Edinburgh Research Explorer Thorium Mono- and Bis(imido) Complexes Made by Reprotonation of cyclo-Metalated Amides

: Molecules containing actinide − nitrogen multiple bonds are of current interest as simple models for new actinide nitride nuclear fuels, and for their potential for the catalytic activation of inert hydrocarbon C − H bonds. Complexes with up to three uranium − nitrogen double bonds are now being widely studied, yet those with one thorium − nitrogen double bond are rare, and those with two are unknown. A new, simple mono(imido) thorium complex and the ﬁ rst bis(imido) thorium complex, K[Th(  NAr)N ″ 3 ] and K 2 [Th(  NAr) 2 N ″ 2 ], are readily made from insertion reactions (Ar = aryl, N ″ = N(SiMe 3 ) 2 ) into the Th − C bond of the cyclometalated thorium amides [ThN ″ 2 (N(SiMe 3 )-(SiMe 2 CH 2 ))] and K[ThN ″ (N(SiMe 3 )(SiMe 2 CH 2 )) 2 ]. X-ray and computational structural analyses show a “ transition-metal-like ” cis -bis(imido) geometry and polarized Th  N bonds with twice the Wiberg bond order of the formally single Th − N bond in the same molecule.

S trong and covalent multiple bond formation between actinide cations and oxo groups is a ubiquitous feature of actinide chemistry, most famously represented by the linear OUO unit in the uranyl dication [UO 2 ] 2+ , in which the ligating atoms strongly donate electron density to the metal through one σ bond and two π bonds in an unreactive and formally triple bond. 1 Studies of multiple bonds between actinide ions and other heteroatoms are still rare 2 but are gaining momentum due to their unique contributions to our understanding of the role of the 5f and other orbitals in covalent actinide bonding. 3 Imido ligands are strong σand π-donors, and uranium imido bonds have a formal order of 3, like uranyl oxos. All known U VI uranyl complexes contain trans-dioxo ions, and nearly 90% of U VI bis(imido) complexes have a close to linear (>150°) geometry; examples of cis-and trans-U VI bis(imido) complexes respectively are (Cp*) 2 U(NR) 2 (A in Figure 1) and the landmark, linear uranyl analogue [U(NR) 2 I 2 (THF) 2 ] (B). 4 Meanwhile, the majority (ca. 95%) of the numerous transition metal bis(imido) complexes exhibit cis geometries (e.g., C in Figure 1).
Recent years have seen a great increase in the number and variety of terminal imido uranium complexes, including bis-(imido) and most recently U VI tris(imido) structures (D in Figure 1). 5 Their bonding has been studied in terms of proving molecular analogues of actinide nitride nuclear fuels, 6 and their reactivities have been studied in the search for catalytic properties which might differ from those of the d-block complexes. 7 An intramolecular ligand C−H bond addition reaction was initiated by diazoalkane binding to a U VI NR bond in the complex (C 5 Me 5 ) 2 U(NC 6 H 2 t Bu 3 )(N 2 CPh 2 ), 8 the strong donation of the N(2p) orbitals and involvement of the uranium 5f orbitals is not conducive to the most desirable intermolecular hydrocarbon C−H bond addition chemistry, for which the more polar transition metal MN bonds have been celebrated. 9 Only four terminal thorium imido complexes have been reported, all of which are supported by the bulky, clam-shaped bis(cyclopentadienyl) co-ligand set (E in Figure 1). 10 Early reports on the small-molecule activation reactivity of E include silane Si−H bond cleavage chemistry. 11 This supports the suggestion of more d-block-like behavior of thorium (cf. uranium), which is indicated by its ground-state [Rn]s 2 d 2 configuration, and a therefore greater potential for the reactivity of the ThNR group. No bis(imido) complexes have been reported to date, which could be due to their anticipated higher reactivity compared to the uranium analogues (calculations on the model [(C 5 H 5 ) 2 Th(NPh) 2 ] 2− ion find much weaker metal−nitrogen bonding than in the uranium complex), 12 and also to the difficulty in stabilizing low coordination numbers at the large Th IV center. 13 Given that A has no equivalent in transition-metal chemistry, we were interested in thorium imido complexes of monodentate, monoanionic ligands which would not be sterically protected or geometrically directed by the "clam-shell" configuration of the bis(cyclopentadienyl) framework. Herein we report a new, straightforward route to amido-imido Th IV complexes and the first thorium bis(imido) complex. We also discuss their bonding and geometries with the support of density functional theory (DFT) calculations.
Treatment of A with KNHDipp (Dipp = 2,6-i Pr 2 C 6 H 3 ) in C 6 H 6 /THF or THF provides straightforward access to the colorless crystalline thorium imide [K(S) x ][Th(NDipp)N″ 3 ] (2, Scheme 1), isolable as a variety of solvates, depending on the workup procedure. 15 For example, from THF solution unsolvated 2a (S = 0) is isolated in quantitative yield, but addition of crown ether (18-c-6) and a small quantity of THF (10% by volume) to the benzene reaction mixture affords the separated ion-pair [K (18-c-6)][Th(NDipp)N″ 3 ] (2b). Complex 2 is highly oxygen and moisture sensitive but thermally stable for at least 14 days in a refluxing toluene/THF solution (ca. 10% THF). NMR spectroscopy and X-ray diffraction (XRD) studies (see below) of 2 confirm the addition of the [HNDipp] group across the Th−C bond to form the ThNDipp aryl imido complex. No NH resonance is observable in the spectra, and in the aliphatic region a lone singlet at δ 0.28 (54H) confirms the equivalence of the two original silylamide ligands and regenerated N″ amido ligand. Finally, 29 Si-INEPT NMR spectra show a single silicon environment with a chemical shift of −11.2 ppm. A number of other arylimido analogues of 2 were also made and are discussed in the Supporting Information (SI).
Analogously, treatment of 1a with 2 equiv of KNHDipp in a C 6 H 6 /THF mixture (ca. 5:1 ratio) yields the first bis(imido) complex of thorium, [K 2 (S) x ][Th(NDipp) 2 N″ 2 ] (3), as a colorless crystalline solid in essentially quantitative yield (Scheme 1). Again, different solvated forms can be isolated, but the simple THF adduct (3a, with S = THF, x = 2, 85% yield) most readily forms single crystals suitable for XRD studies (see Figure 2). The mono-and bis-18-c-6 analogues 3b and 3c are described in the SI, section IVc. The complex is mononuclear in solution as well as in the solid state: 1 H NMR spectra of 3a in C 6 D 6 , solubilized with a few drops of THF, contain a singlet at δ 0.26 (36H), demonstrating a single silylamide environment. 1 H DOSY NMR spectroscopic experiments (SI, section V) show the hydrodynamic radius of 3a (7.1 Å) to be similar to that of 2a (6.9 Å), suggesting that 3 remains monomeric in solution. The highest mass peak in the electron ionization mass spectrum of 3c is at 1509.2 m/z, corresponding to the ion-pair [K (18-c-6)] 2 [Th( NDipp) 2 N″ 2 ]. The MALDI spectrum clearly shows a molecular ion peak at 901.5 m/z, corresponding to [Th(NDipp) 2 N″ 2 − H] − . No ion with higher mass was observed by either method.
Once isolated, 3 is stable for several days in solution, but if stored in contact with the soluble KN″ byproduct, over 8 h at room temperature solutions begin to degrade to multiple products, as determined by 1 H NMR spectroscopy. Other bis(imido) analogues were also targeted, but none was as stable as 3; see SI, section IV for details.
X-ray-quality crystals of 1c were grown by slow cooling of the hexane filtrate of a sample made from ThCl 4 (DME) 2 . X-rayquality crystals of 2b were grown by slow cooling of a C 6 D 6 /THF solution, while single crystals of 3a were grown by slow diffusion of hexane into a C 6 H 6 /THF solution. The molecular structure of 1c (Figure 2a) confirms the metalation of both amido ligands. Although the solid-state structure of A is not known, the two Th− C bonds of 2.562(5) and 2.576(5) Å fall within the expected range for single Th−C bonds (2.4−2.8 Å).
The molecular structure of 2b ( Figure 2b) shows a tetrahedral thorium environment with a Th1−N1 imido bond length of 2.072(3) Å that is significantly shorter than a single Th−N amido bond (range 2.2−2.4 Å for thorium dialkylamides) but is one of the longest Th−N imido bonds observed to date. The four reported terminal thorium imido complexes E all contain two cyclopentadienyl co-ligands, either [(C 5 Me 5 ) 2 ] 10b or

Journal of the American Chemical Society
Communication [(C 5 H 2 ( t Bu) 3 -1,2,4) 2 ]. 11a It is likely that the Th−N lengthening in 2 reflects the electron richness of the metal center due to both the formal negative charge on the Th IV center and the presence of electron-donating silylamide ligands, each of which has been shown to result in a measurable MN lengthening in d-block analogues. 16 The structure of 3a ( Figure 2c) shows a pseudo-tetrahedral thorium environment and a cis arrangement of the two imido ligands with a N4−Th1−N3 bond angle of 97.34(12)°, close to that which has been theoretically predicted (see below). The two thorium imido Th−N bonds (both 2.165(3) Å) are longer than that in 2b (2.072(3) Å) but significantly shorter than the thorium amido Th−N bonds (Th1−N1, 2.448(3) Å and Th1−N2, 2.474(3) Å). Both Th−N imido −C Dipp angles are also very close to linear, as would be anticipated for a Th−N bond order of 2 or 3 (C13−N3−Th1, 169.1(3)°and C25−N4−Th, 1171.4(3)°). One of the two potassium counter-cations (K1) occupies the gap between the two imido groups, with relatively close contacts to both the nitrogen and ipso-aryl carbon atoms of each imido group, a feature observed previously in related uranium systems. 7j The other, K2A/K3 (disordered over two sites), is weakly bound in an approximately bis(arene) sandwich-type geometry through coordination to the imido aryl rings of adjacent molecules, resulting in the formation of one-dimensional chains in the solid state.
The mono-(2) and bis(imido) 3 structures were studied using theoretical methods (DFT) on model compounds 2′ and 3′, with potassium counterions omitted (see SI, section VIII for details and further analysis of the data). The computational method reproduced the geometry of 2 and 3 very well, indicating the suitability of the chosen method. Analysis of the bonding in 2, using NBO, indicates double-bond character in the Th−N imido bond in 2 at the second-order donor−acceptor level, with a polarized covalent σ bond and an ionic π interaction involving donation from the nitrogen 2p orbital toward an empty 6d/5f hybrid orbital on Th. There is a greater polarization than the bonding in the recently reported ThO analogue, as expected, 17 and a greater d orbital involvement than f, in line with previous studies by us. 18 In 3′, these double bonds can be described as a polarized covalent σ bond (90% 2p N; 10% 6d(65)-5f(35) Th) and an ionic π interaction involving donation from the nitrogen 2p orbital toward an empty 6d orbital on Th (second-order NBO, 44.2 kcal/mol). Figure 3 shows the calculated HOMO and HOMO− 1, which are the Th−N imido bonds. Thus, the Th−N imido bonds in both mono-and bis(imido) compounds are correspondingly more polarized than in a typical uranium analogue. 18 The Wiberg bond indices are 0.88 in 3, twice the indices of the Th−N amido bonds, which are 0.47, but significantly lower than 2.0, which would represent a pure covalent double bond.
It is gratifying to find that both the predicted and found bis(imido) geometries are cisoid, 117°and 97.34(12)°(for 3c). The difference of ca. 20°is presumably due to the weak interaction between one potassium countercation and the two imido N atoms that is likely to exaggerate the cis geometry. Optimizations from both cis-and trans-imido starting points converged to the 117°cis geometry, underlining the relative instability of the trans.

Journal of the American Chemical Society
Communication thorium complex, [K 2 (S) x ][Th(NDipp) 2 N″ 2 ] (3), are both readily accessible from the simple, well-known thorium amidometallacycle, Th{N(Si′)(SiMe 2 CH 2 )}N″ 2 (A) and a doubly metalated analogue, K[Th{N(Si′)(SiMe 2 CH 2 )} 2 N″] (1). These are the first thorium-imido-containing complexes to be isolated without the support of the sterically constraining, clam-shellshaped bis(cyclopentadienyl) ligand set. The ThN bonds are relatively long in both 2 and 3, presumably reflecting the electronrichness of the metal center compared to those in the bis(cyclopentadienyl)-supported examples and the formal anionic charge associated with the "ate" nature. Our inability to synthesize more bulky analogues, or to isolate less bulky analogues, suggests that steric protection is a key enabler in the isolation of the bis(imido) unit, in agreement with calculations that suggest the thorium imido bonds are more polarized than their uranium congeners. The ThN bond orders calculated by DFT (Wiberg) methods are twice the value of the single Th−N amido bonds in the same molecule, suggesting a much more polar ThN bond than compared with uranium analogues, more like some of the most reactive early d-block MN complexes. Thus, both experiments and calculations suggest the potential for reactivity closer to the transition-metal imido complexes that can activate hydrocarbon C−H bonds. Work is in progress to explore their capacity for new small-molecule activation chemistry.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.5b06630. Crystallographic files are also deposited in the CCDC, numbers 1062064−1062066. Experimental raw data are openly available through the University of Edinburgh's DataShare archive, DOI:10.7488/ds/289. X-ray crystallographic data for 1c, 2b, and 3a (CIF) Experimental procedures and characterization data for the real and computed complexes (PDF)

■ AUTHOR INFORMATION
Corresponding Author *polly.arnold@ed.ac.uk Notes