Rare-Earth- and Uranium-Mesoionic Carbenes: A New Class of f-Block Carbene Complex Derived from an N-Heterocyclic Olefin.

Neutral mesoionic carbenes (MICs) have emerged as an important class of carbene, however they are found in the free form or ligated to only a few d-block ions. Unprecedented f-block MIC complexes [M(N'')3 {CN(Me)C(Me)N(Me)CH}] (M=U, Y, La, Nd; N''=N(SiMe3 )2 ) are reported. These complexes were prepared by a formal 1,4-proton migration reaction when the metal triamides [M(N'')3 ] were treated with the N-heterocyclic olefin H2 C=C(NMeCH)2 , which constitutes a new, general way to prepare MIC complexes. Quantum chemical calculations on the 5f3 uranium(III) complex suggest the presence of a U=C donor-acceptor bond, composed of a MIC→U σ-component and a U(5f)→MIC(2p) π-back-bond, but for the d0 f0 Y and La and 4f3 Nd congeners only MIC→M σ-bonding is found. Considering the generally negligible π-acidity of MICs, this is surprising and highlights that greater consideration should possibly be given to recognizing MICs as potential π-acid ligands when coordinated to strongly reducing metals.

Over the past three decades the field of stable singlet N‐heterocyclic carbenes (NHCs, I, Scheme 1) has become a burgeoning area.1 Within that time, a variety of experimentally viable classes of carbenes related to I have emerged, including anionic‐NHCs (II),2 cyclic alkylaminocarbenes (CAAC, III),3 and various charge‐neutral mesoionic carbenes (MIC, IV–VI),4 Scheme 1, where for the latter no reasonable canonical resonance forms can be drawn without assigning additional formal charges. A growing number of MICs of type IV are known, but it is notable that all examples to date pertain to either the free carbene, or were formed at and remain coordinated to surprisingly few transition metal ions,4 which contrasts to NHCs that have been coordinated to the majority of metals in the periodic table.1 Where the bonding of these MICs to metals is concerned, complexes are usually considered to have strong MIC→metal σ‐donation. Given that strong σ‐donation, it is surprising that MIC complexes are limited to even only a few transition metals, but this may reflect the limited range of methodologies to deliberately prepare metal‐MIC complexes. Interestingly, any π‐bonding components of metal‐MIC bonds are, unlike NHCs, rarely explicitly considered.5 This is likely because MICs are anticipated to have at best weak π‐acceptor character since the carbene is strongly stabilized by N‐lone pair and vinyl groups, as evidenced by computational comparisons of different classes of carbenes.6

Scheme 1

Prominent classes of NHC‐type 5‐membered‐ring carbenes.

We report herein the synthesis and characterization of rare‐earth–MIC and uranium–MIC complexes, which are the first f‐block‐MIC complexes so by definition a new class of f‐block‐carbene complex.7 The complexes reported herein were prepared by the formal 1,4‐proton migration of an N‐heterocyclic olefin (NHO) that represents a new, general method by which to prepare MIC complexes. Interestingly, quantum chemical calculations suggest that the 5f3 uranium(III) ion engages in a weak π‐back‐bond to the MIC utilizing a 5f electron, whereas the corresponding d0f0 yttrium(III) and lanthanum(III), and 4f3 neodymium(III) benchmarks do not. Considering the generally negligible π‐acidity of MICs,6 this result for uranium is surprising, and highlights that perhaps greater consideration should be given to more widely recognizing MICs as potential π‐acid ligands when coordinated to sufficiently reducing metals.

As part of our continued studies of f‐element–carbon multiple bonding,8 we examined the reactivity of the uranium(III)–triamide complex [U(N′′)3] (1U, N′′=N(SiMe3)2)9 with the NHO H2C=C(NMeCH)2 (2).10 We postulated that an adduct similar to [Nd(N′′)3{H2C=C(NMeCMe)2}] could form,11 which in one resonance form can be represented as a NHC‐protected methylidene, but also that as well as being nucleophilic the methylene groups of NHOs are basic by virtue of the dipolar resonance form H2C−‐C+(NRCH)2.12 Given the existence of MICs, we considered whether 2 could be converted by transfer of an olefinic hydrogen atom to the methylene group in a formal 1,4‐proton shift to give its MIC form with concomitant metal stabilization.

Experimentally, we find that reaction of 1U with 2 gives [U(N′′)3{CN(Me)C(Me)N(Me)CH}] (3U), Scheme 2.13 The crystalline yield of 3U (27 %) is low, but is due to the high solubility of this complex since inspection of the mother liquor by NMR spectroscopy shows that 3U is the major product. How 3U forms is unclear, as no intermediates could be observed in low‐temperature NMR studies. A formal, concerted 1,4‐proton shift might be promoted by 1U, or more likely 1U might C4‐deprotonate 2 and the resulting N′′H could reprotonate the putative [U(N′′)2{CN(Me)C(CH2)N(Me)CH}] at the basic methylene group to re‐establish the third uranium‐amide bond and restore overall charge neutrality to the MIC. This is credible because reactions conducted in D6‐benzene show no evidence for D‐incorporation into 3U. To test the generality of this new reaction, we tested the reactivity of [M(N′′)3] (M=Y, 1Y; La, 1La; Nd, 1Nd) with 2 since the first two are closed‐shell d0f0 analogues and the latter is a 4f3 congener to 5f3 3U. Remarkably, all consistently give the MIC complexes [M(N′′)3{CN(Me)C(Me)N(Me)CH}] (M=Y, 3Y; La, 3La; Nd, 3Nd), isolated in crystalline yields of 23–30 %, Scheme 2.13

Scheme 2

Synthesis of the f‐block‐mesoionic carbene complexes 3M (M=U, Y, La, Nd) from the N‐heterocyclic olefin 2 and the metal triamides 1M.

The molecular structures of 3U, 3Y, 3La, and 3Nd were determined by single‐crystal X‐ray diffraction (Figure 1).14 In multiple crystals, the MIC is disordered, but modeled reliably, over three positions for 3U, two for 3La and 3Nd, but it is ordered for 1Y, which can be related to the size of the metal and pocket that the MIC sits in. There are no other f‐block‐MICs for comparison, and few f‐block‐NHCs have a sterically comparable profile to 3U, 3Y, 3La, and 3Nd, but for example the U−Ccarbene distances in 3U (2.576(12)–2.598(11) Å), can be compared to the longer U−Ccarbene distance in [U(N′′)3{C(NMeCMe)2}] [2.672(5) Å],15 and the U−Cylide bond length of 2.686(6) Å in [U(N′′)3(H2CPPh3)].16 Likewise, the M–MIC distances in 3Y (2.495(7) Å), 3La (2.675(14)/2.699(5) Å), and 3Nd (2.614(12)/2.620(11) Å) are consistently short when compared to respective NHC congeners.17 The MIC ring bond distances are intermediate to single and double bonds, suggesting delocalization. The U−N distances in 3U span 2.359(7)–2.381(7) Å, which suggests the uranium ion retains a+3 oxidation state. For comparison, U−N distances in 1U,9 [U(N′′)3{C(NMeCMe)2}],15 and [U(N′′)3(H2CPPh3)]16 average 2.320(4), 2.362(3), and 2.364(9) Å, respectively, whereas for [U(N′′)3(I)]18 they are 2.238(4) Å. The M−N distances in 3Y, 3La, and 3Nd are also consistent with their MIII natures.

Figure 1

Molecular structure of 3U at 150 K with ellipsoids set to 50 % probability. Hydrogen atoms and disorder components are omitted for clarity. The structures of 3Y, 3La, and 3Nd are very similar.14

NMR spectra of 3U, 3Y, 3La, and 3Nd are consistent with their MIII formulations,19 and a doublet at 173 ppm in the 13C{1H} spectrum of 3Y (J YC=56 Hz, 89Y; I= , 100 %)20 and absence of any carbene resonance for 3La due to the quadrupolar lanthanum (139La; I=7/2, 99.9 %) suggests that the MICs remain ligated in solution for all complexes.

The UV/Vis/NIR spectrum of 3U exhibits broad absorptions in the region 14 800–21 000 cm−1 (ϵ=530–870 L mol−1 cm−1) characteristic of 5f36d0 → 5f26d1 transitions of uranium(III) along with weaker (ϵ<180 L mol−1 cm−1) absorptions in the NIR region are observed.8j, 9b This is similar to that of 1U,9 but distinct to [U(N′′)3(I)].18

The uranium(III) assignment of 3U was further confirmed by SQUID magnetometry and EPR spectroscopy (Figure 2). Powdered 3U exhibits a magnetic moment of 2.82 μB at 300 K (3.31 μB in solution at 298 K), Figure 2 (Left). This is lower than the theoretical magnetic moment of 3.62 μB for a uranium(III) ion (4I9/2 ground spin–orbit multiplet, g=8/11), owing to crystal/ligand field splitting, and is generally in‐line with uranium(III) magnetic moments.21 Characteristic of uranium(III), the magnetic moment of 3 decreases slowly across the entire temperature range, reaching 2.12 μB at 2 K. Furthermore, low‐temperature magnetization data saturate at moderate magnetic fields, consistent with the Kramers nature of the uranium(III) ion, Figure 2 (middle). For comparison, we re‐measured data for 1U,9, 15, 21 and [U(N′′)3(I)],18 giving data consistent with literature values. Importantly, 1U has a similar low‐temperature magnetization profile to 3U, while that of the non‐Kramers [U(N′′)3(I)] fails to saturate up to 7 T. Further support for the uranium(III) oxidation state of 3U comes from low temperature (20 K) EPR spectroscopy at 9.5 GHz (Figure 2, right), where a typical uranium(III) spectrum22 is observed from a powdered sample with effective g‐values of g=4.65, 1.33, and 0.89 arising from the ground Kramers doublet. The non‐Kramers uranium(IV) ion would be expected to be EPR‐silent under these conditions. An isolated Kramers doublet with these g‐values corresponds to a magnetic moment of 2.46 μB, in fair agreement with the experimental magnetic moment of 2.27 μB observed at 20 K. Furthermore, these g‐values and magnetic moments are in reasonable agreement with those determined with CASSCF‐SO calculations,13 which predict g=4.3, 2.4, and 0.7 for the ground Kramers doublet (weighted for the crystal structure MIC disorder; compare to calculated g ∥=0.6 and g ⊥=3.3 for 1U) and magnetic moments of 2.48 and 3.28 μB at 2 and 298 K (Figure 2, left, inset).

Figure 2

Left: Temperature‐dependent SQUID data for powdered samples of the uranium(III) complexes 1U (red ▴) and 3U (black •), and the uranium(IV) complex [U(N′′)3(I)] (blue ▪) recorded in a 0.1 T magnetic field over the temperature range 2 to 298 K; inset: CASSCF‐SO calculated (red —) and experimental (black —) magnetic data for 3U. Middle: Field dependent SQUID data for powdered samples of the uranium(III) complexes 1U (red ▴) and 3U (black •), and the uranium(IV) complex [U(N′′)3(I)] (blue ▪) recorded at 1.8 K over the magnetic field range 0 to 7 T. Right: X‐band (ca. 9.5 GHz) EPR spectrum of a powdered sample of 3U at 20 K.

To probe the nature of the metal–carbene linkages in 3U, 3Y, 3La, and 3Nd, we calculated their electronic structures in detail, noting that d0f0 3Y and 3La and 4f3 3Nd represent closed‐shell and f‐analogues for benchmarking purposes, respectively. The Kohn–Sham molecular orbitals of 3U reveal a U=C donor–acceptor interaction, where resonance forms 3U‐a and 3U‐b can be invoked, Scheme 3. The MIC→U two‐electron σ‐donation is represented by HOMO−16, and HOMO−1 reveals U→MIC one‐electron π‐back‐donation from a uranium 5f‐orbital to an empty carbene p‐character orbital that is generated in resonance form 3U‐b. The HOMO and HOMO−2 account for the remaining two 5f electrons of uranium(III). The donor–acceptor character of 3U stands in contrast to 3Y, 3La, and 3Nd where, as expected, only the σ‐component to the bonding is found in the Kohn–Sham orbitals and thus only resonance form 3M‐a is invoked (Scheme 3). Complexes 3Y and 3La as d0f0 complexes would certainly not be expected to exhibit such donor–acceptor character and indeed the molecular orbital that would constitute a M→MIC back‐bond is in both cases the LUMO+1 orbitals with carbene 2p and 4d (Y) and 5d/4f (La) character that sit about 2.5 and about 3.6 eV above the respective HOMO orbitals. For 3Nd, HOMO−2 to HOMO are the 4f electrons, then LUMO to LUMO+4 are dominated by virtual 4f/amide combinations before the relevant Nd→MIC interaction (5d/2p) is found in LUMO+5, some 2.5 eV above the HOMO. Thus, 3Y and 3La do not have the requisite electrons to back‐bond, and 3Nd has the electrons but they are energetically incompatible with back‐bonding to the MIC.

Scheme 3

Two of the principal resonance forms for the MIC complex 3U.

The computed MDC‐q charges and Nalewajski–Mrozek bond orders of 3U, 3Y, 3La, and 3Nd are instructive and fall into two distinct groups of 3U and 3Y/3La/3Nd. Specifically, the U, Ccarbene, α‐C, and α‐N charges are 1.8, −0.81, −0.05, and −0.31 with U=C, C=C, and C−N bond orders of 1.1, 1.64, and 1.22, respectively. Those of 3Y, 3La, and 3Nd are remarkably invariant with av. M, Ccarbene, α‐C, and α‐N charges of 1.35, −0.6, −0.04, and −0.29 and M−C, C=C, and C−N bond orders of 0.6, 1.73, and 1.26, respectively. If there is no back‐bonding, the carbene should show strong stabilizing interactions with the α‐C and ‐N atoms, and the metal and carbene should have low positive and negative charges, respectively. Conversely, if back‐bonding operates in addition to the σ‐donation then the metal and carbene should have higher positive and negative charges, respectively, reflecting the transfer of electron density back from the metal to carbene, and the carbene should consequently have weaker bonding interactions with the α‐C and ‐N atoms. This is exactly the situation that is suggested by the calculations, consistent with the Kohn–Sham descriptions. We note that the metal–carbene bond order in 3U is nearly twice that of 3Y, 3La, and 3Nd, and, recalling that the U→MIC π‐back‐bond involves a singly occupied 5f‐orbital, that it is greater than one suggests the presence of a two‐fold bonding interaction where each component is polarized and of sub‐integer bond order. Considering the generally accepted negligible π‐acidity of MICs the donor–acceptor bond in 3U is notable, and is also remarkably similar to the donor–acceptor interaction found computationally in [U(N′′)3{C(NMeCMe)2}].15 However, we note that the back‐bond must be weak because we could not freeze‐out rotation of the MIC by the solvent low‐temperature limit (−80 °C) in NMR studies.

NBO analysis of 3U (Figure 3) is also consistent with a U=C donor–acceptor interaction. The MIC→U σ‐donation is returned as essentially electrostatic and so is predominantly carbon‐based. However, the U→MIC π‐back‐donation is found to contain 75 % uranium and 25 % carbene character. The carbene acceptor orbital is a pure 2p orbital, whereas the uranium donor orbital is 90 % 5f and 10 % 6d character. As expected, only electrostatic NBO MIC→M interactions are found for 3Y, 3La, and 3Nd.

Figure 3

Natural bond orbitals calculated for 3U with hydrogen atoms omitted for clarity. Left: the carbene to uranium two‐electron σ‐bond component. Right: the uranium to carbene single‐electron π‐back‐bonding interaction.

Along with the orbital‐based perspectives of DFT and NBO analyses we probed the topological electron density description of the M−C bonds in 3U, 3Y, 3La, and 3Nd. The calculations reveal M−C 3,−1 critical points. The ρ(r)MC values are similar for all complexes (0.08–0.12) suggesting polar interactions, since covalent bonds tend to have ρ(r) >0.1, but we note that 3U has the highest ρ(r)MC value. Most importantly, however, the calculated ellipticity parameters ϵ(r)MC are 0.03, 0.06, and 0.09 for 3Y, 3La, and 3Nd, respectively, but for 3U the ϵ(r)UC value is 0.36. This supports the notion of a polarized two‐fold U=C bonding interaction in 3U because a single σ bond or triple σ–π–π bond present symmetrical electron density distributions around the inter‐nuclear axes (ϵ(r)≈0) whereas σ–π double bonds are asymmetric (ϵ(r)>0). For comparison, calculated C−C ϵ(r)CC values in ethane (H3C−CH3), benzene (C6H6), ethylene (H2C=CH2), and acetylene (HC≡CH) are 0.0, 0.23, 0.45, and 0.0, respectively.23

Inspection of the CASSCF‐SO m manifolds of 1U and 3U reveals that there is only a small change in the energies of the three lowest doublets, whilst the two highest energy states are suppressed by about 400 cm−1 in 3U compared to 1U. We also observe a clear change in the g‐values of the ground doublet, reflecting the departure from axial symmetry in 3U, as confirmed experimentally. These modest changes reflect the coordination of the MIC and also the electronic partial‐cancellation effects of the U=C donor–acceptor interaction, analogous to donor–acceptor net‐cancellation effects on the CO stretching frequency of thorium carbonyls.24

To conclude, we have prepared the first examples of f‐block‐MICs, which thus represent a new class of f‐block carbene. These complexes were prepared by a formal 1,4‐proton migration of an NHO, which therefore represents a new, general way to prepare MIC complexes. Quantum chemical calculations suggest that in addition to a MIC→U σ‐donation there is a weak U(5f)→MIC(2p) π back‐bond; although resonance form 3U‐a most likely dominates, resonance form 3U‐b is non‐negligible. As expected 3Y and 3La exhibit no back‐bonding due to their d0f0 natures, and 3Nd though being 4f3 also does not back‐bond as its valence 4f electrons are energetically incompatible to do this. The donor–acceptor character in 3U is reminiscent of d‐block‐carbonyl and Fischer carbene bonding, though the π‐back‐bond is weak. Considering the generally at best weak π‐acidity of MICs, the computational finding of U(5f)→MIC(2p) π‐back‐bond is surprising and highlights that perhaps greater consideration should be given to more widely acknowledging MICs as potential π‐acid ligands when coordinated to sufficiently reducing metals.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.