Molecular Pnictogen Activation by Rare Earth and Actinide Complexes

This review covers the activation of molecular pnictogens (group 15 elements) by homogeneous rare earth and actinide complexes. All examples of molecular pnictogen activation (dinitrogen, white phosphorus, yellow arsenic) by both rare earths and actinides, to date (2015), are discussed, focusing on synthetic methodology and the structure and bonding of the resulting complexes.


Introduction
Rare earth (scandium, yttrium and the lanthanides) and actinide complexes remain underexplored with respect to the transition metals and main group elements but often demonstrate both unique reactivity and molecular properties.Understanding of the bonding and electronic structure of these complexes has particular significance for separation of metals in nuclear waste streams [1].
Activation of molecular pnictogens (group 15 elements) is an area of growing importance; atmospheric dinitrogen (N2) and white phosphorus (P4) are principal sources of N-and P-containing compounds (e.g., polymers, pharmaceuticals, agrochemicals, explosives, and specialty chemicals) but are both very challenging to selectively activate.Metal-arsenic, -antimony and -bismuth complexes remain rare [2][3][4], while the study of metal-pnictogen complexes, including these heavier pnictogen homologues, is also of fundamental importance with respect to Ln and An-pnictogen bonding and electronic structure.

OPEN ACCESS
Fixation of N2, the six electron reduction to two molecules of more reactive ammonia, is necessary for further formation of N-element bonds.In nature, nitrogenase enzymes containing metalloproteins (Fe, Mo or V) fix N2 through proton-coupled electron transfer under ambient conditions [5,6].In industry, the Haber-Bosch process combines N2 and high purity H2 at high temperatures and pressures over heterogeneous iron-or ruthenium-based catalysts [7][8][9][10].This highly efficient process produces 100 million tons of ammonia per year but is the largest energy-consuming process in the modern world today; the need for direct activation and functionalisation of N2 under mild conditions is a clear goal.Accessing appropriately reactive phosphorus building blocks presents a different set of challenges based on the sustainability and efficiency of chemical transformations required; from phosphate rock minerals which are mined globally on a 225 million ton scale per year (2013) [11], phosphate fertilisers derived from phosphoric acid are the major products with the remainder used for elemental phosphorus production.Organophosphorus compounds are generally derived from PCl3, obtained by the chlorination of P4, and subsequent multi-step procedures [12][13][14].Attention has turned to direct and selective activation of elemental phosphorus under mild conditions; this approach is more atom-efficient (which is important given the limited accessible deposits of phosphate rock), avoids the need for large scale production of PCl3 (which is toxic, corrosive and highly reactive), and is both more economically and environmentally sustainable [15].
This review seeks to cover all examples of molecular pnictogen activation (dinitrogen, white phosphorus, yellow arsenic) by both rare earth and actinide complexes to date, focusing on synthetic methodology and the structure and bonding of the resulting complexes.Only well-defined homogeneous complexes will be discussed; heterogeneous and surface chemistry lie beyond the scope of this review.

Cyclopentadienyl Ancillary Ligands
The first isolated, structurally characterised dinitrogen complex of an f-element metal was reported by Evans and co-workers [72].[(η 5 -C5Me5)2Sm]2(μ-η 2 :η 2 -N2) (4) was isolated by slow crystallisation of a toluene solution of the bent metallocene [(η 5 -C5Me5)2Sm]2 under an N2 atmosphere (Figure 2). 4 exists in dynamic equilibrium with the metallocene starting material involving reversible Sm II /Sm III interconversion.In the solid state, 4 displays tetrahedral coordination around each Sm centre with gearing of the [Sm(C5Me5)2] units and the first example of a co-planar M2N2 diamond core for any metal.The bridging, side-on bound N2 has a short N-N distance of 1.088(12) Å (free N2: 1.0975 Å [66]) and does not imply reduction to N2 2− ; however, recent studies by Arnold and co-workers have shown that N-N bond lengths determined using X-ray diffraction experiments can be underestimated and so may not provide the best way of assessing the level of dinitrogen reduction [87,88].Both the Sm-N/C bond lengths and the 13 C-NMR spectral data support formulation of the complex as [Sm III ]2(N2 2− ).Maron and co-workers have reported calculations on the interaction of N2 with [(η 5 -C5Me5)2Ln] (Ln = Sm, Eu, Yb) [89].Since this landmark discovery, the methodology of using reducing divalent rare earth metal complexes to activate N2 has resulted in analogous cyclopentadienyl complexes of Dy (5, 6) [74] and Tm (7, 20) [75].Structurally, these complexes all demonstrate a common planar Ln2N2 core (Ln-N-N-Ln dihedral angle = 0°), with the arrangement of the cyclopentadienyl ligands being dependent on the metal centre and the nature of the ligand itself (Figure 3).
Complexes 4 and 15-18 were characterised by 15 N-NMR spectroscopy; the first reported examples of such spectra for paramagnetic N2 complexes [68].Trivalent Sm, Ce and Pr were chosen due to the low magnetic susceptibility of these ions (4f 5 Sm III , μ = 0.84 μB; 4f 1 Ce III , μ = 2.54 μB; 4f 2 Pr III , μ = 3.58 μB).Broad singlets at high frequency were observed for 15 (871 ppm), 16 (1001 ppm), 17 (2231 ppm) and 18 (2383 ppm).Consistent with the reversible N2 coordination to 4, only a singlet at −75 ppm corresponding to free N2 is observed at 298 K [68].Cooling resulted in a new resonance at −117 ppm at 263 K which shifted linearly to −161 ppm at 203 K and accounts for bound N2.In the context of pioneering NMR spectroscopic characterisation of organometallic complexes, the solid state 15 N-and 139 La-NMR spectra of 14-15 N 2 have also been reported [90].

Amide Ancillary Ligands
For reference, N-N and M-N(N2) bond lengths obtained from single crystal X-ray diffraction experiments and N-N stretching frequencies (obtained by IR or Raman spectroscopy) are summarised in Table 2.The first definitive evidence for an N2 3− reduction product of dinitrogen was demonstrated by Evans and co-workers [93].The LnA3/M system of Y{N(SiMe3)2}3 with KC8 in thf afforded a mixture of (47) from which 44 and 47 could be isolated (Figure 8).
The EPR spectrum of 44-15 N 2 has a 9-line pattern consistent with a triplet of triplets due to two 15 N and two 89 Y nuclei and has a hyperfine coupling constant of 8.2 G implying a N-centred radical, while 47-15 N 2 shows extra coupling to potassium; all spectra indicate the presence of the N2 3− ion.The N-N bond distances are 1.401(6) and 1.405(3) Å respectively and are intermediate between N-N single bonds (1.47 Å in N2H4) and N=N double bonds (1.25 Å in PhN=NPh) [98].The N-N vibrational stretching frequency in 47 is 989 cm −1 , significantly reduced from 1425 cm −1 for 22. Similarly to complexes 22-36, the Y-N2 bonding interaction in these complexes can be described by the donation from a filled Y 4d orbital into an antibonding N2 πg orbital (HOMO).However, the orthogonal antibonding N2 πg orbital is now also occupied by a single electron.Complexes 48-50 (inner-sphere K + ) and 52-56 (outer-sphere K + ) all display interesting magnetic properties; 48 and 52 have the strongest magnetic exchange couplings in a Gd III complex with exchange constants of −27 cm −1 , and 49, 50 and 53-56 demonstrate single-molecule-magnet behaviour [94,95,99,100].Combination of rare earth complexes which demonstrate both high anisotropy and strong exchange coupling potentially provides a route to single-molecule magnets with high blocking temperatures.The diffuse nature of the N2 3− radical facilitates strong coupling in these systems by overlap of the Ln 4f orbitals with the bridging dinitrogen ligand.53 and 54 exhibit magnetic hysteresis up to record blocking temperatures of 13.9 K (0.9 mTs −1 sweep rate) and 8.3 K (0.08 Ts −1 sweep rate) respectively.Competing Ln III -Ln III antiferromagnetic coupling is observed in complexes at low temperatures in 48-50, which have a non-zero Ln-N-N-Ln dihedral angle, demonstrating the importance of geometry of the Ln2N2 unit to magnetic behaviour.

Complexes Containing an Activated N2 Ligand
Perhaps surprisingly, given the number of examples of dinitrogen activation by rare earth complexes, there are very few examples of N2 activation with actinide complexes despite the presence of uranium in early catalysts for the Haber-Bosch process (Figure 11) [107].Actinide-element bonding in the model system [X3An]2(μ-η 2 :η 2 -N2) (An = Th-Pu, X = F, Cl, Br, Me, H, OPh) has recently been studied using relativistic DFT calculations [108].For reference, N-N bond lengths obtained from single crystal X-ray diffraction experiments, N-N stretching frequencies (obtained by IR or Raman spectroscopy) and 14/15 N-NMR spectroscopic data of actinide N2 complexes are summarised in Table 4.The first example of dinitrogen activation by an actinide complex was reported by Scott and co-workers; the trivalent uranium complex {N(CH2CH2NSiMe2 t Bu)3}U reacts with N2 (1 atm) to yield [{N(CH2CH2NSiMe2 t Bu)3}U]2(μ-η 2 :η 2 -N2) (69) [109].In solution, the reaction is reversible and 69 converts back to the trivalent uranium starting material when freeze-pump-thaw degassed (Scheme 1).The solid state structure of 69 illustrates the side-on binding mode of N2 and features an N-N bond length of 1.109(7) Å. Alongside solution magnetic susceptibility measurements of 3.22 μB per uranium centre, these data agree with the dimer being formulated as [U III ]2(N2 0 ).
To date, the only example of a monometallic f-element complex of N2 was reported by Evans et al.; sterically crowded U(η 5 -C5Me5)3 reacts with N2 (80 psi) to afford (η 5 -C5Me5)3U(η 1 -N2) (73) [112].N2 binding is reversible and lowering the pressure results in N2 dissociation.The solid state structure of 73 shows the N2 ligand is bound end-on and linearly (U-N-N = 180°) and the N-N distance is 1.120( 14) Å which is statistically equivalent with free N2.
Arnold and co-workers reported that the trivalent uranium aryloxides U(OAr)3 (Ar = 2,6-t Bu-C6H3 or 2,4,6-t Bu-C6H2) bind N2 (1 atm) to form the side-on bound N2 adducts [(ArO)3U]2(μ-η 2 :η 2 -N2) (74 and 75 respectively) [88].Though 74 was obtained as a minor product, the more sterically hindered 75 was formed in quantitative yield and was stable under dynamic vacuum and in the presence of coordinating solvents and polar small molecules (CO and CO2) under ambient conditions.N2 loss was observed when 75 was heated to 80 °C in a toluene solution.The solid state N-N bond lengths are 1.163 (19) showing π backbonding from uranium f orbitals into an N2 antibonding πg orbitals and that the interaction is strongly polarised.The bonding description is very similar to that in previously calculated models for 69 (formally N2 0 ) [116,117] and 72 (N2 2− ) [118] which display very different N-N bond lengths.While experimental N-N bond distances determined by X-ray diffraction experiments are undeniably useful for quick comparisons, the bond length is likely underestimated since the data is based on electron density rather than atomic positions and thus may not reflect the level of dinitrogen reduction.In these studies, N-N stretching wavenumbers were more accurately reproduced by calculation than bond length and it is proposed that this would be a more suitable measurement for probing N2 reduction.
The most robust actinide N2 complex prepared to date is [{(Mes)3SiO}3U]2(μ-η 2 :η 2 -N2) (76) (Mes = 2,4,6-Me-C6H2) and is stable both in vacuo and in toluene solution up to 100 °C, at which point U{OSi(Mes)3}4 is slowly formed as the major product (52% conversion after 18 h) [87].76 is isolated from the reaction of U{N(SiMe3)2}3 with 3 equivalents of HOSi(Mes)3 under an N2 atmosphere (1 atm).Raman spectroscopy shows a peak at 1437 cm −1 assigned to the N-N stretching mode, indicating a significant level of reduction with respect to free dinitrogen (2331 cm −1 ) and comparing well with 1451 cm −1 recorded for 75 where reduction to N2 2− was assigned.The N-N distances in the solid state are 1.124( 12) Å (76-Eclipsed) and 1.080(11) Å (76-Staggered), which are statistically equivalent to that of free N2; the disparity in implied reduction of N2 from Raman spectroscopy and X-ray diffraction experiments again highlighting that the latter may not be best suited for assigning reduction in these systems.

Complexes Resulting from N2 Cleavage
Tetra-calix-pyrrole ligands bound to Sm II centres have been demonstrated to activate N2 by Gambarotta and co-workers [101].With U III , an unprecedented example of N-N bond cleavage using an molecular f-element complex was observed; when [K(dme)][( Et2 calix [4]pyrrole)U(dme)] is treated with potassium naphthalenide under an atmosphere of N2, N-N bond cleavage occurs to afford [K(dme)4][{K(dme)( Et2 calix [4]pyrrole)U}2(μ-NK)2] (77) (Scheme 2) [119].77 contains two bridging nitrides (U-N: 2.076(6) and 2.099(5) Å) which have contacts with potassium ions (N-K: 2.554(6) Å) that bridge two pyrrolide units on separate ligands.It was postulated that 77 is a Class 1 U IV -U V mixed valence complex on the basis of an absorption at 1247 nm in the near-IR spectrum which is characteristic of U V .The paramagnetism of 77 resulted in NMR silence in both 15 N-and 14 N-NMR spectra.Scheme 2. N-N bond cleavage using a U III complex to form 71.

White Phosphorus Activation by Rare Earth Complexes
Rare earth complexes resulting from P4 activation are illustrated in Figure 12.For reference, average P-P and Ln-P bond lengths obtained from single crystal X-ray diffraction experiments, and 31 P-NMR spectroscopic resonances are summarised in Table 5.The structural cores of complexes 79-84 are shown in Figure 13 for clarity.Roesky and co-workers reported the first example of a molecular polyphosphide of the rare earth elements, [(η 5 -C5Me5)2Sm]4(μ4-η 2 :η 2 :η 2 :η 2 -P8) (79) [123].The samarocene (η 5 -C5Me5)2Sm activates P4 to yield a P8 4− fragment with a realgar-type structure, a process proposed to be driven by the one-electron oxidation of the divalent samarium metal centre.79 has molecular D2d symmetry and the [Cp*2Sm] units bridge the P8 4− cage with Sm-P distances in the range of 2.997(2) to 3.100(2) Å. DFT calculations support the strongly ionic character of the Sm-P bonds.
The following report of P4 activation came over a decade later and was the first using a uranium complex [129].The U III tris(amide) U{N(R)Ar}3(thf) (R = t Bu or Ad, Ar = 3,5-Me-C6H3) reacts with 0.5 equivalents of P4 to yield [{Ar(R)N}3U](μ-η 4 :η 4 -cyclo-P4) (R = t Bu (87), Ad (88)) which contains a cyclo-P4 2− unit and where the metal centres have been formally oxidised to U IV .In the solid state structures, the average P-P bond distance is 2.159 Å and the P-P-P angle is 90°; both statistically equivalent across the two structures.Resonances at 794 and 803 ppm were observed for 87 and 88 respectively in the 31 P-NMR spectrum.Computational studies implied that the U-P bonding character is largely ionic with the presence of a weak δ-bonding interaction between filled U df hybrid orbitals and the P4 2− LUMO.
Very recently, Liddle and co-workers described the first example of a cyclo-P5 complex resulting from activation of P4 by an f-block complex [131].[{N(CH2CH2NSi i Pr3)3}U]2(μ-η 5 :η 5 -cyclo-P5) (91) was prepared by reaction of {N(CH2CH2NSi i Pr3)3}U with 0.25 equivalents of P4.Spectroscopic and magnetic measurements support oxidation to afford U IV centres and charge transfer resulting in a formal P5 2− ligand in this inverse sandwich complex.Despite the isolobal analogy of cyclo-P5 with the cyclopentadienyl anion, which bonds to metal centres using primarily σ-and π-bonding, calculations on 91 suggest that the principal U-P interactions involve polarised δ-bonding and this can be attributed to the energetically available uranium 5f orbitals of correct δ-symmetry.
Beyond arsenic in group 15 are antimony and bismuth.While activation of molecular forms of these elements is unlikely, it is worth noting that Scheer and co-workers reported a tungsten terminal stibido complex {N(CH2CH2NSiMe3)3}3W≡Sb prepared from reaction of {N(CH2CH2NSiMe3)3}3WCl with LiSb

Conclusions and Perspectives
To date, a wide range of rare earth dinitrogen complexes have been prepared (1-68), including group 3 metal ions and 4f elements at both ends of the periodic table, despite the limited radial extension of the 4f orbitals and the trivalent oxidation state being the most prevalent.In fact, apart from the very first f-element dinitrogen complex (4), all of the other complexes are air-sensitive but stable to N2 dissociation in vacuo.Reduction of N2 to N2 2− has most commonly been achieved with the [A2(thf)xLn]2(μ-η 2 :η 2 -N2) structural motif  whereas examples of reduction to N2 4− have all involved more complex multidentate ligands (61)(62)(63)(64)(65)(66)(67)(68).The nature of the bonding in complexes of the form [A2(thf)xLn]2(μ-η 2 :η 2 -N2) has been extensively studied for both group 3, and closed and open shell 4f n metal ions demonstrating that the Ln-N (N2) bonding is based on a Ln nd-N2 π* interaction.These systems have allowed for the first definitive characterisation of the N2 3− radical reduction product of dinitrogen (44,47) and this has now been extended to many of the rare earth elements (45,46,(48)(49)(50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60).Isolation of the N2 3− radical in homogeneous complexes is significant since it is likely to be a transient species in other transition metal systems and may also have a role in biological N2 fixation.
In terms of reactivity, the rare earth N2 complexes prepared thus far react with N2 dissociation rather than N2 cleavage and functionalisation [77,[159][160][161].Related to this, the understanding of the nature of bonding of Ln-N multiple bonds is of fundamental interest and the isolation of terminal imido (Ln=NR) complexes has only recently been reported [162][163][164].It is also significant to consider the reactivity of Ln-N2 complexes in the context of other low electron count early transition metal complexes; here, N2 cleavage or functionalisation can be achieved through ligand induced reductive cleavage [165][166][167][168][169][170][171][172][173], and more generally, N-N bond scission of reduced N2 derivatives can be attained through metal-ligand cooperativity [174,175].
Actinide dinitrogen complexes are much rarer, with just 8 examples of well-defined molecular uranium complexes (69)(70)(71)(72)(73)(74)(75)(76).Of these, only the heterobimetallic U-Mo end-on N2 complexes (70,71) and recently prepared U IV aryloxide and siloxide side-on N2 complexes (75,76) are thermally robust and stable when exposed to vacuum.Key steps forward have been made in the understanding of U-N (N2) bonding in the side-on N2 complexes as a polar covalent U 5f-N2 π* interaction, and in the rationalisation of the differences between solid state N-N bond lengths and the overall electronic structure of these complexes.This understanding, in combination with well-designed ligand sets may lead the way in the preparation of other isolable actinide dinitrogen complexes for further study.
Though the isolated actinide N2 complexes, like the rare earth N2 complexes, tend to react with N2 loss rather than N2 functionalisation or cleavage, there are two reports of such reactivity.Importantly, in both cases, the putative An(N2) complex was not observed.Cleavage of N2 by a uranium tetra-calix-pyrrole complex results in a bimetallic complex with bridging nitrides (78) whereas a thorium bisphenolate complex activates and functionalises N2 to a parent amide ligand [NH2] − (79) through an unknown mechanism.Both reactions occur in the presence of an external reductant.Examples of isolable terminal uranium nitrides (U≡N), derived from NaN3, have only recently been reported [176][177][178], but it has already been demonstrated that these systems are capable of nitride functionalisation.78 and 79 remain standout examples in demonstrating that actinide complexes can both cleave and functionalise N2, but also highlight how much more remains to be understood in this field.P4 activation by rare earth complexes has led to both P8 4− ions with realgar-type structures (79, 80), and P7 3− ions (81-84) using cyclopentadienyl and amido ancillary ligands.Promisingly, functionalisation of the P7 3− unit in 81 and 82 was found possible using Me3SiI to afford P7(SiMe3)3.This is an interesting prospect for the synthesis of organophosphorus compounds from a P4 building block.Actinide P4 activation results in a more diverse array of phosphorus ligands; P3 3− (85), P6 4− , (86), cyclo-P4 2− (87-89), P7 3− (90), and cyclo-P5 2− (91).Similar to rare earth chemistry, the P7 3− ions in 90 could be functionalised by P-Si, P-C or P-Li bond formation to afford P7R3 units and this reaction cycle could be completed with two turnovers.91 is the first example of an f-element being able to fragment and catenate P4 to cyclo-P5 2− .Despite the parallels with the cyclopentadienyl ligand, calculations suggest the U-P (cyclo-P5) interaction to be based on polarised δ-bonding and electronic structure in these systems can be described as [U IV ]2(P5 2− ).Putting this area into perspective, transition metals have already been shown to activate molecular phosphorus and, in limited examples, to result in further functionalisation.
There remains only a lone example of arsenic activation by a thorium butadiene complex leading to a P6 cage (92) and no examples using rare earth metals.The first examples of crystallographically characterised uranium arsenide (U-AsH2), arsenidene (U=AsH) and arsenido (U≡AsK) complexes have only recently been reported, using KAsH2 as a source of arsenic [179].These compounds raise the question of the diverse reactivity that actinide complexes could be expected to show with molecular arsenic and whether the formation of an unsupported, terminal actinide arsenide bond (M≡As) is accessible.
It is clear that molecular pnictogen activation by rare earth and actinide metal complexes is an exciting field of study which remains underdeveloped with respect to transition metals and main group elements.These unique metals offer the potential of new reactivity and functionalisation chemistry with the pnictogen elements, while the fundamental study of M-pnictogen bonds remains important.

Figure 13 .
Figure 13.Overview of the LnxPn structural cores resulting from P4 activation by rare earth complexes.

Figure 15 .
Figure 15.Overview of the AnxPn structural cores resulting from P4 activation by actinide complexes.
Second independent molecule in unit cell; b trans arrangement of thf. a
a Referenced to CH315NO.

Table 5 .
Summary of rare earth P4 activation complexes.