A Blueprint for the Stabilization of Sub‐Valent Alkaline Earth Complexes

Abstract The study of sub‐valent Group 2 chemistry is a relatively new research field, being established in 2007 with the report of the first Mg(I) dimers. These species are stabilized by the formation of a Mg−Mg covalent bond; however, the extension of this chemistry to heavier alkaline earth (AE) metals has been frustrated by significant synthetic challenges, primarily associated with the instability of heavy AE−AE interactions. Here we present a new blueprint for the stabilization of heavy AE(I) complexes, based upon the reduction of AE(II) precursors with planar coordination geometries. We report the synthesis and structural characterisation of homoleptic trigonal planar AE(II) complexes of the monodentate amides {N(SiMe3)2}− and {N(Mes)(SiMe3)}−. DFT calculations showed that the LUMOs of these complexes all show some d‐character for AE = Ca−Ba. DFT analysis of the square planar Sr(II) complex [Sr{N(SiMe3)2}(dioxane)2]∞ revealed analogous frontier orbital d‐character. AE(I) complexes that could be accessed by reduction of these AE(II) precursors were modelled computationally, revealing exergonic formation in all cases. Crucially, NBO calculations show that some d‐character is preserved in the SOMO of theoretical AE(I) products upon reduction, showing that d‐orbitals could play a crucial role in achieving stable heavy AE(I) complexes.


Computational analysis
Density Functional Theory (DFT) studies were undertaken using Gaussian 16 with BP86 optimised structures, alongside electronic structure analyses completed with NBO7 (see Tables 1-2 and Supporting Information for full computational details).Relative free energies of the [AE{N(SiMe 3 ) 2 } 3 ] À monoanion and the proposed reduced [AE{N(SiMe 3 ) 2 } 3 ] 2À dianion (Scheme 3) were calculated and confirm the reduction to be exergonic in nature (Ca: À 20.5 kcal/mol; Sr: À 20.2 kcal/mol; Ba: À 22.1 kcal/ mol -see Table S6 in Supporting Information). [59]Molecular Orbital analysis of the [AE{N(SiMe 3 ) 2 } 3 ] À monoanions and [AE {N(SiMe 3 ) 2 } 3 ] 2À dianions showed some d z2 character in a lowestunoccupied molecular orbital (specifically LUMO + 6) [60] of the monoanions (see Table 1).Natural Bond Orbital (NBO) calculations of the monoanion complexes in [AE{N(Mes)(SiMe 3 )} 3 ] À confirm that the LUMO has some d z2 character (NBO d%: 0.4-2.3;see Supporting Information), whilst in [AE{N(SiMe 3 ) 2 } 3 ] À this is observed in higher MOs (LUMO + 6; NBO d%: 1.3-2.1;see Supporting Information); the differing orbital profile could be tentatively ascribed to the structural differences between the two families of compounds, particularly with regards to CÀ N distances and involvement of aryl substituents (see above) in contrast to the silyl groups.Furthermore, the singly-occupied molecular orbital, SOMO(α), of the dianionic reduced species shows electron density residing on the AE metal centre (as shown in Table 2), with a higher d-character in [AE{N-(Mes)(SiMe 3 )} 3 ] 2À (NBO d%: 6.8-10.9%; see Supporting Informa  S8).These structural differences could be due to mesityl substituent conformations in each amide ligand as observed in precursors 3-AE (see above), with this puckering affecting an increase or decrease in the d-character of calculated SOMOs based on steric effects.A similar puckering was also observed in the solid state structure of [Y{N(SiMe 3 ) 2 } 3 ] À ([Kr]4d 1 ) reported by Evans and co-workers, [41] which could be considered as isoelectronic to [Sr{N(SiMe 3 ) 2 } 3 ] 2À .Conversely, [Sc{N(SiMe 3 ) 2 } 3 ] À ([Ar]4d 1 ), isoelectronic to [Ca{N-(SiMe 3 ) 2 } 3 ] 2À , displays a regular trigonal planar geometry. [39]or completeness, the same computational studies were performed on Mg analogues, which also afforded an exergonic reaction profile for the reduction to Mg(I) (À 14.5 kcal/mol), though to a lesser extent compared to the heavier analogues, and reassuringly no d-orbital contributions observed in the MOs.Interestingly, in the case of [AE{N(Mes)(SiMe 3 )} 3 ] À mono-  anions the ΔG of the reduction is affected by the orientation of the mesityl substituents.For the Mg analogue, a large ΔG value (À 32.9 kcal/mol) is obtained from the conformation with two mesityl substituents below the coordination plane and one situated above.Conversely, a smaller ΔG value (À 16.8 kcal/mol) is obtained by adopting a conformation where all three mesityl substituents are positioned above the coordination plane as seen in 3-Ca.The latter is more in line with the energies calculated for the [AE{N(SiMe 3 )} 3 ] À /2À series, thus highlighting the role played by the ligand substituents and their conformations in the stability of reduced species.
Owing to the results obtained with trigonal planar AE trisamide complexes, we postulated that a more generalised equatorial ligand field could induce an increase in d-character for the LUMO of divalent Ca, Sr and Ba complexes, thus opening more possibilities in terms of ligand design.To verify this, we turned our attention towards complexes featuring a square planar coordination geometry.Square planar Mg(II) porphyrin complexes are relatively common, owing to the rigid conformation of the ligand dictating the geometry around the metal center; [61][62][63][64][65][66][67][68][69] however, these species are not suitable for our molecular design because of the redox activity of porphyrins and their incompatibility with large AE metals.Besides constrained square planar porphyrin complexes, a square planar geometry is a very rare coordination mode for the AE metals owing to the lack of crystal field stabilization.As a result there are a handful of near square planar magnesium complexes supported by nitrogen-based ligands in the literature, [70][71][72][73][74][75] and, to the best of our knowledge, only one example of a heavy AE complex in a square planar geometry, [Sr{N(SiMe 3 ) 2 } 2 (dioxane) 2 ] n (4) reported by Lappert and co-workers. [76]Therefore, we decided to use 4 as a model for expanding our strategy through   S8); and in this case, this structural puckering seems necessary to accommodate the presence of the additional charge.

Discussion
Our modelling shows that the constrained structural geometries imposed by sterically demanding ligands can have a significant effect on the of d-orbitals, even in the case of the AE metals for which there are negligible crystal field effects.
Though NBO analysis reveals a predominant s-character for the LUMO and SOMO(α) in all precursors and reduced species respectively, the involvement of d-orbitals in frontier orbitals is significant.It is noteworthy that d-character in Sc 2 + , Y 2 + and La 2 + species (isoelectronic with Ca + , Sr + and Ba + respectively) has been evaluated via X-ray absorption methods, [42] but, to the best of our knowledge, details of NBO or Mulliken analysis of frontier orbitals have not been reported.In the case of the Lu 2 + complex [K(crypt)][Lu(Cp') 3 ] ([Xe]5f 14 4d 1 ) Mulliken population showed 42 % metal character and 48 % d-character in the SOMO(α). [34]In our studies, the NBO analysis of the SOMO(α) of [AE{N(R)(SiMe 3 )} 3 ] 2À and [Sr{N(SiMe 3 ) 2 } 2 (dioxane) 2 ] À shows vary-ing degrees of d-orbital participation, with additional p-orbital contribution as high as 31 % in [Ba{N(SiMe 3 ) 2 } 3 ] 2À (see Supporting Information).This is in part reminiscent of sp 3 d and sp 3 d 2like hybridizations that have been previously invoked to describe deviation from linearity in the geometry of AE metallocenes and halides, [77][78][79][80][81] but have not been taken into consideration as a potential characteristic of monovalent AE species.
Following our computational analysis, we carried out several attempts to reduce 1-AE and 3-AE, however we have not been able to produce our target reduced species.In the case of 1-Ca, we performed attempted reductions with KC 8 either with 18crown-6 in THF, or in benzene without any sequestering agent (for the latter we used [Ca{N(SiMe 3 ) 2 } 3 K] as starting material, rather than 1-Ca).In both cases we recovered unreacted starting material and K[N(SiMe 3 ) 2 ], together with trace amounts of other unknown species that we could not identify.These results were obtained even under strict temperature control in the attempt to minimize decomposition reactions (see Supporting Information Figures S22 and S25).Similarly, when 3-Ca was reacted under the same conditions and 1 H NMR analysis reaction mixtures revealed the presence of unreacted starting material and K[N(Mes)(SiMe 3 )], together with other unknown species (see Supporting Information Figures S23 and S27).We also attempted the reductions of 1-Sr and 3-Sr obtaining similar results (Figures S24 and S26).Furthermore, we attempted the reduction of 4 in various conditions, but also in this case we either recovered unreacted starting material or obtained an intractable mixture of products (see Supporting Information Figures S28 and S29).
Reasons for the lack of success with these reductions could be ascribed to the reducing power of the reagents employed and also the incompatibility of the ligand systems with reductive chemistry, owing to the facile elimination of potassium amide by-products and the propensity of silylamide ligands to undergo facile degradations via intra-and intermolecular CÀ H activation. [40,82] Additionally, the ligands used in this work display various degrees of flexibility, which can affect the energy profile of the reductions and stability of the resultant reduced species.The stabilization of dianionic complexes as a result of these reductions (e.g.[AE{N(SiMe 3 )(R)} 3 ] 2À ) could also be deemed problematic in terms of charge distribution.Therefore, the use of neutral precursors would be highly desirable, as demonstrated by the more favourable reactivity profile of the reduction of [Sr{N(SiMe 3 ) 2 } 2 (dioxane) 2 ] n .Despite the lack of success in isolating monovalent AE complexes, our theoretical modelling based on trigonal planar and square planar precursors provides a blueprint for further exploitation in synthetic chemistry.Nonetheless, more suitable ligand systems should be employed to implement this blueprint, which needs to possess a relatively innocent redox profile, paired with sufficient conformational rigidity and steric protection to enforce highly equatorial geometries.

Conclusions
We have reported the synthesis of a series of trigonal planar, heavy AE(II) silylamide complexes; DFT calculations have been performed on these complexes and the theoretical AE(I) species that could form upon chemical reduction.We have demonstrated that Ca(II), Sr(II) and Ba(II) complexes with highly equatorial ligand environments (i.e. trigonal planar, square planar) have readily accessible d z2 orbitals, and that these should be occupied for their respective AE(I) analogues, providing feasible synthetic targets.Together, this work has presented a combined synthetic strategy and theoretical modelling that provide a blueprint for targeting sub-valent heavy AE species that do not form AEÀ AE bonds.[41] We are exploring ligand variations and reduction protocols, including different reducing agents and the use of solid-state reactions, in order to overcome the synthetic challenges that have hampered the isolation of heavy AE(I) complexes to date.

Experimental Section
General methods: THF and toluene were passed through columns containing molecular sieves, then stored over a potassium mirror (toluene), or over 4 Å molecular sieves (THF) and thoroughly degassed prior to use.Hexane and diethyl ether were purchased anhydrous from Tokyo Chemical Industry, dried over activated molecular sieves for 7 days, then stored over a potassium mirror.
[a] Computed Mg shown is based on 3-Mg with only two of the three Mesityl substituents of the amide ligands pointing in the same direction, with weak agostic Mg•••Me Si interactions.The LUMO has no notable interesting features and hence is not given.A conformation equivalent to 3-Ca (with all three Mesityl groups pointing in the same direction) is given in the Supporting Information and has a ΔG of À 16.8 kcal/mol.[b]Computed Ba structure obtained by removing THF from 3-Ba • (THF) (see Supporting Information, ΔG obtained from THF-free species to compare directly with MgÀ Sr entries.computational studies.Gratifyingly, MO analysis affords a LUMO with d-character, which was also confirmed by NBO calculations (d%: 1.3).The computed reduction of 4 to monovalent [Sr{N(SiMe 3 ) 2 } 2 (dioxane) 2 ] À (Scheme 4) is exergonic (À 26.1 kcal/ mol) and the resulting SOMO(α) also displays d-character (d%: 7.8, Table3 and SupportingInformation). Analogously to the [AE{N(R)(SiMe 3 )} 3 ] À radicals, the optimised geometry of [Sr {N(SiMe 3 ) 2 } 2 (dioxane) 2 ] À exhibits a slight pyramidalization (Sr•••N^O^N^O = 0.565 Å) compared to the structure of the divalent precursor 4 (Sr•••N^O^N^O = 0.056 Å, Table
H} experiments, or externally referenced to SiMe 4 ( 29 Si { 1 H}).FTIR spectra were recorded on a Bruker Alpha II spectrometer with Platinum-ATR module.Elemental microanalyses were carried out by the Elemental Analysis Service at London Metropolitan University.n BuLi was used as received.18-crown-6 was dissolved in anhydrous diethyl ether and dried over 4 Å molecular sieves for 7 days before removing the solvent in vacuo and drying under vacuum.[2.2.2]-cryptand was dried in vacuo for 4 h prior to use.All AE iodides were baked at 200 °C for 4 h prior to use.HN-(Mes)(SiMe 3 C{ 1 H} NMR (125 MHz, 298 K, C 6 D 6 ) δ/ppm = 7.36 (Si(CH 3 ) 3 ), 70.63 (OCH 2 ).No signals could be obtained from 29 Si{ 1 H} NMR experiments.