charged N (in red) in tBu N − is completely buried by ligand bulk and the positively charged surface (in blue) is highly unfavorable for interactions with cations.

Conclusion
The superbulky amine (tBu 3 Si) 2 NH ( tBu NH) can be easily obtained by a synthetic method reported by Wiberg and coworkers.However, deprotonation of the relatively acidic N-H functionality turned out to be extremely difficult and could not even be achieved with nBuLi/KOtBu superbase mixtures.This stands in complete contrast to the facile deprotonation of the somewhat smaller amine ligand (iPr 3 Si) 2 NH ( iPr NH) which reacts smoothly with nBuLi or KH. 18,42The origin for its reluctance to be deprotonated lies in steric congestion and poor accessibility.However, using a blue electride solution of Cs + /e − in THF resulted in slow deprotonation and formation of ( tBu N) Cs.This reagent is also the key to ( tBu N)Li which could be obtained by reaction of ( tBu N)Cs with LiI.However, in contrast to the facile salt metathesis reactions with ( iPr N)K, 20,43 the application of ( tBu N)Cs in the synthesis of other metal complexes is limited.
These solvent-free superbulky amide complexes could be obtained in crystalline form by recrystallization from either apolar hexanes or weakly coordinating polar solvents like chlorobenzene.Although the N atom in the anion tBu N − is completely shielded by large bulky tBu 3 Si-substituents, small cations like Li + can be embedded between substituents and form relatively short N-Li bonds (1.913 Å).This only results in very slight bending of the Si-N-Si backbone (167.4°) and therefore an unusual T-shaped coordination geometry around N. Replacing Li + for the much larger Cs + cation led to cleavage of the metal-N bond and formation of an ion-pair, even in the absence of stabilizing solvent molecules.
Structural comparison of a range of Li and Cs amide complexes with ligands of increasing bulk ( Me N < iPr N < tBu N) shows that the tBu N − anion can be considered a WCA, at least for large cations like Cs + but not for Li + .Reduction of the ligand bulk to iPr N already results in N-Cs bonding.The presence of polar solvents like ethers or amines leads to cleavage of the tBu Nmetal bond, also for metals like Li + , Mg 2+ or Sr 2+ .
DFT analysis of a series of Li and Cs amide complexes with ligands of increasing bulk support these experimental observations: ( tBu N)Li optimizes as a contact ion-pair with a short N-Li bond whereas ( tBu N)Cs optimizes as a separated ion-pair with a long N/Cs distance.Calculations also show that increasing the Si-N-Si angle results in more ionic and shorter Si-N bonding.Although the negative charge on N is largest in the bulkiest linear amide anion, tBu N − , this is the anion showing the poorest ability to coordinate to metals.This can be explained partially by steric arguments but also nds it origin in the non-directional distribution of electron density along the N atom.
The very poor donor ability of the tBu N − anion can be exploited in the search for new non-or weakly coordinating anions.It is a rare example of a WCA that is free of halogens.We are currently investigating potential applications of tBu N − and similar bis(silyl)amide anions as WCAs.

Fig. 1 Scheme 3
Fig. 1 (a) ORTEP representation for the crystal structure of (tBu 3 Si) 2 NH ( tBu NH).The H atom at N is disordered and was not located.(b) Spacefilling model for the crystal structure of (tBu 3 Si) 2 NH ( tBu NH).

Scheme 5
Scheme 5 Comparison of DFT-optimized Li and Cs amide complexes and the free anions Me N − , iPr N − and tBu N − (B3PW91/def2tzvp including GD3BJ dispersion corrections) showing distances (Å), angles (°), Wiberg bond indices (WBIs) and values for r(r) and the Laplacian V 2 r(r) in the bond-critical point (a.u.).Values for crystal structures are given between square brackets.

Table 1
Comparison of the calculated N-M distances (Å) in ( tBu N)M complexes (M = Li-Cs) with ionic radii for six-coordinate metal cations and N-M distances in monomeric ( Me N)M complexes with multidentate ligands a