One ligand fits all: lanthanide and actinide sandwich complexes comprising the 1,4-bis(trimethylsilyl)cyclooctatetraenyl (=COT 00 ) ligand †‡

The series of anionic lanthanide( III ) sandwich complexes of the type [Ln(COT 00 ) 2 ] (cid:2) (COT 00 = 1,4-bis(trimethyl-silyl)cyclooctatetraenyl dianion) has been largely extended by the synthesis of eight new derivatives ranging from lanthanum to lutetium. The new compounds [Li(DME) 3 ][Ln(COT 00 ) 2 ] (Ln = Y ( 1 ), La ( 2 ), Pr ( 3 ), Gd ( 4 ), Tm ( 6 ), Lu ( 8 )) and [Li(THF) 4 ][Ln(COT 00 ) 2 ] (Ln = Ho ( 5 ), Tm ( 7 )) were prepared in good yields following a straightforward synthetic protocol which involves the treatment of LnCl 3 with 2 equiv. of in situ prepared Li 2 COT 00 in either DME (=1,2-dimethoxyethane) or THF. The neutral actinide sandwich complexes An(COT 00 ) 2 (An = Th ( 9 ), U ( 10 )) and An(COT 000 ) 2 (COT 000 = 1,3,6-tris(trimethylsilyl)cycloocta-tetraenyl dianion; An = Th ( 11 ), U ( 12 )) were synthesized in a similar manner, starting from ThCl 4 or UCl 4 , respectively. The COT 00 ligand imparts excellent solubility even in low-polar solvents as well as excellent crystallinity to all new compounds studied. All twelve new f-element sandwich complexes have been structurally authenticated by single-crystal X-ray diﬀraction. All are nearly perfect sandwich complexes with little deviation from the coplanar arrangement of the substituted COT 00 rings. Surprisingly, all six [Li(DME) 3 ][Ln(COT 00 ) 2 ] complexes covering the entire range of Ln 3+ ionic radii from La 3+ to Lu 3+ are isostructural (space group P % 1). Compound 10 is the first uranocene derivative for which 13 C NMR data are reported.


Introduction
Second only to the omnipresent cyclopentadienyl complexes, the dianionic 10p-cyclooctatetraenyl ligand C 8 H 8 2À , commonly abbreviated as COT, plays an important role in the organometallic chemistry of lanthanides and actinides for almost 50 years. There is a general understanding that the large, flat C 8 H 8 2À ring is ideally suited for overlapping with the f-orbitals of the large lanthanide and actinide ions. 1 Early work in this area was mainly focused on complexes bearing unsubstituted COT ligands. 2 Scheme 1 shows some prototypical lanthanide COT complexes which are considered milestones in the development of organolanthanide chemistry using COT ligands.
[Li(TMEDA)] 2 (COT 00 ) was shown to be an inverse sandwich complex with the two Li + ions coordinated to the bridging 1,4-bis(trimethylsilyl)cyclooctatetraene dianion ring in an Z 3 -allyllike fashion. 28a [Li(THF) 2 ] 2 [Li 2 (COT 00 ) 2 ] contains two Li + ions sandwiched between two COT 00 rings and two Li(THF) 2 + units attached to the outside of the COT 00 rings. 28b In the present study, however, it was found to be more convenient to use in situ-prepared THF solutions of Li 2 COT 00 rather than isolated samples. Accordingly, the anionic lanthanide sandwich complexes 1-8 were prepared by treatment of selected anhydrous lanthanide trichlorides, LnCl 3 , with 2 equiv. of Li 2 COT 00 in THF solution as outlined in Scheme 3. In the case of the THF adducts [Li(THF) 4 ][Ln(COT 00 ) 2 ] (Ln = Ho (5), Tm (7)), purification was achieved by recrystallization of the crude products from toluene. The DME adducts [Li(DME) 3 ][Ln-(COT 00 ) 2 ] (Ln = Y (1), La (2), Pr (3), Gd (4), Tm (6), Lu (8)) were obtained by extraction of the reaction products with toluene followed by recrystallization from DME after addition of n-pentane. The products were isolated in moderate to good yields (57-75%) in the form of yellow or yellow-green (Tm: red), highly air-sensitive crystalline solids. It has been noted earlier that DME is the solvent of choice for crystallizing these anionic lanthanide sandwich complexes. 14,20 The DME solvates are readily crystallized and the resulting crystals do not lose DME even under vacuum or upon prolonged storage in the dry-box. In contrast, crystals of the THF adducts are less stable with respect to loss of solvent and become opaque upon storing in the dry-box. Meaningful NMR spectra could be obtained only for the diamagnetic products [Li(DME) 3 ][Y(COT 00 ) 2 ] (1), [Li(DME) 3 ][La(COT 00 ) 2 ] (2), and [Li(DME) 3 ][Lu(COT 00 ) 2 ] (8) as well as for the paramagnetic praseodymium derivative 3. In all four cases the 1 H and 13 C NMR data were in good agreement with the formation of the expected anionic sandwich complexes. The observation of only one signal in the 29 Si NMR spectra (1: 0.7 ppm, 2: 0.5 ppm, 3: À46 ppm, 8: 0.8 ppm) indicated high purity of the materials. Moreover, the IR spectra of the DME adducts on one hand and the THF adducts on the other hand were found to be almost superimposable. All the new complexes were structurally characterized through single-crystal X-ray crystallography. Crystallographic data for 1-8 are summarized in Tables 1 and 2. The most significant bond lengths and angles are listed in Table 3.
As can be seen from Table 3  In a similar manner, the closely related neutral actinidocenes An(COT 00 ) 2 (An = Th (9), U (10)) have also been prepared. As outlined in Scheme 4, these sandwich complexes were prepared in a straightforward manner by reaction of anhydrous ThCl 4 or UCl 4 with 2 equiv. of in situ-prepared Li 2 COT 00 . Due to the high solubility of all the reactants in THF, the reactions were finished after 2 h stirring at r.t. In contrast, reactions of AnCl 4 with the unsubstituted K 2 COT normally take days. 11,12 Bright yellow Th(COT 00 ) 2 (9) and dark green (dichroitic red/green) U(COT 00 ) 2 (10) were both isolated in high yields of ca. 80%. Purification could be achieved either by highvacuum sublimation at 240 1C or by slow crystallization from the oily crude products. In this context it is interesting to note that Murugesu et al. very recently prepared compound 10 via a two-step synthesis where U III I 3 (1,4-dioxane) 1.5 and [Li(THF) 2 ] 2 [Li 2 (COT 00 ) 2 ] 28b were first combined in THF to afford the anionic uranium(III) sandwich complex [Li(DME) 3 ]-[U(COT 00 ) 2 ] which was then oxidized to the uranium(IV) sandwich 10 using FeCl 2 . 27 For comparison, two neutral actinide sandwich complexes comprising the bulky 1,3,6-tris(trimethylsilyl)cyclooctatetraenyl ligand (COT 0 0 0 ) have also been prepared. These compounds have earlier been mentioned in two communications, but structural characterization through X-ray diffraction was lacking. 24 Both compounds were prepared according to the straightforward synthetic protocol illustrated in Scheme 5. In this case, the use of the potassium precursor K 2 COT 0 0 0 provided products 11 and 12 in yields of around 80% after crystallization from concentrated solutions in n-pentane. Like their tetrasubstituted   congeners 9 and 10, thorium compound 11 forms bright yellow crystals, while crystals of 12 appear dichroitic red/green. Both complexes are highly soluble in common organic solvents, including hydrocarbons. All four silyl-substituted actinidocenes 9-12 have been structurally characterized through single-crystal X-ray diffraction. Crystallographic data for 9-12 are summarized in Table 2; selected bond lengths and angles are listed in Table 4. The molecular structures are depicted in Fig. 3 and 4. As can be seen from the structural data listed in Table 4, the overall structural features of all four actinidocene derivatives studied here are very similar. According to the unsymmetrical substitution pattern on the cyclooctatetraenyl rings leading to steric interactions, all complexes show a slight distorsion from the ideal linear arrangement with Ctr-M-Ctr angles of about 1741. As expected, evidence for actinide contraction is found which is reflected in B5 pm shorter M-C as well as in B7 pm shorter M-Ctr distances in the uranium complexes as compared to the thorium species (Table 5).
In the following, the structural and spectroscopic characterization of 10 as a typical example will be discussed in more detail. The molecular structure of 10 can be clearly described as being of the well-known uranocene type (Fig. 5). Accordingly, in the molecular structure the central uranium atom is placed between the two cyclooctatetraenyl rings with U-Ctr distances of 1.913 or 1.921 Å, comparable to previously reported uranocene derivatives 29 (Table 4). However, the trimethylsilyl substituents in 1,4-positions of the cyclooctatetraenyl ring lead to an arrangement in the solid state where on one side of the molecule a stronger steric interaction between the two cyclooctatetraenyl rings results. Si1 and Si4 are found to be in closer steric environment than Si2 and Si3, giving rise to a significant repulsion on this side of the rings. This has an influence on the bond lengths and angles in that the two cyclooctatetraenyl rings do not bind symmetrically to the central uranium atom. The U-C bond distances cover a range between 2.642 and 2.690(4) Å with the longer bond lengths found on the side with the stronger steric interactions, whereas the shortest U-C bond length is observed for U1-C22 with 2.642(4) Å. Accordingly, the two cyclooctatetraenyl rings are not coordinated coplanar with Table 4 Overview of all known anionic lanthanide sandwich complexes of the type [Li(THF) 4 ][Ln(COT 00 ) 2 ] (denoted THF) and [Li(DME) 3 ][Ln(COT 00 ) 2 ] (denoted DME). X: compounds described in this work Scheme 4 Synthetic route to the neutral actinidocenes An(COT 00 ) 2 (An = Th (9), U (10)).
Scheme 5 Synthesis of the neutral actinidocenes An(COT 0 0 0 ) 2 (An = Th (11), U (12)).   respect to the uranium center. This results in a Ctr-U-Ctr angle of 7.01 and a tilt angle between the two ring planes of 7.41 with the opening to the side of Si1/Si4 (Table 4 and Fig. 5). This is further reflected in the corresponding distances between opposing carbon atoms of the two COT 00 rings in the staggered structure. With 4.047 and 4.070 Å the distances C1-C17 and C2-C18 are remarkably longer than those between C5 and C21 or C6 and C22, which are with 3.627 or 3.614 Å significantly shorter. These structural findings clearly show that compound 10 shows typical uranocene structural features 29a but with a significant distorsion caused by steric effects due to the trimethylsilyl substituents at the COT rings. A significantly stronger tilting of the two cyclooctatetraenyl rings has been observed in the 1,4-bis(triphenylsilyl)-substituted system where ring-to-ring C-C distances between 3.468 and 4.247 Å and a tilt angle of 11.41 have been found. 30 The spectroscopic data of the complexes 9-12 are in good agreement with their structural features. As expected, the IR spectra of 9-12 are all very similar, showing the comparable molecular constitution of these actinidocenes. Frequencies arising from the COT 00 ligand increase slightly upon complexation as compared to K 2 COT 00 . However, the spectra are more complicated than those of the unsubstituted actinidocenes as the SiMe 3 -substituents give rise to strong absorptions themselves and cause a distorsion from the ideal D 8h -symmetry observed in the actinidocenes, leading to a higher number of observed frequencies. 31 However, the general congruency of the IR and FIR spectra clearly shows the similarity in the structural features of the complexes 9-12. The other spectroscopic data will be highlighted taking again compound 10 as example. In contrast to the corresponding Th-complex 9, the uranocene derivative 10 exhibits a 5f 2 -electron configuration causing paramagnetism and an intensely red color in transmission. These findings are confirmed by the UV-vis data (Fig. 6), showing that below 450 nm the absorption of the complex is strongly increasing.
The absorptions at 592, 618, 635 nm are caused by strong charge transfer transitions typical for actinocene complexes, however being more intense in symmetry-distorted systems. 32 In the range of 800 to 2000 nm, the UV-vis spectrum does not show any significant differences between the solid state and the solution, indicating that the solid state structure is retained in solution and no adduct formation takes place. Accordingly, the absorptions at 980, 1322, 1486, 1710, 1755, 1793, 1865 nm are caused by f-f transitions, which are characteristic of U(IV)-organometallics. 33 The f-f transitions are in this case of higher intensity than for the unsubstituted uranocene due to the observed distortion of the complex symmetry by the SiMe 3 substituents, which causes an increase in the intensity for the symmetry-forbidden f-f transitions. These are, however, between 10 to 100 times less intense than the charge transfer absorptions.
The paramagnetism of 10 is also clearly seen from its NMR data (Fig. 7-9), where for all signals a typical upfield shift is observed. 7a,34 In good agreement with the solid state structure, the 1 H NMR spectrum of 10 (Fig. 7) shows four well-separated singlets at À9.99, À25.20, À39.63, and À45.62 ppm. The latter three each correspond to four ring protons, whereas the first  resonance can be clearly assigned to the protons of the SiMe 3 substituents. In the two-dimensional HH-correlated spectrum the resonances at À39.63 and À45.62 ppm (b-position to the SiMe 3 -substituents) are assigned to the protons in the (CH) 4chain of the COT 00 ring, whereas the resonance at À25.20 ppm corresponds to the ring protons positioned between the two trimethylsilyl substituents in 1,4-positions (Fig. 8). This assignment is in good agreement with the published data where a strong influence of the paramagnetism on the chemical shifts in uranocene derivatives is described. 34 However, in this paper, for the first time, the 13 C chemical shifts of a uranocene complex are reported. The carbon resonance of the SiMe 3 groups was localized at À3.5 ppm. The proton resonance at À25.20 ppm exhibits a cross peak at 325.9 ppm in the 13 C frequency, whereas the two coupling H-atoms of the aromatic ring at À39.63 ppm and À45.62 ppm give rise to carbon resonances as well at low field shifts with 293.8 and 270.3 ppm, respectively (Fig. 9). The observation of carbon frequencies at these low fields is in agreement with theoretical predictions. 35

Conclusions
In summarizing the results reported here, the series of anionic lanthanide(III) sandwich complexes of the type [Ln(COT 00 ) 2 ] À (COT 00 = 1,4-bis(trimethylsilyl)cyclooctatetraenyl dianion) has been largely extended through the synthesis of eight new derivatives ranging from lanthanum to lutetium. Surprisingly, neither the ionic radius nor the oxidation state of the f-element ion (Ln 3+ / An 4+ ) have a pronounced influence on the structural features of the compounds [Li(DME) 3 ][Ln(COT 00 ) 2 ] (1-8; Ln = Y, La, Pr, Gd, Tm, Lu), [Li(THF) 4 ][Ln(COT 00 ) 2 ] (5, 7; Ln = Ho, Tm), An(COT 00 ) 2 (9, 10; An = Th, U) and An(COT 0 0 0 ) 2 (11,12; An = Th, U). In all cases the slight deviation from the ideal sandwich structure is in the same range. Through this comparative study anionic sandwich complexes containing the [Ln(COT 00 ) 2 ] À anions have now become available for the entire series of rare-earth metals. This should allow for more detailed investigations e.g. of the magnetic properties in the course of future studies.

Crystal structure determination
The intensity data of the lanthanide sandwich complexes 1-8 were collected on a Stoe IPDS 2T diffractometer with MoKa radiation. The data were collected with the Stoe XAREA program using o-scans. 39 The space groups were determined with the XRED32 program. The structures were solved by direct methods (SHELXS-97) and refined by full matrix leastsquares methods on F 2 using SHELXL-97. 40 Data collection parameters are summarized in Tables 1 and 2. Single-crystal X-ray analyses of the actinide complexes 9-12 were performed on a Bruker Apex II Quazar diffractometer at given temperature, collecting two or four spheres of data with an irradiation time of 10 to 40 s per frame, applying a combination of oand j-scans. Maximum y-values were in the range of 281. Completeness of data to y r 251 was higher than 99%. For more information refer to Table 2. Integration of the data proceeded with SAINT, 41 the data were corrected for Lorentzand polarisation effects, and an experimental absorption correction with SADABS 41 was performed. For searches relating to single-crystal X-ray diffraction data, the Cambridge Structural Database was used. The structures have been solved by direct methods and refined to a minimum R-value with SHELXL-2013 42 via full-matrix least-squares on F 2 . In the case of compound 9, a second type of crystals could be isolated with a different elementary cell showing a strong disorder. The data have been deposited at the CCDC with the CCDC 1049928 but will not be discussed here in detail as due to the disorder the overall standard deviations for all values are significantly higher.