Group 4 Metal and Lanthanide Complexes in the Oxidation State +3 with Tris(trimethylsilyl)silyl Ligands

A number of paramagnetic silylated d1 group 4 metallates were prepared by reaction of potassium tris(trimethylsilyl)silanide with group 4 metallates of the type K[Cp2MCl2] (M = Ti, Zr, Hf). The outcomes of the reactions differ for all three metals. While for the hafnium case the expected complex [Cp2Hf{Si(SiMe3)3}2]− was obtained, the analogous titanium reaction led to a product with two Si(H)(SiMe3)2 ligands. The reaction with zirconium caused the formation of a dinuclear fulvalene bridged complex. The desired [Cp2Zr{Si(SiMe3)3}2]− could be obtained by reduction of Cp2Zr{Si(SiMe3)3}2 with potassium. In related reactions of potassium tris(trimethylsilyl)silanide with some lanthanidocenes Cp3Ln (Ln = Ce, Sm, Gd, Ho, Tm) complexes of the type [Cp3Ln Si(SiMe3)3]− with either [18-crown-6·K]+ or the complex ion [18-crown-6·K·Cp·K·18-crown-6] as counterions were obtained. Due to d1 or fn electron configuration, unambiguous characterization of all obtained complexes could only be achieved by single crystal XRD diffraction analysis.


INTRODUCTION
Investigations on the chemistry of group 4 silyl complexes were started in the late 1960s, with some work on silyl titanium chemistry 1−6 and Lappert's contributions of zirconocene and hafnocene complexes. 7−10 Systematic studies of zirconocene and hafnocene silyl complexes were carried out by Tilley and co-workers, 11−17 who especially studied aspects of σ-bond metathesis and the catalytic dehydrocoupling polymerization of hydrosilanes catalyzed by these compounds. 16−19 While in the initial papers by Harrod and co-workers 20−23 on the dehydrocoupling polymerization of hydrosilanes titanium was acting as the catalytically active element, Tilley's mechanistic studies were carried out using hafnium or zirconium. Starting out from CpCp*M(Cl)Si(SiMe 3 ) 3 (M = Zr, Hf) it was shown that σ-bond metathesis reaction with a hydrosilane leads to (Me 3 Si) 3 SiH and a new metal silyl complex, which in reaction with another hydrosilane forms a disilane and a metal hydride. 18 We were curious whether the same chemistry would also work for titanium but quickly realized that CpCp*Ti(Cl)Si-(SiMe 3 ) 3 or even Cp 2 Ti(Cl)Si(SiMe 3 ) 3 is not easily available. NMR spectroscopic analysis of the reactions of CpCp*TiCl 2 or Cp 2 TiCl 2 with (Me 3 Si) 3 SiK (Scheme 1) did not show the expected signals but only a number of oligosilanes. However, upon crystallization of the reaction mixture an NMR-silent silylated titanium(III) species was detected using single crystal XRD analysis. 24 Subsequent studies revealed that silylated titanocenes with Ti(IV) tend to undergo reductive elimination to "Cp 2 Ti(II)", which in a subsequent comproportionation with Cp 2 TiCl 2 gives Cp 2 TiCl or its respective KCl adduct K[Cp 2 TiCl 2 ] which in a final step can react with (Me 3 Si) 3 SiK to K [Cp 2 Ti(Cl)Si(SiMe 3 ) 3 ]. 25 In order to study the chemistry of silylated Cp 2 Ti(III) complexes, we reacted α,ω-oligosilanyldiides with (18-crown-6)·K [Cp 2 TiCl 2 ] or (tmeda)·Li- [Cp 2 TiCl 2 ] to titanacyclosilanes with titanium in the oxidation state +3. 25 Further investigations revealed that analogous metallacyclosilanes could be obtained also with Zr(III) and Hf(III). 25 In the current paper we wish to report on reactions of (Me 3 Si) 3 SiK with (18-crown-6)·K [Cp 2 MCl 2 ] (M = Ti, Zr, Hf) to obtain d 1 -complexes of the type K[Cp 2 M{Si-(SiMe 3 ) 3

} 2 ].
Examples of compounds with lanthanide−silicon (Ln−Si) bonds are still scarce. Among all 4f-elements, samarium and the late lanthanide metals ytterbium and lutetium are best investigated for this class of compounds. In their landmark contributions, Schumann and co-workers were the first to employ a common method for the preparation of early transition-metal complexes, treating rare-earth halide complexes with the lithium silanide Me 3 SiLi. Reactions with complexes of the type Cp 2 Ln(μ-Cl) 2 30 More recently, Sgro and Piers reported the synthesis of yttrium and gadolinium silyl complexes by reacting potassium tris-(trimethylsilyl)silanides with the respective triiodidies. 31 Earlier, Radu and Tilley prepared similar compounds by σ-bond metathesis reactions of Cp* 2 LnCH(SiMe 3 ) 2 with SiH 2 (SiMe 3 ) 2 to obtain Cp* 2 LnSiH(SiMe 3 ) 2 (Ln = Sm, Nd). 36−38 Our own attempts in this field were mostly concentrating on reactions of oligosilanides with iodides of Sm(II), Yb(II), and Eu(II). 33−35 However, we also reported a study on the synthesis of metallacyclosilanes with lanthanidocenes, which were formed as ate-complexes reactions of α,ωoligosilanyl dianions with lanthanidocenes Cp 3 Ln. 29 In the course of these reactions, one cyclopentadienyl group was eliminated and a lanthanide ate-complex with two Cp and two silyl ligands was obtained.
Starting with titanium, we initially attempted reaction of 2a with (tmeda)·Li [Cp 2 TiCl 2 ] without much success. Eventually, we found that optimum conditions require reaction of 1a at low temperature with donor-free 2a. Nevertheless, the reaction did not give the expected product K[Cp 2 Ti{Si(SiMe 3 ) 3 } 2 ], but instead complex 3a with two H(Me 3 Si) 2 Si groups was obtained (Scheme 2). Due to the fact that 3a is a paramagnetic NMRsilent complex, its identity could only be determined using single crystal XRD analysis ( Figure 1). The experiment was repeated several times to exclude possible hydrolysis as a cause for the Me 3 Si to H exchange. It is not quite clear how the trimethylsilyl groups are lost; however, there is some precedence for similar reactivity that was observed as a side reaction in the synthesis of zirconium disilene complexes. 41 Complex 3a crystallizes in the monoclinic space group P2 1 / c, where the counterion to the ate-complex [Cp 2 Ti{Si-(SiMe 2 H) 3 } 2 ] − is the inverse sandwich [{K·(18-crown-6)} 2 Cp] + in which a Cp − is coordinated on both sides by a potassium ion which on the outer side is coordinated by a crown ether unit. We and others have observed this counterion already before for group 4, 24,25 cobalt, 42 iron, 43 and fblock 29,44−46 ate-complexes and Zintl anions. 47,48 Most of the known Si−Ti bond distances containing titanocenes involve Ti(III). However, Si−Ti bond lengths seem to be much more sensitive to the substitution pattern on silicon and sterics than to the oxidation state of titanium. This can be derived nicely from the series Cp 2 Ti(PMe 3 )SiH 3 (2.594(2) Å), 49 Cp 3 SiTi(NEt 2 ) 3 }. 54 For the case of the analogous reaction of (18-crown-6)· K[Cp 2 HfCl 2 ] (1c) with donor-free (Me 3 Si) 3 SiK (2a) no unpredicted side reaction was observed and the expected product K[Cp 2 Hf{Si(SiMe 3 ) 3 } 2 ] 3b with two (Me 3 Si) 3 Si groups was obtained (Scheme 2). Again, the structure of 3b was unambiguously determined using single crystal XRD analysis ( Figure 2). For 3b, which crystallizes in the monoclinic space group P2 1 , K·(18-crown-6) acts as the cationic counterion. This structure provides the opportunity of direct comparison of the anionic d 1 -complex 3b with its neutral d 0 -counterpart Cp 2 Hf{Si(SiMe 3 ) 3 } 2 . 55 Surprisingly, we found that the Hf−Si bond lengths of 3b (2.821(6) and 2.829(6) Å) are shorter than those of the neutral compound, for which a distance of 2.850(4) Å was observed. This is really unexpected as the general trend within the cyclic disilylmetallates is that bonds to the d 1 -metal atoms are longer than those to the d 0metal atoms in neutral complexes. 25 For the cyclic cases of hafnacyclopentasilanes the neutral compounds featured Si−Hf bond lengths of 2.791(14) and 2.823(15) Å 56 compared to 2.849(2) Å for the analogous Hf(III) compound. 25 The Si− Hf−Si angle of 3b is 127.56 (16)°which is larger than 117.79 (14)°as was observed for the bis[tris(trimethylsilyl)silyl]hafnocene. 55 Again this is against the trend that we observed for the cyclic compounds where the neutral compound exhibits a Si−Hf−Si angle of 96.41(5)°5 6 whereas the respective hafnate displayed 89.10(6)°. 25 It is likely that the two unusual observations are connected. A closer inspection of Hf−Cp distances seems to provide an explanation. For the cyclic Hf(III) 25 and Hf(IV) 56 cases the Cp centroid -Hf distances are 2.187/2.189 Å 25 and 2.181 Å, 56 respectively. As expected, the values for Hf(III) are slightly longer but in essence the numbers are similar. For Cp 2 Hf{Si-(SiMe 3 ) 3 } 2 the Cp centroid −Hf distance is 2.177 Å, 55 while for 3b this distance is elongated to 2.204/2.212 Å. We assume that the large Si−Hf−Si angle of 3b causes some population of antibonding Cp-Hf orbitals. This increases the Cp−Hf distance and thus allows the Si(SiMe 3 ) 3 substituents to approach closer.
The potassium ion of the cationic counterion part of 3b displays a weak interaction to one of the methyl groups ( Figure  2). This is quite common, and we 57 and also others 58,59 have observed similar potassium C−H interaction in the solid state on occasion.
The attempt to prepare the zirconium analog of 3b by reaction of 18-crown-6·K[Cp 2 ZrCl 2 ] (1c) with (Me 3 Si) 3 SiK took an entirely different course. Despite the fact that we previously observed that reaction of α,ω-oligosilanyl dianions with 1c gave the expected metallacyclosilane with Zr(III), compound 3c as isolated from the reaction of 1c with two equivalents of (Me 3 Si) 3 SiK is a dinuclear complex with a fulvalene ligand bridging the two Zr atoms, each of which is carrying an additional Cp ligand. The two Zr atoms are further bridged by a chloride ligand and one Zr atom bears a tris(trimethylsilyl)silyl ligand, whereas the other Zr has a bond to the Cp ligand of its neighbor (Scheme 3, Figure 3).
While we do not know exactly how this complex is formed, it seems reasonable to assume that the expected intermediate K [Cp 2 Zr(Cl)Si(SiMe 3 ) 3 ] is involved in its formation. It is likely that this compound is not as easily silylated as the intermediate in the reactions with the α,ω-oligosilanyl dianions since the two silanide units are not connected and thus for the current case an entropic disadvantage can be expected. If access to the Zr atom of K[Cp 2 Zr(Cl)Si(SiMe 3 ) 3 ] is sterically hindered, (Me 3 Si) 3 SiK might act as a base, deprotonating a Cp ligand, and this way the reaction takes a different course than

Inorganic Chemistry
Article expected. This assumption is supported by the NMR spectroscopic detection of a substantial amount of (Me 3 Si) 3 SiH formed during the reaction.
Repeating the reaction of 1c with (Me 3 Si) 3 GeK led to 3d (Figure 4), which is analogous to 3c but contains a Ge(SiMe 3 ) 3
A look at the solid state structure of compound 3c reveals that Zr1 and Zr2 are bridged by a chloride. Depending on the assignment of a covalent interaction of the Cl atom with one Zr atom and a dative interaction with the other Zr atom, we can categorize the Zr atom with the covalent interaction as Zr(IV) and the other one as Zr(III). The Cl−Zr distances of 2.573(1) Å (Zr2−Cl) and 2.627(2) Å (Zr1−Cl) suggest that Zr1 should be assigned Zr(III). Nevertheless, the Si−Zr distance of 2.858(2) Å is not really significant as we observed very similar distances for Si−Zr(IV) bond lengths of 2.853(2) Å in Cp 2 Zr(Cl)Si(SiMe 3 ) 2 (SiMe 2 Thex) 55 64 coordinating to Zr(IV). Nevertheless, the value for the Ge−Zr distance of 3d is only slightly larger than that of Si−Zr for 3c. Zr−Zr distances in 3c and 3d are almost identical (3.295(11) Å and 3.299(11) Å, respectively) and are thus only slightly elongated compared to Fv[Cp 2 Zr(μ-Cl 2 )] (3.233(2) Å). 62 The Zr−Cl distances in the latter complex are between 2.571(2) and 2.591(2) Å, which are close to what we observe for the Zr2−Cl bond lengths of 3c and 3d.
Our previous study had shown that apart from reactions of silanides with group 4 metallates, it is also possible to access silylated group 4 Zr(III) complexes by reduction of the respective Zr(IV) silyl complex. We therefore subjected Cp 2 Zr{Si(SiMe 3 ) 3 } 2 to reaction with elemental potassium in the presence of crown ether and indeed obtained dark red crystals of (18-crown-6)·K[Cp 2 Zr{Si(SiMe 3 ) 3 } 2 ] (3e) (Scheme 4, Figure 5).
Hypersilylated Lanthanidocenates. In this paper we also want to report on related chemistry of lanthanides. Reactions of a number of lanthanidocenes Cp 3 Ln (Ln = Ce, Sm, Gd, Ho, Tm) occur with (Me 3 Si) 3 SiK·18-crown-6 (2a) (Scheme 5). The reactions proceeded in all cases in a way that hypersilylated lanthanidocenates were formed. However, the nature of the positively charged counterion was different for the particular examples (Scheme 5).
All of the investigated lanthanides contain unpaired felectrons and thus are paramagnetic. As straightforward NMR spectroscopic evaluation thus was not possible, we based our

Inorganic Chemistry
Article analysis on single crystal XRD diffraction of the obtained metallates (see Table 1 for a compilation of acyclic and cyclic cases of Si-Ln metallates including oligosilanyl ligands).
Reaction of 2a with Cp 3 Ce gave the complex [18-crown-6· K·Cp·K·18-crown-6][Cp 3 CeSi(SiMe 3 ) 3 ] (4) (Scheme 5, Figure 6). This is somewhat surprising because the presence of the extra CpK in the complex cationic counterion clearly indicates a more complex reaction than indicated in Scheme 5. The fact that two K·18-crown-6 ether units are present in the product suggests that two equivalents of 2a is required for product formation and that one Cp 3 Ce molecule is losing at least one of its Cp units. While we have observed frequently that silanides can replace cyclopentadienides from early metal complexes, 24,25 the current case is special as the formed product contains four Cp units from a starting material containing only three of those. However, this behavior is not totally unprecedented as similar examples for terbium and erbium have been reported by the Evans (Tb) 44 and Zheng (Er) 45 groups.
The structure of the anionic metallate part of 4 ( Figure 6) is not unexpected, since also the reaction of Cp 3 Ce with an 1,4oligosilanyldiide did not give the anticipated ceracyclopentasilane but led to the formation of two silanylene bridged silyl cerate units. 29 However, two K·18-crown-6 units were the cationic counterions in this case. The Si−Ce distance of 3.155(2) Å is somewhat shorter than the one found for this bridged compound (3.228(2)Å), reflecting the fact that the bridging ligand is sterically somewhat more demanding than the Si(SiMe 3 ) 3 group. 29 No other examples of silylated cerium compounds are known.
Reaction of 2a with Cp 3 Sm proceeded similarly to what was observed for Cp 3 Figure 7). The same complex cationic counterion as for 4 was observed. The Si−Sm distance of 5 is 3.103(2) Å, which is substantially longer than the 2.880 Å reported by Schumann 33 In our previous account on the reactions of Cp 3 Ln with 1,4dipotassium-tetramethyl-1,1,4,4-tetrakis(trimethylsilyl)tetrasilane 2c we have not included Cp 3 Sm as starting material. 29 In order to have a suitable comparison compound we caught up on the synthesis of samaracyclopentasilane 9 (Scheme 6, Figure 8), which was obtained by reaction of Cp 3 Sm with oligosilanyl dianion 2c.

Reaction of 2a with Cp 3 Gd gave [18-crown-6·K·Cp·K·18crown-6][Cp 3 GdSi(SiMe 3 ) 3 ] (6) (Scheme 5,
In the case of synthesis of the thulium complex 8 (Scheme 5) two different kinds of crystals were formed which could be separated under the microscope and led to structures 8a ( Figure 11) and 8b ( Figure 12). The difference between 8a and 8b is located in the cationic part: 8a crystallizes in infinite chains with one K·18-crown-6 unit bridging two Cp-ligands; thus, two of the three cyclopentadienyls of each Tm are coordinating to a potassium ion. For the structure of 8b also infinite chains in the crystal are observed, but while 8a only contains K·18-crown-6 units as counterions, in 8b one K·18crown-6 unit and one of the inverse sandwich units [18-crown-6·K·Cp·K·18-crown-6] are present. One potassium ion of the [18-crown-6·K·Cp·K·18-crown-6] unit interacts weakly with a trimethylsilyl group (see bottom left side of Figure 12).
Consistently, the Si−Tm distances of 3.018(3) Å in 8a and 3.014(2) Å in 8b are longer than the 2.980(1) and 2.966(2) Å observed for the thullacyclopentasilane ate-complex. 29 In the course of the single crystal XRD analysis of complexes 4, 5, 6, 7, 8a, 8b, and 9, we were interested in the presence of the additional Si(SiMe 3 ) 3 group coordinating to the Cp 3 Ln unit causing a large distortion of the Cp-Ln distances. Table 1 lists the distances between Cp centroids and the respective Ln ions for the neutral lanthanidocenes 65   shown that these compounds are NMR-silent. The same was found true for the complexes of the current study. EPR spectroscopy of cyclic compounds 25 indicated some delocalization of the electron between the metal and the attached silicon atoms.
Our attempts to do EPR spectroscopy of the hypersilylated lanthanidocenates were futile. As lanthanides are in the regime of strong spin−orbit interaction, due to very short relaxation times EPR spectra frequently can be observed only at temperatures below 20 K. 69 On the other hand, NMR spectroscopy of paramagnetic compounds is well established. However, not all paramagnetic compounds are simple to measure. 70 As we have reported NMR data for the related metallacyclopentasilane complexes, 29 it seemed reasonable to expect the same for the current complexes. We have therefore extensively tried to obtain NMR spectra for the metallates of the type [Cp 3 CeSi(SiMe 3 ) 3 ] − . For {K 2 (18-c-6) 2 -Cp}[Cp 3 CeSi(SiMe 3 ) 3 ] (4) no meaningful NMR spectra could be obtained at all. For the other compounds, especially complexes 7 (Ho) and 8 (Tm), the situation was different. We were able to get fairly meaningful 1 H NMR spectra, and using 2D-NMR techniques (HSQC and HMBC) we tried to get 13 C and 29 Si data. However, we also found that chemical shifts of the compounds are extremely concentration dependent with strongly shifted signals. We assume that this effect was much less pronounced with the bidentate ligand used before because the latter is connected to the metal fragments electrostatically. While we realize that concentration

Inorganic Chemistry
Article dependent magnetic behavior is interesting, we intend to study it in more detail in a future investigation.

CONCLUSION
Some years ago we could show that reactions of group 4 metallocene dichlorides (M = Zr, Hf) with oligosilanyldiides give metallacyclosilanes. Later, we found that double silylation of Cp 2 TiCl 2 is more difficult as the two silyl ligands tend to undergo reductive elimination to a cyclosilane and "Cp 2 Ti". The latter reacts with Cp 2 TiCl 2 to Cp 2 TiCl or more likely to an adduct thereof such as [Cp 2 TiCl 2 ] − . Further reaction with disilanide then gave titanacyclosilanes with Ti(III). Alternatively, these compounds and also analogous Zr
X-ray Structure Determination. For X-ray structure analyses the crystals were mounted onto the tip of glass fibers. Data collection was performed with a BRUKER-AXS SMART APEX CCD diffractometer using graphite-monochromated Mo Kα radiation (0.71073 Å). The data were reduced to F 2 o and corrected for absorption effects with SAINT 75 and SADABS, 76,77 respectively. Structures were solved by direct methods and refined by full-matrix least-squares method (SHELXL97). 78 If not noted otherwise all non-hydrogen atoms were refined with anisotropic displacement parameters. All hydrogen atoms were located in calculated positions to correspond to standard bond lengths and angles. All diagrams were drawn with 30% probability thermal ellipsoids, and all hydrogen atoms were omitted for clarity. Crystallographic data (excluding structure factors) for the structures of compounds 3a, 3b, 3c, 3d, 3e, 4, 5, 6, 7, 8a, 8b and 9 reported in this paper have been deposited with the Cambridge Crystallographic Data Center as supplementary publication nos. CCDC-1891714 (3a), 767188 (3b), 767186 (3c), 1891716 (3d), 1904333 (3e), 1891721 (4), 1891720 (5), 1891718 (6), 1891719 (7), 1891715 (8a), 1891717 (8b), and 1891722 (9). Copies of data can be obtained free of charge at: http://www.ccdc.cam.ac.uk/products/csd/request/. Figures of solid state molecular structures were generated using Ortep-3 as implemented in WINGX 79 and rendered using POV-Ray 3.6. 80 Dicyclopentadienylbis{bis(trimethylsilyl)silyl}titanate(III) (3a). To a green suspension of 1a (453 mg, 0.82 mmol) in toluene (10 mL) was added dropwise a solution of 2a (1.64 mmol) in pentane (10 mL) at −90°C. After 2 h the reaction mixture was allowed to come to rt and the stirring was continued for another 16 h. The mixture was filtered and the solvent reduced to 3 mL. Crystallization was achieved by overlaying of 10 mL of pentane within 24 h. Crystalline red-brown needles of 3a (673 mg, 68%) were obtained. Mp.: 153−155°C. Anal. Calcd for C 51  Dicyclopentadienylbis{tris(trimethylsilyl)silyl}hafnate(III) (3b). To a solution of 1b (368 mg, 0.54 mmol) in toluene (5 mL) was added dropwise a solution of 2a (1.08 mmol) in pentane (5 mL) at −60°C. After 1 h the reaction mixture was allowed to come to rt and the stirring was continued for another 3 h. The solid components were removed by filtration and the solvent reduced to 3 mL. Crystallization was achieved by overlaying by 5 mL of pentane within 24 h. Crystalline orange needles of 3b (354 mg, 81%) were obtained. Mp.: 174−176°C. Anal. Calcd for C 40  Bis(trimethylsilyl)silyl Zirconium Fulvalene Complex (3c). To a solution of 1c (150 mg, 0.25 mmol) in toluene (5 mL) a solution of 2a (0.50 mmol) in pentane (5 mL) at −60°C was added dropwise. After 1 h the reaction mixture was allowed to come to rt and the stirring was continued for another 3 h. The solid components were removed by filtration and the solvent reduced to 3 mL. Crystallization was achieved by overlaying by 5 mL pentane within 24 h. Crystalline

Inorganic Chemistry
Article