[U(III) {N(SiMe2 tBu)2 }3 ]: a structurally authenticated trigonal planar actinide complex.

We report the synthesis and characterization of the uranium(III) triamide complex [U(III) (N**)3 ] [1, N**=N(SiMe2 tBu)2 (-) ]. Surprisingly, complex 1 exhibits a trigonal planar geometry in the solid state, which is unprecedented for three-coordinate actinide complexes that have exclusively adopted trigonal pyramidal geometries to date. The characterization data for [U(III) (N**)3 ] were compared with the prototypical trigonal pyramidal uranium(III) triamide complex [U(III) (N")3 ] (N"=N(SiMe3 )2 (-) ), and taken together with theoretical calculations it was concluded that pyramidalization results in net stabilization for [U(III) (N")3 ], but this can be overcome with very sterically demanding ligands, such as N**. The planarity of 1 leads to favorable magnetic dynamics, which may be considered in the future design of U(III) single-molecule magnets.

Investigations into low-coordinate metal complexes (defined herein as coordination number, CN<4) are legion, because they can exhibit interesting properties,1 including small-molecule activation chemistry2 and single-molecule magnet (SMM) behavior.3 Low CN complexes usually contain sterically demanding ligands to prevent oligomerization,1 in which bulky monodentate amides are frequently utilized.4 The bulky silylamide {N(SiMe3)2}− (N“) has provided landmark low CN complexes; for example, three-coordinate [MIII(N”)3] complexes of Group 13 (M=Al, Ga, In, Tl)5 and first row d-block (M=Ti–Co)6 metals are trigonal planar (D3) in the solid state, but Group 3,6a, 7 lanthanide (Ln),7 and actinide (An)8 [MIII(N“)3] complexes exhibit trigonal pyramidal (C3) solid-state geometries, although they have zero dipole moment in solution, inferring that they may become planar in this phase.9 Pyramidal geometries persist for [LnIII(N”)3] (Ln=Ce, Pr) in the gas phase,10 but [ScIII(N“)3] vapors are D3, with crystalline/gas-phase discrepancies for this complex attributed to crystal-packing effects.11 It is noteworthy that complexes, such as [LnII(N”)(μ-N“)2Na] (Ln=Eu, Yb) and [SmII(N”)(μ-N“)2M] (M=Na, K), have trigonal planar Ln coordination spheres,[12] but this geometry has not been previously observed in An complexes.

f-Block metal centers favor high CNs, because Ln and An cations have relatively large ionic radii and bonding regimes that are dominated by electrostatic contributions.13 Low CN UIII chemistry is burgeoning, driven by interesting small molecule activation reactions[14] and intrinsic SMM behavior.15 Structurally characterized three-coordinate An complexes to date adopt exclusively trigonal pyramidal geometries rather than trigonal planar or T shaped (C2),16 although matrix isolation experiments[17] and calculations18 have shown that monomeric UO3 is T shaped. Both covalent[19] and electrostatic10 arguments account for the trigonal pyramidal geometry of [UIII(N“)3],8, 20 hence, the most influential factor of these two for causing pyramidalization has never been established. Herein, we report the structurally characterized An complex, [UIII(N**)3] (1, N**=N(SiMe2tBu)2−), which adopts an unprecedented trigonal planar geometry for an actinide triamide complex. Complex 1 is closely related to [UIII(N”)3], allowing the contributions to pyramidalization to be assessed, together with the impact of geometry on magnetic (including dynamic) and electronic properties of UIII complexes, for the future rational design of useful An materials.

Complex 1 was prepared by a modification of the revised synthesis of [UIII(N“)3].8c Compound [UIII(I)3(THF)4]8c was reacted with 1.5 equivalents of [K{N(SiMe2tBu)2}]2 in THF, followed by work-up and recrystallization from hexane to give 1 as dark purple needles in 62 % yield (Scheme 1).21 Absorbances in the FTIR spectrum of 1 at , 825, and 761 cm−1 are attributed to the UNSi2 stretching modes of the silylamide ligand. The asymmetric stretch (950 cm−1) is 40 cm−1 lower than that observed for [UIII(N”)3] (990 cm−1),8a which is of a similar magnitude to the differences between previously reported planar and pyramidal [M(N“)3] MNSi2 asymmetric stretches (ca. 50 cm−1).5b, 6a

Scheme 1

Synthesis of 1.

The 1H NMR spectrum of 1 exhibits two resonances at δ=3.8 (ν½=206 Hz) and −47.0 ppm (ν½=4597 Hz) in a 54:36 ratio that are assigned to the tBuSi and Me2Si protons, respectively. The Me2Si resonance of 1 is much broader than the analogous resonance for [UIII(N“)3] (δ −11.4, ν½=15 Hz),8 but variable-temperature (VT) studies gave a sharper resonance at 353 K (δ=−32.9 ppm, ν½=266 Hz).21 A wide-scan 13C NMR spectrum of 1 exhibited two resonances for the Me2Si (δ=−2.1 and 1.5 ppm) and tBuSi quaternary carbons (δ=18.2 and 32.0 ppm), but only one for the tBuSi primary carbons (δ=26.4 ppm). In contrast, in the 13C NMR spectrum of [UIII{N(SiPhMe2)2}3], the Me2Si group resonates at δ=−57.1 ppm.22 A resonance was observed in the 29Si NMR spectrum of 1 at δ −296.0 ppm (ν½=73 Hz), which has not been reported for similar systems,8, 22 but is typical for a UIII complex.23

The electronic absorption spectrum of 121 exhibited 5f3→5f26d1 transitions at 20 000 (ε=776 m−1 cm−1) and 22 500 cm−1 (ε=770 m−1 cm−1) that are typical of UIII[24] and comparable to a broad absorption observed for [UIII{N(SiPhMe2)2}3] at 21 500 cm−1 (ε=430 m−1 cm−1).22 In the 7 000–13 000 cm−1 region, weak Laporte forbidden 5f→5f transitions were observed (ε=15–64 m−1 cm−1).25 Similar weak absorptions were observed for most UIII complexes, such as [U(I)3(THF)4]8c, 26 and [UIII{N(SiPhMe2)2}3],22 and strong absorptions in this region are very rare.[27]

The crystal structure of 1 was determined and is depicted in Figure 1, with selected metrical parameters.28 Complex 1 crystallizes in the C2/c space group, with a twofold axis bisecting the U(1)–N(1) bond. This contrasts to [Fe(N“)3],9 [EuIII(N”)3],29 [UIII(N“)3],8d and [PuIII(N”)3],8e which all crystallize exclusively in the P31c space group, and [UIII{N(SiPhMe2)2}3], which crystallizes in R3.22 The U atom of 1 is almost ideally trigonal planar, with U–N bonds that are statistically identical within experimental uncertainty [U-N range 2.403(3)–2.415(6) Å]. These distances are longer than those observed in [UIII(N“)3] [2.320(4) Å]8d and [UIII{N(SiPhMe2)2}3] [2.34(2) Å],22 which can be attributed to the greater interligand repulsion in 1 arising from the sterically demanding tBu groups. The U centroid/N(1)-N(2)-N(2A) mean plane distance in 1 is 0.008(2) Å, and the N-U-N bond angles (range 119.1(2)–120.47(9)°) sum to 360°; in contrast, [UIII(N”)3] and [UIII{N(SiPhMe2)2}3] exhibit U centroids 0.456(1) and 0.874 Å from the N3 planes, and the N-U-N angles average 116.24(7) (Σ angles 348.72(7)°) and 106.88° (Σ angles 320.64°), respectively.8d, 22 The UNSi2 fragments of 1 are essentially planar and all bisect the UN3 plane (range 53.23–61.35°) to form a molecular propeller.

Figure 1

Molecular structures of 1 a) top view and b) along twofold axis, with selected atom labelling. Displacement ellipsoids are set at the 40 % probability level, and hydrogen atoms are removed for clarity. Selected bond lengths [Å] and angles [°]: U(1)–N(1) 2.403(3), U(1)–N(2) 2.415(6); N(1)-U(1)-N(1′) 119.05(19), N(1)-U(1)-N(2) 120.47(9).

The pyramidal geometries of [UIII(N“)3] and [UIII{N(SiPhMe2)2}3] are predicted by the polarized-ion model, whereby net stabilization was achieved by dipole formation.8d, 22 [UIII(N”)3] exhibits unequal U-N-Si angles (108.50(7) and 125.25(7)°), because one Si–C bond for each N“ ligand is relatively close to the U center [U⋅⋅⋅Cγ 3.05 Å; U⋅⋅⋅Si 3.29 Å].8d These can be attributed to stabilizing agostic M⋅⋅⋅Si–Cγ interactions, as have been discussed for [UIII{CH(SiMe3)2}3]30 and [SmIII(N”)3].[31] The shortest U⋅⋅⋅Cγ and U⋅⋅⋅Si distances in 1 are 3.119–3.301 Å and 3.433–3.510 Å, respectively, and they are not correctly orientated to interact with the U center. Although there is no evidence for agostic U⋅⋅⋅Si–Cγ interactions in 1, stabilizing U⋅⋅⋅C–H contacts cannot be discounted.

Unrestricted DFT calculations were carried out on full models of 1 and [UIII(N“)3].21 The geometry-optimized structures reproduce the experimental structures with good agreement, despite the slight deviation from planarity for the model of 1 (discrepancies attributed to this being a gas-phase calculation, which does not account for crystal-packing forces), providing qualitative models (bond lengths within 0.05 Å, angles within 1°, U centroid/N3 mean plane distance: 1 0.132 Å, [UIII(N”)3] 0.393 Å). In both models, the HOMO, HOMO−1 and HOMO−2 represent the three unpaired UIII 5f electrons (1: 93.93, 94.71, 90.09; [UIII(N“)3] 86.81, 86.32, 84.17 % U 5f, respectively). Both models exhibit essentially insignificant degrees of U 6d/5f orbital contributions to the U–N bonds, with the HOMO−3, HOMO−4. and HOMO−5, representing the π components (1: 5.27/0, 1.57/0, 0/1.31; [UIII(N”)3] 4.29/0, 0/2.06, 1.63/1.39 % U 5f/6d, respectively) and the HOMO−6, HOMO−7, and HOMO−8 the σ components (1: 0/2.29, 0/2.12, 1.20/0; [UIII(N“)3] 0/5.04, 0/5.26, 2.14/0 % U 5f/6d, respectively). This concurs with gas-phase photoelectron spectroscopy (PES) studies of [U(N”)3], which have shown that π bonding between the ligand and U center is insignificant in this complex.32 The calculated uranium spin densities (MDC-m α spin, 1=−3.26; [UIII(N′′)3]=−3.26) are identical, which also supports similar bonding patterns for 1 and [UIII(N“)3].

Ab initio calculations on [AnIII(CH3)3] (An=U, Np, Pu)33 and [AnIII(NH2)3] (An=U, Np)34 have shown that the involvement of An 6d orbitals in the U–X (X=C, N) σ components may be associated with pyramidalization in the absence of steric contributions. Thus, given the similar bonding within 1 and [UIII(N“)3] together with the small U 6d/5f contributions to the U–N σ and π components, we suggest that the experimentally determined trigonal planar geometry of 1 results from steric interactions involving the large N** ligands. These interactions could predominate over crystal packing forces, which are often only approximately 10 kJ mol−1.35 We conclude that there are minor differences in bonding between 1 and [UIII(N”)3], therefore, the planar geometry of 1 derives principally from steric effects involving the ligands.

The solution magnetic moment of 1 was calculated to be 2.59 μB in [D6]benzene at 298 K by using the Evans method.[36] Magnetometry measurements on a powdered sample of 1 suspended in eicosane gave a magnetic susceptibility temperature product, χT, of 1.07 cm3 Kmol−1 (2.92 μB) at 298 K,21 which corresponds well with the solution measurement considering weighing errors and the difference in phase. These values are lower than for a free-ion 5f3 4I9/2 ground state (3.69 μB), because not all crystal field levels are thermally occupied,37 but are typical for UIII complexes described in the literature (range 2.13–4.63 μB).8, 15, 22, 25, 26, 30, [38] The χT value of 1 decreases to 0.41 cm3 Kmol−1 at 2 K; ac measurements give a low-temperature plateau in the in-phase χ′T at 0.48 cm3 Kmol−1[21] consistent with thermal depopulation into a Kramers doublet ground state.3, 13 Low-temperature EPR spectra of 1 are consistent with UIII,[27] and simulation gives geff=3.55, 2.97, and 0.553 for the ground Kramers doublet (the latter is observed at high field at X-band, but is beyond the magnetic field range at Q band; Figure 2 a).

Figure 2

a) X- (9.5 GHz) and Q-band (34 GHz; inset) EPR spectra of 1 at 5 K. Lower spectra are simulations as Seff=1/2. Magnetic-susceptibility data for 1: b) magnetic hysteresis at 1.8 K, sweep rate 13 G s−1; c) in-phase (χ′); and d) out-of-phase (χ“) components of the ac susceptibility measured in an applied dc field of 600 G and an oscillating field of 1.55 G.

Compound [UIII(N“)3] is an SMM,15 hence, we have performed low-temperature ac measurements on 1 to probe differences in the dynamic magnetic behavior as a result of the higher symmetry. Compound 1 is also an SMM, with clear frequency-dependent behavior (Figure 2 c and d).21 Under the optimal dc field of 600 G, the magnetization relaxes much slower than in [UIII(N”)3], and maxima in the out-of-phase susceptibility χ′′(T) are seen to significantly higher temperatures for 1 than for [UIII(N“)3] at equivalent frequencies (e.g., 3.5 vs. 2.1 K, respectively, for 1.4 kHz). An Arrhenius treatment21 of the higher-temperature ac data gives an energy barrier of Ueff=21.4±0.2 K for 1. Although this is lower than that reported for [UIII(N”)3] (31 K), the latter value was derived from an extremely limited temperature range15 and should be treated with some caution. The relaxation time (τ) at 2 K is 2.6 ms for 1; from the previously reported data15 we find 0.3 ms for [UIII(N“)3] at 2 K, an order of magnitude quicker. The pre-factor τ0 for 1 is greater by four orders of magnitude (3.1×10−7 cf. 10−11 s for [UIII(N”)3]).15 Moreover, the frequency dependence of χ′ and χ“ at 1.8 K for 121 reveal a single relaxation process with a narrow distribution in relaxation times (α=0.001–0.03 from Cole–Cole analysis), an order of magnitude lower than in [UIII(N”)3] (α=0.09–0.34).15 In fact, the difference in dynamics is sufficient that magnetization hysteresis is observed for 1 at 1.8 K on a conventional superconducting quantum interference device (SQUID) magnetometer (Figure 2 b), while it is not for [UIII(N“)3].

In the trigonal planar geometry of 1, with no axial ligands, we expect a low J state of UIII to be stabilized by the crystal field. This is supported by the EPR analysis: if we assume a 4I9/2 ground term,39 with gJ=8/11, the J=±1/2 doublet is calculated to have g=3.65, g=0.73 (all other doublets have g=0), in good agreement with experiment. |J|=1/2 is also the ground doublet of the (pyramidal) 4f3 complex [NdIII(N“)3] from optical studies.40 Hence, 1 and [UIII(N”)3] are SMMs despite their easy-plane anisotropy: this highlights the complexity of interpreting f-block relaxation data,41 particularly when relatively low (tens of K) energy barriers are involved. At this stage, we can speculate that the “cleaner” and slower relaxation of 1 compared with [UIII(N“)3] on flattening the geometry is because of quenched mixing. In D3 |J|=1/2 cannot mix with any other doublet within the 4I9/2 term, whereas in C3, it can mix with both |J|=5/2 and 7/2.

To conclude, we have prepared and fully characterized an unprecedented trigonal planar actinide triamide complex. Differences in the spectroscopic and magnetic data between 1 and [UIII(N“)3] can be attributed to differences in symmetry that may be useful to consider in the future design of UIII SMMs with greater relaxation times. Computational analyses of 1 and [UIII(N”)3] have shown only minor differences in their calculated bonding schemes, therefore, the energy gained by pyramidalization, which leads to favorable agostic M⋅⋅⋅Si–Cγ interactions in [UIII(N“)3],8d, 32, 33 can be overcome by sterically demanding ligands, such as N**.

Experimental Section

Synthesis of 1: THF (20 mL) was added to a precooled (−78 °C) mixture of [K{N(SiMe2tBu)2}]2 (1.007 g, 1.5 mmol) and [U(I)3(THF)4] (0.907 g, 1 mmol). The reaction mixture was allowed to warm to RT slowly with stirring over 48 h, with precipitation of a pale solid. Volatiles were removed in vacuo, and the dark purple solid was extracted with hexanes (3×10 mL). Recrystallization from hexanes (5 mL) at −30 °C gave 1 as dark purple needles (0.605 g, 62 %).1H NMR (400.13 MHz, [D6]benzene, 25 °C, TMS): δ=−47.04 (br s, ν½=4597 Hz, 36 H; Si(CH3)2), 3.79 ppm (br s, ν½=206 Hz, 54 H; SiC(CH3)3); 13C{1H} NMR (100.61 MHz, [D6]benzene, 25 °C, TMS): δ=−2.13 (Si(CH3)2), 1.45 (Si(CH3)2), 18.22 (SiC(CH3)3), 26.40 (SiC(CH3)3), 31.98 ppm (SiC(CH3)3); 29Si{1H} NMR (79.48 MHz, [D6]benzene, 25 °C, TMS): δ=−296.04 ppm (br. s, ν1/2=73 Hz); FTIR (Nujol); (s), 1247 (s), 1002 (s), 950 (m, asym. str., UNSi2), 825 (s, sym. str., UNSi2), 761 (s, sym. str., UNSi2), 655 (m), 604 (s) cm−1; μeff=2.59 μB (Evans method); elemental analysis calcd for C36H90Si6N3U (971.67 g mol−1): C 44.5, H 9.34, N 4.33; found: C 38.29, H 9.10, N 4.22. Low carbon values were obtained upon repeating the analysis multiple times on different batches and is ascribed to 1 being a silicon-rich molecule, as was observed previously.42