Non-traditional ligands in f-block chemistry

The molecular chemistry of the f-elements, i.e. the lanthanides and actinides, is traditionally dominated by the use of carbon, nitrogen, oxygen or halide ligands. However, the use of metal-based fragments as ligands is underdeveloped, which contrasts to the field of d-block metal–metal complexes that have developed extensively over the last 50 years. Consequently, the use of metal-based fragments as ligands to the f-elements may be regarded as ‘non-traditional’. This review outlines the development of compounds that possess f-element–metal bonds that may be described as polarized covalent or donor–acceptor in nature. For this review, the f-element is defined as (i) a group 3 or lanthanide metal: scandium, yttrium and lanthanum to lutetium or (ii) an actinide metal: thorium or uranium, and the metal is defined as a d-block transition metal, or a p-block triel (group 13, aluminium or gallium), a tetrel (group 14, silicon, germanium or tin), or a pnictide (group 15, antimony or bismuth) metal. Silicon, germanium and antimony are traditionally classified as metalloids, but we include them in this review for completeness. This review focuses on complexes that have been unambiguously structurally authenticated by single crystal X-ray diffraction studies, and novel aspects of their syntheses, properties and reactivities are highlighted.


Introduction
The f-elements are traditionally described as hard electropositive ions that preferentially bind to hard Lewis basic ligands. Consequently, the vast majority of f-element complexes contain lanthanide or actinide centres coordinated by anionic carbon, nitrogen, oxygen or halide ligands. For example, the uranyl [UO 2 ] 2C unit accounts for more than 60 per cent of all structurally characterized molecular uranium complexes. Since these ligand classes are dominant, they may be referred to as traditional in nature.
The study of metal-metal bonds is of significant fundamental interest because it is a fertile territory for furthering our understanding of structure and bonding, catalysis, metal surface chemistry and bio-inorganic chemistry. Historically speaking, major advances regarding our understanding of chemical bonding have emerged from metal-metal chemistry, as exemplified by rhenium-rhenium (Cotton et al. 1964), zinc-zinc (Resa et al. 2004), magnesium-magnesium (Green et al. 2007), chromium-chromium (Nguyen et al. 2005) and hexa-indium (Hill et al. 2006) complexes. However, while metal-metal bonding in the d-and p-blocks of the periodic table is widespread and well understood, and now even known in the s-block, very little is known about metal-metal complexes where one metal is an f-element. The dearth of f-element-metal complexes compared with d-and p-block metal-metal complexes renders their assignment as 'nontraditional' a valid classification.
The f-block, and especially the actinide elements, comprises a collection of metals that are, generally speaking, far less researched than all other areas of the periodic table. The area can legitimately be described as a frontier area because our level of understanding of it is so far behind the rest of the periodic table.
Given the paucity of f-element-metal chemistry, and the advances that have historically resulted from investigating metal-metal chemistry, the study of f-element-metal bonds represents an excellent opportunity to advance our understanding of chemical bonding and catalysis at the foot of the periodic table. Furthermore, this is of direct relevance to the remediation of nuclear waste and the design of new and more efficient catalytic cycles. Herein, we present a review of the emerging area of f-element-metal bonds, focusing principally on complexes that have been unambiguously structurally authenticated by single crystal X-ray diffraction studies, and we highlight novel aspects of their syntheses, properties and reactivities.

(a ) Lanthanide-transition metal complexes
The first structurally authenticated lanthanide(II)-transition metal bond was reported by Shore (Deng & Shore 1991). Reduction of Fe 3 (CO) 12 by three equivalents of ytterbium metal in liquid ammonia resulted in the formation of [(NH 3 ) x YbFe(CO) 4 ] (1), which was reported to be a highly air-sensitive yellow solid. Orange crystals of [{(MeCN) 3 YbFe(CO) 4 } 2 $MeCN] N (2) were isolated from a cold solution of 1 in acetonitrile. The X-ray crystal structure of 2 revealed a polymeric zigzag ladder structure (figure 1a). Each ytterbium centre is coordinated by three acetonitrile ligands and two isocarbonyl linkages from two different iron carbonyl units. Under C 3v symmetry, the d 10 [Fe(CO) 4 ] 2K dianion formally possesses a lone pair directed towards an axial site, which results in the coordination sphere of ytterbium being further supplemented by an Fe/Yb dative donor-acceptor interaction that forms the rungs of the ladders by linking the two strands of the ladder together. Shore (Deng et al. 1996) subsequently reported a lattice solvent-free modification of 2, namely [(MeCN) 3 YbFe(CO) 4 ] N (3). The X-ray crystal structure of orange 3 revealed a polymeric sheet network (figure 1b). The removal of lattice solvent clearly allows the number of bridging isocarbonyl groups to increase from 2 to 3. Thus, individual ladders become cross-linked by the additional isocarbonyl to extend the onedimensional ladders into two-dimensional sheets. The Yb-Fe bond lengths of 3.010(1) and 3.046(1) Å for 2 and 3, respectively, are within the Yb-Fe sum of covalent radii (3.39 Å) (Cordero et al. 2008).
The first unsupported lanthanide-transition metal bond was reported by Beletskaya and Voskoboynikov (Beletskaya et al. 1993). Reaction between [(h 5 -C 5 H 5 ) 2 LuCl(THF)] and Na[Ru(h 5 -C 5 H 5 )(CO) 2 ] in THF afforded colourless crystals of [(h 5 -C 5 H 5 ) 2 Lu(THF)-Ru(h 5 -C 5 H 5 )(CO) 2 ] (4; scheme 1). A single crystal X-ray diffraction study of 4 revealed an unsupported Lu-Ru bond length of 2.995(2) Å that is slightly shorter than the sum of the covalent radii of lutetium and ruthenium (3.33 Å) (Cordero et al. 2008). Complex 4 was reported to exhibit Fourier transform infrared (FTIR) v CO bands at 2027 and 1965 cm K1 ; the corresponding Lu-Fe complex was characterized in THF solution by FTIR (v CO bands at 1946 and 1859 cm K1 ), but it could not be isolated in crystalline form due to quantitative decomposition within 1 hour of preparation.
Diaconescu reported a series of scandium-alkyl complexes supported by a ferrocene diamide ligand that was shown to exhibit weak, dative Fe/Sc interactions (figure 3; Carver et al. 2008). For example, the complexes [Sc{fc( NSiMe 2 Bu t ) 2 }(CH 2 C 6 H 3 -3,5-Me 2 )(THF)] (9; fcZ[Fe(C 5 H 4 ) 2 ]) and [Sc{fc(NSiMe 2 Bu t ) 2 }(Me)(THF) 2 ] (10) were crystallographically characterized, revealing Sc-Fe distances of 3.167(2) and 3.258(1) Å, respectively, which compares with a value of 3.02 Å for the sum of the covalent radii of scandium and iron (Cordero et al. 2008). DFT calculations revealed the Sc-Fe distance to  be highly sensitive to the presence or absence of coordinating ether solvents, and the orbital contributions from scandium to the molecular orbitals involved in the donor-acceptor interaction were found to lie in the range of 2.8-13.9%.
An unsupported Nd-Fe bond has been very recently reported by Arnold and McMaster (Arnold et al. 2009). Reaction of half an equivalent of the neodymium amido-tethered N-heterocyclic carbene (NHC) precursor [{Nd(L 0 )(N 00 )(mKI)} 2 ] [L 0 ZBu t NCH 2 CH 2 {C(NCSiMe 3 CHNBu t )}; N 00 ZN(SiMe 3 ) 2 ] with K[Fe(h 5 -C 5 H 5 ) (CO) 2 ] gave, after the elimination of potassium iodide and work-up, yellow crystals of [Nd(L 0 )(N 00 ){Fe(h 5 -C 5 H 5 )(CO) 2 }] (11) (scheme 3). The FTIR v CO bands at 1915 and 1845 cm K1 compare well with the iron analogue of 4, and show the iron centre in 11 to be more electron rich than that of 4, which is commensurate with the presence of a strongly nucleophilic NHC ligand at neodymium. A DFT study on 11 concluded that the Nd-Fe bond is highly polarized. Indeed, the polarization of the Nd-Fe bond is so great that it may be regarded as being essentially electrostatic in nature. This accounts for the electron-rich nature of the iron centre, and the low-stretching frequencies of the CO groups, as a consequence of extensive Fe-CO back bonding, which was confirmed by the inspection of the Kohn Sham molecular orbitals. The Nd-Fe bond length of 2.9942(7) Å is within the sum of the covalent radii for neodymium and iron (3.33 Å) (Cordero et al. 2008) and perhaps reflects the low coordination number of 4 for neodymium in this complex since the DFT study indicated negligible covalency in the Nd-Fe bond.

(b ) Lanthanide-triel complexes
The first dative lanthanide-triel bond was reported by Roesky (Gamer et al. 2006). Solvent-free reactions between a quarter molar equivalent of [{Al(h 5 -C 5 Me 5 )} 4 ] with one molar equivalent of [Ln(h 5 -C 5 Me 5 ) 2 ] (LnZEu; Yb) at 1208C in an evacuated ampoule afforded the target complexes [(h 5 -C 5 Me 5 ) 2 Ln-Al(h 5 -C 5 Me 5 )] (LnZEu: 12; Yb: 13) as red and green crystals, respectively (scheme 4). The aluminium(I) fragment is neutral and binds to the lanthanide centre in a dative donor-acceptor manner with minimal charge-transfer and orbital contributions. The dative interaction is clearly weak as 12 and 13 dissociate in solution to [{Al(h 5 -C 5 Me 5 )} 4 ] and [Ln(h 5 -C 5 Me 5 ) 2 ]. Single crystal X-ray diffraction studies showed Eu-Al and Yb-Al bond lengths of 3.365(2) and 3.198(2) Å, respectively, which are both longer than the sum of the covalent radii for each pair (3.19 and 3.08 Å, respectively) (Cordero et al. 2008). DFT calculations confirmed the assignment of oxidation states as Ln(II)-Al(I) and revealed very low Ln-Al bond dissociation energies of 35 and 37 kJ mol K1 that are in line with the instabilities of 12 and 13 in solution.
The first polarized-covalent lanthanide-triel bond was reported by Arnold and Jones (Arnold et al. 2007). Reaction of half an equivalent of [{Nd(L 0 )(N 00 )(mKI)} 2 ] with the gallium(I ) heterocycle [Ga(NArCH) 2 ] [K(tmeda)] (ArZ2,6-Pr i 2 C 6 H 3 ; tmedaZtetramethylethylenediamine) afforded the complex [Nd(L 0 )(N 00 ){Ga(NArCH) 2 }(THF)] (14) as red crystals (scheme 5). In contrast to 12 and 13, complex 14 was prepared in a strong donor solvent in high yield. The Nd-Ga bond length of 3.220(2) Å is identical to the sum of the covalent radii for neodymium and gallium (3.22 Å) (Cordero et al. 2008). A DFT study on a model complex of 14 showed a high, for a lanthanide, Nd-Ga Wiberg bond order of 0.83 and a bond composed of 87 per cent gallium (4s4p 1.67 ) and 13 per cent neodymium (6s6p 0.01 5d 0.36 ) character. The bond dissociation energy for the model of 14 was calculated to be 386 kJ mol K1 , which is an order of magnitude greater than that in 12 and 13 and consistent with the far greater solution stability of 14 compared with 12 and 13. Subsequent to 12, 13 and 14, Roesky reported the synthesis of Eu-Ga and Yb-Ga bonds (Wiecko & Roesky 2007). In contrast to 12 and 13, which had to be prepared under solvent-free conditions, but in common with 14, the purple-red [(h 5 -C 5 Me 5 ) 2 Eu{Ga(h 5 -C 5 Me 5 )} 2 ] (15) and red [(h 5 -C 5 Me 5 ) 2 YbGa(h 5 -C 5 Me 5 )(THF)] (16) complexes were prepared in solution (scheme 6). However, while 14 could be prepared in strong donor media such as THF, the presence of even stoichiometric quantities of THF apparently resulted in decomposition during the preparation of 15 and 16 and the use of arene solvents was necessary. Gallium is slightly larger than aluminium, so, in 15, two gallium(I) centres coordinate to europium, whereas only one aluminium(I) centre coordinates to the europium centre in 12. Europium(II) is larger than ytterbium(II) due to the lanthanide contraction, so for 16 only one gallium(I) group can be accommodated. However, the coordination sphere of ytterbium is clearly not satisfied, which results in coordination of a molecule of THF. The Eu-Ga and Yb-Ga bond lengths were found to be 3.250(2)/3.391(2) and 3.287(2) Å, respectively. As for 12 and 13, the observed Ln-Ga bond lengths are significantly longer than the sum of the covalent radii for each pair (3.20 and 3.09 Å, respectively) (Cordero et al. 2008).
Very recently, Jones has reported a series of polarized-covalent Sm-Ga, Eu-Ga, Tm-Ga and Yb-Ga bonds using lanthanide(II) di-iodides and the gallium(I) heterocycle employed in the synthesis of 14 ). Previous attempts to prepare homoleptic lanthanide(II)-gallium(I) bonds had met with failure. However, it was discovered that the addition of an excess of the potentially bidentate amine tmeda allowed isolation of the complexes [Ln{Ga(NArCH) 2 } 2 (tmeda) 2 ] (LnZSm: 17; Eu: 18; Yb: 19) as dark green, orange and red-orange crystals, respectively (scheme 7). In each case, a mutually trans octahedral complex was authenticated by single crystal X-ray diffraction. The Sm-Ga, Eu-Ga and Yb-Ga bond lengths of 3.312(2), 3.312(2) and 3.226(2) Å, respectively, are longer than the sum of covalent radii for the respective pairs (Sm-Ga 3.20, Eu-Ga 3.20 and Yb-Ga 3.09 Å) (Cordero et al. 2008), but are, unsurprisingly, shorter than the corresponding dative bonds in 15 and 16. The corresponding attempt to prepare a Tm-Ga bond resulted in oxidation of thulium(II) to thulium(III) and loss of one gallium centre to give [Tm{Ga(NArCH) 2 }{(NArCH) 2 }(tmeda)] (20). The Tm-Ga bond length of 2.974(2) Å is markedly shorter than the corresponding bond lengths in 17-19, and is well within the sum of the covalent radii of thulium and gallium (3.12 Å) (Cordero et al. 2008).
The first lanthanide(II)-stannyl derivative was reported by Cloke and Lawless (Cloke et al. 1991). Reaction between ytterbium(II) di-iodide and an in situ prepared [K{Sn(CH 2 Bu t ) 3 }] afforded orange crystals of [Yb{Sn(CH 2 Bu t ) 3 } 2 (THF) 2 ] (23) after work-up (scheme 8   X-ray structure of 23 revealed polarized-covalent Yb-Sn bond lengths of 3.216(1) Å, which is within the sum of the covalent radii for ytterbium and tin (3.26 Å) (Cordero et al. 2008). The incipient basicity of the tin group, which may be regarded as a 'heavy alkyl', was exploited in protonolysis reactions to prepare a range of ytterbium(II) complexes that were characterized by multi-nuclear NMR spectroscopy.
In the same year of the report of an Yb(II)-Sn(IV) bond by Cloke and Lawless, Bochkarev reported two complexes containing Yb(II)-Sn(IV) bonds (figure 5; Bochkarev et al. 1991a,b). The reaction of ytterbium naphthalenide with tetraphenyltin and/or hexaphenyldistannane afforded two products that were separated by fractional crystallization. The first product, isolated in 36 per cent yield as yellow crystals, was the complex [Yb(SnPh 3 ) 2 (THF) 4 ] (24) that was shown to be a mutually trans octahedral complex exhibiting Yb-Sn bond lengths of 3.305(1) Å. The second product was isolated in 44 per cent yield as dark ruby, almost black, crystals, and was shown to be [(Ph 3 Sn)(THF) 2 Yb(mKPh) 3 Yb(THF) 3 ] (25). The Yb-Sn bond length in 25 was found to be 3.379(1) Å, which is substantially longer than that observed in 23 or 24 and it exceeds the sum of the covalent radii (3.26 Å) (Cordero et al. 2008).
As part of a programme aimed at investigating s-bond metathesis reactions involving silicon, Tilley reported the Sm-Si complex [Sm(h 5 -C 5 Me 5 ) 2 {SiH(SiMe 3 ) 2 }] (26) (figure 6; Radu & Tilley 1992;Radu et al. 1996b). Complex 26 was prepared by the reaction of five equivalents of SiH 2 (SiMe 3 ) 2 with [Sm(h 5 -C 5 Me 5 ) 2 {CH(SiMe 3 ) 2 }] and isolated as red crystals. The lutetium analogue was also isolated as blue-green crystals, but no X-ray structure has been reported. An excess of silane reagent was found to be necessary to suppress the thermal decomposition of the samarium alkyl reagent. The preparation of 26 is interesting because the s-bond metathesis reaction observed is in competition with alkyl transfer to silicon with concomitant hydride transfer to samarium; the bulky alkyl group apparently suppresses this reaction enabling the formation of 26. Tilley noted the reaction to be of second order with a variable induction period, which implies a relatively complex mechanism. The X-ray crystal structure of 26 revealed it to be dimeric in the solid state; the Sm-Si bond length  is 3.052(8) Å and the dimer is formed by Sm/H 3 C-Si contacts of 2.970(2) Å. The longer Sm-Si bond length in 26 compared with 21 can be attributed to the greater steric demands of C 5 Me 5 compared with C 5 H 5 . In solution, 26 was found to be monomeric and far more reactive than the corresponding alkyl. For example, 26 reacted rapidly with hydrogen to give [{Sm(h 5 -C 5 Me 5 ) 2 (m-H)} 2 ] and SiH 2 (SiMe 3 ) 2 , with ethylene to give polyethylene, and with MesSiH 3 (MesZ2,4,6-Me 3 C 6 H 3 ) to give [{Sm(h 5 -C 5 Me 5 ) 2 (m-H)} 2 ], SiH 2 (SiMe 3 ) 2 and MesH 2 SiSiH 2 Mes.
Tilley subsequently investigated the reactivity of the Sc-Si complex [Sc(h 5 -C 5 H 5 ) 2 (THF){Si(SiMe 3 ) 3 }] (27) (scheme 9) (Campion et al. 1993). Complex 27 was prepared from the reaction between [{Sc(h 5 -C 5 H 5 ) 2 (m-Cl)} 2 ] and two equivalents of [Li{Si(SiMe 3 ) 3 }] and isolated as a yellow crystalline solid with a Sc-Si bond length of 2.862(2) Å ( 29 Si: d 31.0 and K126.2 ppm for the trimethylsilyl and central silicon centres, respectively). Complex 27 was found to react with carbon monoxide to afford, in the presence of 2-methyltetrahydrofuran, the orange double-insertion product complex 28. Removal of reaction volatiles and solvent from 28 resulted in the formation of green 29, which was isolated as green crystals from pentane. Complex 29 can be prepared directly if the carbonylation reaction is carried out in non-polar solvents. Isonitriles were also found to react with 27 to give the yellow h 2 -iminosilaacyl insertion product 30 that was isolable. Complex 30 was, however, found to react further to give the fused-ring system 31; the mechanism was proposed to follow the route outlined in scheme 9, but attempts to validate this mechanism by trapping experiments were unsuccessful.
A novel half-sandwich ytterbium(II) complex containing an Yb-Si bond was reported by Lawless (figure 8; Corradi et al. 1996). Complex 41 was prepared from the reaction of [Yb(h 5 -C 5 Me 5 ) 2 (OEt 2 )] with [Li{Si(SiMe 3 ) 3 }(THF) 3 ] in toluene and was isolated as orange needles. The retention of one C 5 Me 5 ligand is noteworthy because attempts to increase the yield of 41 by adding excess lithium reagent did not result in an increased yield of 41 or the formation of the bis-silyl compound. The complex was first characterized by 171 Yb and 29 Si NMR spectroscopy. The 171 Yb NMR spectrum exhibited a singlet at d 421 ppm with the anticipated satellites from ytterbium-silicon coupling ( J YbSi Z829 Hz). The 29 Si NMR spectrum showed signals at d K2.9 and K158 ppm; the former was assigned to the trimethylsilyl groups ( J YbSi Z11.5 and J SiC Z40 Hz) and the latter to the central silicon that confirmed the ytterbium-silicon coupling constant and also displayed silicon-silicon coupling ( J SiSi Z26 Hz). The study provided the first example of ytterbium-silicon coupling. An X-ray diffraction study showed the Yb-Si bond length to be 3.032(3) Å, which is only slightly longer than the sum of the covalent radii for ytterbium and silicon (2.98 Å) (Cordero et al. 2008). The first dative, lanthanide-silylene complexes were reported by Lappert ( figure 9; Cai et al. 2000). The isoleptic complexes [Ln(h 5 -C 5 H 5 ) 3 {Si(NCH 2 Bu t ) 2 C 6 H 4 -1,2}] (LnZY: 42; Yb: 43) were prepared by mixing the appropriate triscyclopentadienyl lanthanide complex with the silylene in toluene, and they were isolated as colourless and green crystals, respectively. Despite the high yttriumsilicon coupling constant in the 29 Si NMR spectrum of 42 ( 29 Si: d 119.5 ppm, J YSi Z59 Hz), the silylene was found to be labile and, at 338 K, the sole siliconcontaining species in solution was the free silylene, whereas at 188 K, the ratio of 42 to free silylene was 6 : 10; at 298 K this ratio was 1 : 14. However, in the case of complex 43, free silylene was not detected in the temperature range 238-298 K and it was only observed in significant quantities at 343 K. A single crystal X-ray diffraction study showed the Y-Si and Yb-Si bond lengths to be 3.038(2) and 2.984(2) Å, respectively, which are essentially the same as the sum of the covalent radii for each pair (3.01 and 2.98 Å) (Cordero et al. 2008).
In a continuation of his previous studies, Tilley reported the Lu-Si complex [Lu(h 5 -C 5 Me 5 ) 2 (SiH 2 -2-OMeC 6 H 4 )] (44) (figure 10; Castillo & Tilley 2001). Complex 44 was prepared from the reaction between [Lu(h 5 -C 5 Me 5 ) 2 (m-H) 2 ] and two equivalents of H 3 Si-2-OMeC 6 H 4 and isolated as colourless crystals ( 29 Si : d K39.6 ppm). A single crystal X-ray diffraction study showed an Lu-Si bond length of 2.823(5) Å, which is within the sum of the covalent radii of lutetium and silicon (2.98 Å) (Cordero et al. 2008) and shorter than the Lu-Si bond distance in 22, which is consistent with the neutral and ate natures of 44 and 22, respectively. Thermolysis of 44 at 808C resulted in the formation of [Lu(h 5 -C 5 Me 5 ) 2 (C 6 H 4 OMe)] and formal extrusion of 'SiH 2 ', which resulted in the appearance of SiH 4 and H 3 SiSiH 3 , as well as other unidentified silanes that are likely to be dehydrocoupling products. The synthesis and structures of two lanthanide-silyl complexes [{Ln(h 5 -C 5 Me 5 ) 2 (SiH 3 )(K)(THF)} N ] (LnZorange-red Eu: 45; dark red Yb: 46) (figure 11) were reported by Hou (Hou et al. 2003). The analogous dark red samarium complex was also made, but poor crystal quality prevented its structural characterization. The complexes were prepared from the reaction of [Ln(h 5 -C 5 Me 5 ) 2 (THF) 2 ] with '[K(SiH 2 Ph)]'. However, it is likely that the [K(SiH 2 Ph)] contained a significant proportion of [K(SiH 3 )], which may have been produced during its preparation from KH and PhSiH 3 . Complexes 45 and 46 form polymeric honeycomb two-dimensional sheets that can be viewed as individual chains of Cp Ã LnCp Ã K units (Cp Ã ZC 5 Me 5 ), which are cross-linked by bridging 'inter-chain' SiH 3 units. The Eu-Si and Yb-Si bond lengths of 3.239(3) and 3.091(3) Å are longer than the sum of the covalent radii for the respective pairs (3.09 and 2.98 Å, respectively) (Cordero et al. 2008), which reflects the bridging nature of the SiH 3 units. Complexes 45 and 46 are highly active ethylene and styrene pre-catalysts. In principle, an anionic polymerization mechanism could operate where a one electron transfer from the Ln(II) centre to a monomer initiates the reaction. Alternatively, migration of the SiH 3 unit could initiate polymerization. For ethylene polymerization, the europium(II) and ytterbium(II) complexes gave good activities despite electron transfer being difficult, which indicates that the latter mechanism was in operation. For samarium, either mechanism could account for its high activity (94 g mmol K1 h K1 ). The europium activity (75 g mmol K1 h K1 ) was much greater than that for ytterbium (9 g mmol K1 h K1 ), which was rationalized by the larger ion size of the former compared with the latter. The polymerization of 700 equivalents of styrene by the samarium, europium and ytterbium complexes was accomplished in quantitative (10 min), 91 (1 hour) and 89 per cent (5 hours) yields, respectively, and the generation of block copolymers was also possible.
Other than complexes 42 and 43 reported by Lappert, the only other example of a lanthanide-silylene bond is the complex [Sm(h 5 -C 5 Me 5 ) 2 {Si(NBu t CH) 2 }] (47) reported by Evans ( figure 12; Evans et al. 2003). The paramagnetic nature of the samarium(II) centre prevented a signal being observed in the 29 Si NMR spectrum, but variable temperature 1 H NMR spectra showed that the silylene remained coordinated up to 343 K, which contrasts to 42. However, despite this apparent stability, the silylene ligand in 47 is displaced by THF, indicating that THF is a stronger donor; for comparison, THF is usually displaced by NHCs. Complex 47 was isolated as purple crystals and its structure was determined by X-ray diffraction, revealing a Sm-Si bond length of 3.191(1) Å that is marginally longer than the sum of the covalent radii of samarium and silicon (3.09 Å) (Cordero et al. 2008). Interestingly, the silylene is not coordinated symmetrically and is instead coordinated in a skewed manner to accommodate a CH 3 $$$Sm contact of 3.396(4) Å from one of the N-tert-butyl methyl groups.
The syntheses and structures of two scandium-silyl complexes, which are closely related to 26 and 27, were reported by Tilley (figure 13; Sadow & Tilley 2005). The complexes [Sc(h 5 -C 5 Me 5 ) 2 (SiH 2 SiPh 3 )] (48) and [Sc(h 5 -C 5 Me 5 ) 2 {SiH(SiMe 3 ) 2 }] (49) were prepared from methane elimination reactions between [Sc(h 5 -C 5 Me 5 ) 2 (CH 3 )] and Ph 3 SiSiH 3 or (Me 3 Si) 2 SiH 2 , respectively, and were isolated as yellow crystalline solids. A fivefold excess of silane was used to suppress the formation of [Sc(h 5 -C 5 Me 5 ) 2 (H)] and RMeSiH 2 from the other competing s-bond metathesis reaction. Single crystal X-ray diffraction experiments yielded an Sc-Si bond distance of 2.797(1) Å for 48; the Sc-Si bond in 49 was found to be disordered, so the refined distances of 2.804 and 2.859 Å are not meaningful. The Sc-Si bonds in 48 and 49 are, nevertheless, shorter than the Sc-Si bond distance in 27, but they are all slightly longer than the sum of the covalent radii for scandium and silicon (2.81 Å) (Cordero et al. 2008). Both 48 and 49 are significant because they represent one-half of a possible catalytic cycle for the alkylation of silanes by s-bond metathesis reactions where the other half of the catalytic cycle provides an alkyl that is derived from s-bond metathesis between a lanthanide hydride and an alkane (e.g. methane). As part of a study aimed at investigating the interaction of the so-called hypersilyl anion with homoleptic lanthanide silyl amides, Niemeyer (2006) reported the complex [{Yb(N 00 ) 2 [Si(SiMe 3 ) 3 ]K} N ] [N 00 ZN(SiMe 3 ) 2 ] (50) (figure 14). Complex 50 was prepared by the addition of [K{Si(SiMe 3 ) 3 }] to [Yb(N 00 ) 2 ] and was isolated as deep orange crystals. The reaction is noteworthy because addition of [K{Si(SiMe 3 ) 3 }] to [Ln(N 00 ) 3 ] (LnZY or Yb) proceeded to give deprotonation of an amide trimethylsilyl methyl group by the hypersilyl anion to afford the cyclometallated complexes [Ln{CH 2 SiMe 2 N(SiMe 3 )}(N 00 ) 2 ] (LnZY or Yb). The Yb-Si bond distance was found to be 3.039(2) Å by an X-ray diffraction study and this is almost identical to the Yb(II)-Si bond lengths observed in 34 and 41. The 171 Yb NMR spectrum of 50 exhibited a singlet at d 1057 ppm with satellites (J YbSi Z716 and 2 J YbSi Z8.9 Hz) and the 29 Si NMR spectrum exhibited three signals at d K148.6 (central silicon, J SiSi Z27.9 Hz), K13.9 (silyl amide silicon) and K4.7 (trimethylsilyl silicon) ppm.

(d ) Lanthanide-pnictide complexes
Only two lanthanide metal/metalloidal pnictide complexes have been structurally authenticated, both reported by Evans. The reaction of [Sm(h 5 -C 5 Me 5 ) 2 ] with [Bi(Ph) 3 ] afforded a mixture of products that could be separated by differential solubility in toluene and hexane. The hexane-soluble portion The Sm-Bi bond distances that were found to be 3.265(1), 3.283(1), 3.291(1) and 3.311(1) Å, and are all well within the sum of the covalent radii for samarium and bismuth (3.46 Å) (Cordero et al. 2008). The Bi-Bi distance of 2.851(1) Å is within the sum of the covalent radii of bismuth (2.96 Å) and, considering the Bi-Bi unit is formally dianionic, it is consistent with some multiple bond character, since it is shorter than typical Bi-Bi single bond distances, although it is not particularly short either. In a reaction analogous to the preparation of 51, but with a different outcome, the complex [{Sm(h 5 -C 5 Me 5 ) 2 } 3 (m-h 2 : h 2 : h 1 -Sb 3 )(THF)] (52) (figure 15) was isolated from the reaction between [Sm(h 5 -C 5 Me 5 ) 2 ] and [Sb(Bu n ) 3 ] as dark red crystals (Evans et al. 1992). An X-ray diffraction study revealed a tri-samarium tri-antimony assembly where a Sb 3 3K unit bridges three samarium centres in a m-h 2 : h 2 : h 1 manner, and complex 52 can be regarded as a lanthanide adduct of a trapped Zintl trianion. The five Sm-Sb bond distances span a relatively narrow range [3.162(1)-3.205(1) Å], which are all well within the sum of the covalent radii for samarium and antimony (3.37 Å) (Cordero et al. 2008). The Sb-Sb distances are indistinguishable, measuring 2.689(1) and 2.686(1) Å, and this implies some multiple bond character to the Sb-Sb bonds.

(e ) Actinide-transition metal complexes
The first proposal for the synthesis of an actinide complex with a bond to a transition metal fragment was made in the report of the complex [U{Mn(CO) 5 } 4 ] (53) by Stone (Bennett et al. 1971). However, the bright orange solid that was isolated was extremely air sensitive and apparently decomposed during analysis. For example, the only signals that could be unambiguously identified in the mass spectrum were [Mn(CO) n ] C fragments and attempts to record IR spectra in chloroform showed v CO bands consistent with [Mn(CO) 5 Cl]. Unfortunately, a definitive X-ray crystal structure was not obtained, so whether any U-Mn bonds were present in 53 remains unknown since the likelihood of isocarbonyl bridging to the oxophilic uranium(IV) centre must be regarded as high.
The first definite report of an unsupported actinide-transition metal bond was made by Marks ( figure 16; Sternal et al. 1985). The Th-Ru complex [Th(h 5 -C 5 Me 5 ) 2 (I){Ru(h 5 -C 5 H 5 )(CO) 2 }] (54) was isolated as pale yellow needles from the reaction between [Th(h 5 -C 5 Me 5 ) 2 (I) 2 ] and Na[Ru(h 5 -C 5 H 5 )(CO) 2 ]. Interestingly, the bis-ruthenium derivative was not accessible, even when an excess of the ruthenium reagent was employed, which indicates the sterically crowded coordination sphere at thorium. The FTIR v CO bands at 2023 and 1959 cm K1 are comparable with the v CO bands observed for 4. An X-ray diffraction study revealed a Th-Ru bond length of 3.028(2) Å, which is well within the sum of the covalent radii for thorium and ruthenium (3.52 Å) (Cordero et al. 2008). A variable temperature NMR study of 54 revealed a dynamic equilibrium in solution that was rationalized as rotation about the Th-Ru bond. Viewed down the Th-Ru vector, the Ru-h 5 -C 5 H 5 and Th-I bonds may be arranged in an anti or gauche conformation, the latter of which may be represented by two enantiomorphous forms. In d 8 -THF, the ratio of anti to gauche forms was 1.4 : 1, whereas in d 8 -toluene the ratio was approximately 15 : 1. The FTIR spectrum of 54 indicated that heterolysis in THF did not occur. However, reaction of 54 with Bu t OH afforded the expected compounds [Th(h 5 -C 5 Me 5 ) 2 (I)(OBu t )] and [Ru(h 5 -C 5 H 5 )(CO) 2 H]. Marks also reported a range of complexes of the general form [An(h 5 -C 5 H 4 R) 3 {M(h 5 -C 5 H 5 )(CO) 2 }] (AnZTh, U; RZH, Me; MZFe, Ru) (Sternal & Marks 1987). None of these complexes were characterized by X-ray crystallography. However, FTIR data supported their formulations since the data were in good agreement with 4 and 54. NMR data of these complexes showed that a dynamic process analogous to 54 was operating, i.e. hindered rotation around the An-M vector, which was rationalized as a dynamic equilibrium between anti and gauche conformations with an average DG ‡ of 54.4 (G2.9) kJ mol K1 . Alcoholysis of these complexes gave the expected An-alkoxides and the corresponding transition metal hydride; reaction with ketones gave the corresponding An-enolate and transition metal hydride. A detailed study of the thermochemistry of An-M complexes (Nolan et al. 1991) concluded that complexes featuring an actinide-transition metal bond are promising candidates for thermodynamically plausible stoichiometric and catalytic reactions. The complex [Th(h 5 -C 5 Me 5 ) 2 (m-PPh 2 ) 2 Ni(CO) 2 ] (55) (figure 17) in which a supported Th-Ni bond was suggested was reported by Ryan (Ritchey et al. 1985). Complex 55 was isolated as an orange crystalline solid from the reaction between [Th(h 5 -C 5 Me 5 ) 2 (PPh 2 ) 2 ] and [Ni(COD) 2 ] (CODZcyclo-octadiene) under an atmosphere of carbon monoxide. The 31 P NMR spectrum exhibited a resonance at d 143 ppm, which is noteworthy because phosphides usually resonate in the range of d 50 to K300 ppm, whereas dimeric transition metal complexes that contain bridging phosphide units and a metal-metal bond more usually exhibit 31 P resonances in the range of d 300-50 ppm. However, these data alone are not convincing evidence of a Th-Ni bond, as exceptions to these correlations are known. The structure of 55 was determined by X-ray crystallography that revealed a Th-Ni distance of 3.206(2) Å, which is within the sum of the covalent radii of thorium and nickel (3.3 Å) (Cordero et al. 2008). The Th-P-Ni bond angles were noted to be particularly acute (approx. 768) and Ryan suggested that this was due to a Th-Ni bond, since complexes where no metal-metal bond is present exhibit significantly larger M-P-M 0 angles (larger than 1008) and the energetic penalty for an acute Th-P-Ni angle would be low since the Th-P bond may be considered predominantly ionic in nature. Ryan and Sattelberger subsequently reported the complex [Th(h 5 -C 5 Me 5 ) 2 (m-PPh 2 ) 2 Pt(PMe 3 )] (56), which exhibits a supported, dative donor-acceptor Th-Pt bond (figure 17; Hay et al. 1986). Complex 56 was prepared analogous to 55 from the reaction between [Th(h 5 -C 5 Me 5 ) 2 (PPh 2 ) 2 ] and [Pt(COD) 2 ] in the presence of trimethylphosphine and was isolated as a red-brown solid. Analogously to 55, 56 exhibited a resonance at d 149.3 ppm attributable to the phosphide centres and, additionally, a triplet at d K3.3 ppm attributed to the phosphine centre ( 2 J PP Z7 Hz). Each of these resonances was flanked by 195 Pt satellites (J PtP Z2459 and 2556 Hz, respectively); these coupling constants are low for platinum and were interpreted as evidence for a four-coordinate platinum centre, i.e. three Pt-P bonds and one Th-Pt bond. An X-ray diffraction study revealed a Th-Pt bond distance of 2.984(1) Å that is significantly within the sum of the covalent radii of thorium and platinum (3.42 Å) (Cordero et al. 2008). As for 55, acute Th-P-Pt bond angles were observed, which also indicates the presence of a Th-Pt bond. The nature of the Th-Pt bond was explored by ab initio calculations on a model complex that indicated the highest occupied molecular orbital (HOMO) to be a metal-metal bond with Pt 5d x 2 -y 2 and Th 6d x 2 -y 2 characters. Mulliken population analysis revealed charge densities of ThC1.48 and Pt 0.0 and orbital populations of Th 6p 6.00 , 7s 0.35 , 7p 0.35 , 6d 1.61 , 5f 0.22 and Pt 6s 0.77 , 6p 0.00 , 5d 9.21 . Taken together, the interaction can be regarded as a dative donor-acceptor bond formed from donation from platinum into the empty d-shell on thorium. The ferrocenophane complex [U(fc) 3 Li 2 (py) 3 ] (57; pyZC 5 H 4 N) was reported by Ephritikhine ( figure 18; Bucaille et al. 2000). Complex 57 was isolated as red crystals from pyridine after dissolution of the green-yellow precipitate isolated from the reaction between 1,1 0 -dilithioferrocene and uranium tetrachloride. An X-ray diffraction study revealed three uranium-iron interactions that range from 3.122(2) to 3.165(2) Å that are all shorter than the sum of the covalent radii for uranium and iron (3.28 Å) (Cordero et al. 2008 (Monreal et al. 2007) and [U{fc(NSiMe 2 Bu t ) 2 }(CH 2 C 6 H 5 )(OEt 2 )][BPh 4 ] (59)  figure 19). Complex 58 was prepared from the salt elimination/disproportion reaction between UI 3 (THF) 4 and [K 2 (OEt 2 ) 2 {fc(NSiMe 2 Bu t ) 2 } 2 ] in toluene or diethyl ether, followed by chemical oxidation with iodine and subsequent reaction with NaBPh 4 yielding 58 as black crystals. An X-ray diffraction study of 58 showed an average uranium-iron distance of 2.96(1) Å that is well within the sum of covalent radii for uranium and iron (3.28 Å) (Cordero et al. 2008). Cyclic voltametry studies showed distinct, reversible Fe(II)/Fe(III) oxidations, and this, together with magnetic, electron paramagnetic resonance and IR studies, indicated a direct uraniumiron interaction. A DFT study on a model cationic thorium system indicated modest iron and thorium contributions to the iron-thorium interaction (HOMO: 11.8% Th, 11.0% Fe, 10.5% Fe; HOMO-1: 5.2% Th, 37.3% Fe, 26.3% Fe; HOMO-2: 8.7% Th, 21.8% Fe, 15.1% Fe). Complex 59 was prepared in an analogous reaction to the first step in the preparation of 58. The reaction between UI 3 (THF) 4 and [K 2 (OEt 2 ) 2 {fc(NSiMe 2 Bu t ) 2 } 2 ] in THF affords a mixture of 58 and [U{fc(NSiMe 2 Bu t ) 2 }(I) 2 (THF)], which could be separated due to the insolubility of the latter in hexane. Straightforward salt elimination with benzyl potassium afforded the di-benzyl complex that was converted to 59 by treatment with [Et 3 NH][BPh 4 ]. The structure of 59 was confirmed by X-ray crystallography, which confirmed a uranium-iron distance of 3.071(2) Å that is longer than in 58, but still well within the sum of the covalent radii of uranium and iron (3.28 Å) (Cordero et al. 2008). A DFT study on the model complex [U{fc(NSiH 3 ) 2 }(CH 3 )(OMe 2 )] C showed a dative, uranium-iron donor-acceptor interaction similar to 58. The room temperature magnetic moment of 59 (2.4 m B ) was observed to be much lower than the di-benzyl precursor (3.2 m B ). The former value is much lower than the theoretical free ion moment for a 3 H 4 f 2 U(IV) centre (3.58 m B ), which indicates the quenching of the orbital-angular momentum by low symmetry or higher covalency than in the free ion. It was therefore proposed that the lower magnetic moment of 59 was due to the cationic nature of 59 resulting in a shorter uranium-iron distance and better orbital overlap.
(f ) Actinide-triel complexes The first actinide-triel complex has been recently reported by Arnold (figure 20). The reaction between one molar equivalent of [U(h 5 -C 5 H 4 SiMe 3 ) 3 ] with a quarter molar equivalent of [{Al(h 5 -C 5 Me 5 )} 4 ] in toluene at 608C afforded the complex [U(h 5 -C 5 H 4 SiMe 3 ) 3 {Al(h 5 -C 5 Me 5 )}] (60), which was isolated as dark brown crystals from cold (K808C) pentane (Minasian et al. 2008). An X-ray diffraction study revealed an unsupported uranium-aluminium bond with a uranium-aluminium bond distance of 3.117(3) Å, which is within the sum of the covalent radii of uranium and aluminium (3.17 Å) (Cordero et al. 2008). Signals attributable to bridging hydrides were not observed in the NMR and FTIR spectra, [M C ]-SiMe 3 was observed in the electron ionization (EI) mass spectrum, and CHCl 3 was not produced upon addition of CCl 4 to 60. A DFT study on the model complex [(h 5 -C 5 H 5 ) 3 U-Al(h 5 -C 5 H 5 )] showed a uranium-aluminium bonding interaction of essentially s-bonding character from a lone pair on aluminium to uranium and no occupied orbitals with uranium-aluminium p-bonding character were observed. Natural bond orbital analysis revealed charges of 1.982 and 0.651 for uranium and aluminium, respectively, at the B3LYP level of theory (unrestricted). A charge of 0.560 was computed for the aluminium fragment, indicating a net charge transfer of 0.091 to uranium, which is small. The Wiberg bond index of 0.436 indicates the bond is mainly electrostatic with some covalent character.
The first actinide-triel complex to exhibit sand p-bonding character, and the only other actinide-triel complex, has been very recently reported by Liddle and Jones (figure 21;Liddle et al. 2009 (61) as orange blocks following work-up and recrystallization. Analytical data were supportive of the proposed formulation; in particular, the FTIR spectrum exhibited no peaks where hydrides would be anticipated, [M C ]-THF was observed in the EI mass spectrum and CHCl 3 was not produced on addition of CCl 4 . Single crystal X-ray diffraction showed two molecules to be present in the asymmetric unit of crystalline 61, which were found to be almost identical except for differing uranium-gallium bond lengths. Uranium-gallium bond distances of 3.221(2) and 3.298(2) Å were observed, which are both marginally longer than the sum of the covalent radii of uranium and gallium (3.18 Å) (Cordero et al. 2008). This may reflect the sterically demanding nature of the gallyl heterocycle and Tren TMS fragments. The room temperature magnetic moment of 2.46 m B is significantly below the theoretical free ion value of 3.58 m B for the 3 H 4 ground state of f 2 uranium (cf. 59). A DFT study on the model complex [{N(CH 2 CH 2 NSiH 3 ) 3 }U{Ga[N(2,6-Me 2 C 6 H 3 ) CH] 2 }(OMe 2 )] showed not only the anticipated s-bonding, but also an occupied p-bonding component ( figure 22). The former component (HOMO-3) was found to be an essentially sp-hybridized gallium lone pair donation into vacant uranium valence orbitals. The latter component (HOMO-2) was shown to be p-donation from the gallyl heterocycle into principally the empty U 5f z 2 y orbital. Since the gallyl heterocycle in 61 is valence isoelectronic and isolobal to the popularly employed NHC ligand class, 61 raises the possibility that p-donation of NHCs to metals may be more widespread than previously recognized. Indeed, the nature and capacity of NHCs to engage in p-bonding (as donor or acceptor) has only been debated in recent years. This invokes the possibility that the selectivity of NHCs for uranium over the lanthanides in solution extraction studies could originate from selective p-donor phenomenon that has not yet been observed for the lanthanides. However, it should be noted that the nature of the uranium- gallium bond in 61 is still mainly electrostatic, since the orbital contribution to the total attractive interaction is 28 per cent. Additionally, complex 61 represents a model for the U(IV)-CO K, unit, which is yet to be isolated.
(g ) Actinide-tetrel complexes The first, and to date only, structurally characterized uranium-silicon bond was reported by Cummins (figure 23; Diaconescu et al. 2001). Red crystals of [{(3,5-Me 2 C 6 H 3 )N(Bu t )} 3 U{Si(SiMe 3 ) 3 }] (62) were isolated from diethyl ether after work-up from the reaction between [{(3,5-Me 2 C 6 H 3 )N(Bu t )} 3 U(I)] and [(THF) 3 Li{Si(SiMe 3 ) 3 }]. An X-ray diffraction study revealed a uranium-silicon bond length of 3.091(3) Å, which is slightly longer than the sum of the covalent radii of uranium and silicon (3.07 Å) (Cordero et al. 2008), and which may reflect the sterically demanding nature of the amide and silyl groups. A DFT study on a model of complex 62 [(H 2 N) 3 U-SiH 3 ] showed a uranium-silicon bonding orbital of mainly silicon 3p z (49%) and uranium 6d z 2 (15%) composition. Surprisingly, the uranium-silicon bond in 62 was found to not react with CO, H 2 , nitriles or  isonitriles, with the exception of Pr i NC that, when added in 1.5-2 equivalents, gave a complex mixture of intractable products. The first, and thus far only, structurally authenticated uranium-tin bond was reported by Porchia (figure 24;Porchia et al. 1986). The reaction between [(h 5 -C 5 H 5 ) 3 U(NEt 3 )] and HSnPh 3 afforded a brown powder in quantitative yield. Interestingly, reaction of HSnPh 3 with [(h 5 -C 5 H 5 ) 3 U(Me)] or [(h 5 -C 5 H 5 ) 3 U(BH 4 )], or reaction of [(h 5 -C 5 H 5 ) 3 U(Cl)] with [Li(SnPh 3 )], appeared too slow or was complicated by the formation of other inseparable products. Similar experiments with HSnBu 3 were apparently unsuccessful. Solutions of the brown powder afforded stable, bright green solutions in aromatic solvents, but rapid reaction occurred in diethyl ether or THF. However, brown crystals of the heavy alkyl complex [(h 5 -C 5 H 5 ) 3 U(SnPh 3 )] (63) were obtained and subjected to an X-ray diffraction study to reveal a uranium-tin bond length of 3.166(1) Å, which is well within the sum of the covalent radii for uranium and tin (3.335 Å) (Cordero et al. 2008). Porchia also reported spectroscopic characterization of the uranium-silicon (Porchia et al. 1989) and uranium-germanium (Porchia et al. 1987) complexes [(h 5 -C 5 H 5 ) 3 U(EPh 3 )] (EZSi: 64; Ge: 65). Complex 64 was prepared from [(h 5 -C 5 H 5 ) 3 U(Cl)] and [Li(SiPh 3 )], but it was noted that the lithium reagent had to be freshly prepared to obtain 64 in good yield; complex 65 was prepared likewise from [(h 5 -C 5 H 5 ) 3 U(Cl)] and freshly prepared [K(GePh 3 )]. Both compounds 64 and 65 were found to react cleanly with HSnPh 3 to afford 63, which was attributed to the greater acid strength of HSnPh 3 compared with the silicon and germanium congeners. Complex 64 was also found to be unstable during preparation; if the synthesis was performed at room temperature or the reaction mixture (K308C) was allowed to warm, the siloxide compound [(h 5 -C 5 H 5 ) 3 U(OSiPh 3 )] was found to form, which was confirmed by its independent synthesis and characterization by X-ray crystallography, and 1 H NMR and FTIR spectroscopies. Although no reactivity studies of 63 have been reported, complexes 64 and 65 were found to react with 2,6-dimethylphenyl isocyanide to give the corresponding imino-silyl or -germyl insertion products.
(h ) Actinide-pnictide complexes No molecular complexes containing uranium-antimony or -bismuth bonds, unsupported or supported, dative or polarized-covalent, have yet been reported.

Concluding remarks
Although this review has covered 65 f-element-metal complexes, this number is dwarfed by the number of metal-metal complexes in the d-and p-blocks.
Furthermore, when f-element-metalloidal complexes are exempted from the count, there are only ten dative and nine covalent lanthanide-metal complexes and six dative and four covalent actinide-metal complexes reported to date. This relative paucity underscores the fact that the area is underdeveloped and will require much more development before reaching maturity. Nevertheless, some of the complexes reported already exhibit novel bonding and reactivity patterns that have far-reaching implications in a wider context. The elaboration of this chemistry holds great promise for advancing our understanding of the fundamentals of structure and bonding at the bottom of the periodic table, which is historically behind all other areas, and for the generation of new catalytic cycles that may yet provide feasible new chemistry that is unavailable by other, more traditional, methodologies.
The author thanks the Royal Society for the award of a University Research Fellowship and the UK EPSRC and University of Nottingham for funding.