Chemistry of a 1,5-Oligosilanylene Dianion Containing a Disiloxane Unit

Synthesis of a number of disiloxane containing cyclo- and bicyclooligosilanes is described starting from the dipotassium 1,5-oligosiloxanylene diide derived from 1,3-bis[tris(trimethylsilyl)silyl]tetramethyldisiloxane. In addition, the use of this particular fragment as ligand for zinc and group 4 metallocene complexes was studied. Both types of compounds exhibit marked structural differences compared to related compounds containing Si-Si-Si units instead of the Si-O-Si fragment.


■ INTRODUCTION
Over the past years, we have utilized oligosilanylene diides 1−4 for the synthesis of longer oligosilane chains, 5−10 cyclosilanes, 2,7,11−14 heterocyclosilanes, 11,12,15,16 and as ligands for silyl transition metal complexes 1,4,17−22 and silylated low valent main group compounds. 23−30 Usually methylated oligosilanylene units were used as the connecting units between the two silyl anionic atoms of these compounds. Such spacer parts generally do not interact with a newly incorporated heteroatom and are mainly responsible for conformational properties. However, our recent studies concerning the use of silanides as ligands for lanthanide complexes 31−34 have brought about the necessity of incorporating additional donor sites into the ligand backbone. These additional donor sites for the metal atom should avoid or diminish the coordination of solvent molecules like THF or DME to the metal atoms. Solvent free lanthanide complexes allow the use of vacuum during workup procedures and do not restrict the solvent use in order to ensure a homogeneous product distribution. For this reason, we have prepared several different silylated siloxanes, and in doing so, disiloxane 1 (Scheme 1) turned out to be an easily available ligand with great opportunities for further transformations, leading to a variety of interesting new compounds. Furthermore, theoretical 35,36 and synthetic aspects 37−41 of siloxanes have gained considerable attraction in recent times. Despite the large structural variety of oligosilanes that have been prepared over the past years, compounds with Si−Si bonds and Si-O-Si units are not very abundant. While such compounds are available by controlled hydrolysis of α,ω-dichlorooligosilanes, 42 examples with even slightly more complex molecular architecture are rather rare. Nevertheless, Krempner et al. have shown that dendritic oligosilanes with discrete disiloxane units are interesting compounds for the modeling of oxygen defects in silicon nanomaterials 43 and von Hanisch and co-workers have recently incorporated oligosiloxane units into crown ethers. 44 Upon treatment with 2 equiv t BuOK, disiloxane 1 can be converted to the respective oligosilanylene diide 2 (Scheme 1). 33 In the reactions of dianion 2 with YbI 2 ·(THF) 2 and SmI 2 · (THF) 2 , it acted as a tridentate ligand to Ln(II), leading to complexes 3 (Scheme 2). 33 The fact that the lanthanide ion coordinates to the very weakly basic siloxane oxygen 45 is likely caused by the ion's very strong Lewis acidity.
With a convenient access to disiloxane 1 and the respective dianionic derivative 2, we thought it would be interesting to use these as precursors for the design of oligosilanes with even more siloxane units and also for the formation of additional silyl-metal complexes.
■ RESULTS AND DISCUSSION Disiloxane Containing Oligosilanes. With compound 2 readily available, we decided to study its chemistry in more detail. By addition of 1,2-dibromoethane, oxidative coupling of the two silanide moieties 46 was achieved, yielding oxacyclopentasilane 4 (Scheme 3). In the case of using a slight excess of 1,2-dibromoethane, in addition to 4 also the 1,5-dibromide 5 was formed as a side product, which was converted to 4 by reaction with added potassium graphite (Scheme 3).
The 29 Si NMR spectrum of compound 4 (Table 1) features expected values for SiMe 3 (−9.6 ppm), and Si(SiMe 3 ) 2 (−132.2 ppm). For the disiloxane unit, a resonance at 20.9 ppm was observed, which is somewhat downfield shifted compared to acyclic products 1 and 5 (Table 1), as can be expected for the diminished Si-O-Si angle in a cyclic compound.
The eight-membered ring of 6 allows for a widened Si−O− Si angle, and accordingly, the 29 Si NMR chemical shift of the siloxane silicon atoms (11.6 ppm) is close to that of the acyclic compound 1 (13.4 ppm) (Table1). Correspondingly, the 29 Si NMR spectrum of the respective 1,4-dianionic compound 7 resembles that of compound 2 ( Table 1).
The 3,7-dioxabicyclo[3.3.0]octasilane 8 is structurally very similar to 4. This is clearly reflected by its 29 Si NMR spectrum which resembles that of 4. In a similar sense, compound 9 is structurally related to 6. The 29 Si NMR resonances of the trimethylsilyl groups of 1, 33 6, and 9 experience upfield shift in this order. Compound 9 is a rare example of a tricyclic oligosiloxane. A somewhat related bicyclo [3.3.3]pentasiloxane was recently obtained by Iwamoto and co-workers using mCPBA oxidation of a 1,3-bis(trimethylsilyl)bicyclo[1.1.1]pentasilane. 37 Compound 10 features a very simple 29 Si NMR spectrum with only two lines; the typical upfield resonances for the anionic silicon atoms (−186.7 ppm) are accompanied by a peak at 15.5 ppm for the SiO units. The compound might be regarded as a building block for the synthesis of low dimensional materials such as one-dimensional nanorods consisting of bridgehead connected bicyclo[3.3.3]trisiloxane units. 37 Facile protonation of oligosilanylene diide 2 yielded the respective 1,5-dihydrosilane 11. Reaction with tetrachloromethane converted 11 to the 1,5-dichlorooligosilane 12 (Scheme 5). 47 Further reaction of 12 with excess diethylamine gave 1,5-bis(diethylamino)oligosilanyldisiloxane 13, 47 which upon reaction with aqueous methanol led to the rather unexpected formation of 1,4-dioxacyclohexasilane 14. We assume that 14 forms via an intermediate oligosilane diol, which in the presence of Et 2 NH is partly deprotonated. Attack of the respective siloxide at a SiMe 2 unit leads to a rearranged oligosilane diol, which upon water elimination can cyclize to 14 (Scheme S1).   29 Si (Si q ) 29 Si (Si-E) Quite typically, dihydrooligosilane 11 was obtained as an oil. Its 29 Si NMR spectroscopic properties are very much as expected. The Si-H resonance at −116.5 ppm is close to the respective signal of (Me 3 Si) 3 SiH (−115.4), and also the trimethylsilyl signal at −12.6 ppm is in line with the −10.9 ppm observed for (Me 3 Si) 3 SiH. 46 In a similar way, the 29 Si NMR signature of oligosilanyldichloride 12 (5.9 (SiO), −14.9 (SiMe 3 ), −19.7 (SiCl) ppm) reflects the similarity of 12 to (Me 3 Si) 3 SiCl (−11.6 (SiMe 3 ), −13.3 (SiCl) ppm). 48  Metal Complexes with Disiloxane Containing Oligosilanyl Ligands. Silylated lanthanides are an interesting field of research pioneered by Schumann and co-workers. 49,50 Oligosilylated examples are still investigated by us 31−34 and others. 51−54 As mentioned, we initially devised the synthesis of oligosilanylene diide 2 to employ it as a ligand for Ln(II)-silyl complexes. 33 As compound 7 can be regarded as a derivative of 2, containing an additional disiloxane unit, we reacted it with YbI 2 (Scheme 6). 1 H NMR studies showed that the obtained product 15 was indeed coordinating to both oxygen atoms as only two THF or one DME molecules were shown to occupy the remaining two of the six coordination sites of Yb.
The clean reaction of 2 with YbI 2 encouraged us to study its coordination chemistry also with other divalent metal halides (Scheme 7). Not unexpectedly, 2 can be cleanly transmetalated to the respective magnesium compound 16 by reaction with MgBr 2 ·Et 2 O. 4,55 16 exhibits the typical 29 Si NMR spectroscopic signature known for oligosilanyl magnesium compounds. The signal at −166.9 ppm (Table 1) reflects the diminished anionic character compared to 2. While the influence of the negative charge on the directly metalated silicon atom is most pronounced, a downfield shift for attached trimethylsilyl groups compared to the neutral precursor molecules is usually observed.
Conversion of oligosilanides with zinc halides to silyl zinc compounds is a well established process. 19,34,56−60 Reaction of 2 with ZnCl 2 was thus attempted (Scheme 7). We expected a six-membered ring to be formed in the reaction; 19 however, the obtained product 17 is a 12-membered ring with close to linear Si-Zn-Si coordination geometry. Earlier studies have already shown a pronounced tendency of the Si-Zn-Si unit to  19,56 Cases with significant bending of the Si-Zn-Si unit are almost always accompanied by coordination of one or more Lewis bases to the involved Zn atom. The main reason compound 17 forms is likely that not only the Si-Zn-Si unit preference for linear arrangement but also the Si-O-Si part's tendency for engaging in larger angles. 29 Si NMR resonances at −5.8 (SiMe 3 ) and −142.0 (SiZn) ppm are close to the respective −7.2 and −123.9 ppm observed for (Me 3 Si) 3 SiZnSi(SiMe 3 ) 3 . 56 In contrast to the reaction of 2 with ZnCl 2 , analogous reactions with zirconocene and hafnocene dichlorides gave compounds 18 and 19 with six-membered rings (Scheme 7). At first glance, this is not unexpected. However, our previous attempts to react Cp 2 MCl 2 (M = Zr, Hf) with an oligosilanyl 1,5-diide caused eventual formation of M(III) complexes. 17 If we would envision a similar course as for the previous reaction, we would have expected that compound 4 would form in the reaction by reductive elimination from 18 and 19. Although it is not quite clear why compounds 18 and 19 are stable toward the elimination process, it seems likely that the ring strain of compound 4 is higher than that of 1,1,2,2-tetrakis(trimethylsilyl)hexamethylcyclopentasilane. The reason for this increased strain seems to be the enhanced tendency of the Si-O-Si unit to acquire angles larger than tetrahedral. 29 Si NMR chemical shifts of silylated zirconocenes and hafnocenes typically are much deshielded compared to the respective silanides. For structurally related 1-zircona-and 1hafna-2,2,5,5-tetrakis(trimethylsilyl)tetramethylcyclopentasilanes, 1 values of −65.2 and −52.2 ppm, respectively, were observed. The resonances for 18 (−71.5 ppm) and 19 (−45.7 ppm) are similar, but the difference between the two metals is more pronounced.
Crystal Structure Analysis. The molecular structure of 4 was determined using single crystal XRD analysis ( Figure 1).
The five-membered ring is almost planar (sum of angles is 537°) which is caused by a large Si−O−Si angle of 132.4° (  Table 2). As a consequence of the planar arrangement, the Me 3 Si−Si−Si−SiMe 3 torsional angles are small (16.2°and 17.0°), causing some steric interaction between the vicinal trimethylsilyl groups. The Si(1)−Si(4) distance is therefore slightly elongated (2.3887(6) Å).
Although compound 6 contains an eight-membered ring, in the solid state, a fairly wide Si−O−Si angle of 153.7°causes the molecular structure ( Figure 2) to engage in a conformation that is similar to a six-membered ring chair conformer.
Compounds similar to 14 are not abundant. The structurally related 1,4-dioxaoctamethylcyclohexasilane was prepared by hydrolylsis of 1,2-dichlorotetramethyldisilane 61 a long time ago, and its structure was determined by XRD methods more recently. 62 The structure is quite similar to that of 14 (

■ CONCLUSION
The current work continues our studies of the transformation of siloxane 1 to higher oligosiloxanes and illustrates the use of these compounds as ligands for metal complexes. Utilizing 1, we could demonstrate that cyclic and bicyclic oligosilanes with one or more siloxane units can be prepared. Most of these compounds still contain peripheral trimethylsilyl units and thus can be converted to synthetic building blocks by simple reaction with potassium tert-butoxide.
Reactions of the siloxane containing dipotassium oligosilanylene diide 2 with magnesium and zinc halides proceeded smoothly, but for both metals, no interaction with the siloxane oxygen was detected. Somewhat unexpectedly, reactions of 2    with zirconocene and hafnocene dichlorides occurred to the respective 1-metalla-4-oxacyclohexasilanes. We initially assumed that the latter compounds would undergo reductive elimination to form an oxacyclopentasilane. A likely reason for the stability of the 1-metalla-4-oxacyclohexasilanes is ring strain in the potential reaction product caused by a strong tendency of Si-O-Si units to acquire larger than tetrahedral angles. The synthesized metallaoxacyclosilanes as well as the oxacyclo-and bicyclosilanes exhibit structural features that are different from isostructural homocyclo-and bicyclosilanes, which is mostly caused by Si−O−Si angles significantly larger than the corresponding Si−SiMe 2 −Si angles.
1 H (300 MHz), 13 C (75.4 MHz), and 29 Si (59.3 MHz) NMR spectra were recorded on a Varian INOVA 300 spectrometer and are referenced to tetramethylsilane (TMS) for 1 H, 13 C, and 29 Si. If not noted otherwise, the used solvent was C 6 D 6 and samples were measured at rt. In the case of reaction samples, a D 2 O capillary was used to provide an external lock frequency signal. To compensate for the low isotopic abundance of 29 Si, the INEPT pulse sequence 64,65 was used for the amplification of the signal for some of the spectra.
Elemental analyses were carried out using a Heraeus VARIO ELEMENTAR instrument. For a number of compounds, obtained elemental analysis showed too low carbon values, which is a typical problem for these compounds likely caused by silicon carbide formation during the combustion process. Multinuclear NMR spectra ( 1 H, 13 C, 29 Si) of these compounds are presented in the Supporting Information (SI) as proof of purity.