Oligosilanylated Antimony Compounds

By reactions of magnesium oligosilanides with SbCl3, a number of oligosilanylated antimony compounds were obtained. When oligosilanyl dianions were used, either the expected cyclic disilylated halostibine was obtained or alternatively the formation of a distibine was observed. Deliberate formation of the distibine from the disilylated halostibine was achieved by reductive coupling with C8K. Computational studies of Sb–Sb bond energies, barriers of pyramidal inversion at Sb, and the conformational behavior of distibines provided insight for the understanding of the spectroscopic properties.


INTRODUCTION
Over the last 50 years, the chemistry of oligosilanes has experienced tremendous progress. 1−4 In the course of this development, numerous examples of compounds substituted with heteroatoms were prepared and studied. The chemistry of group-15-substituted oligosilanes is dominated by aminosilanes and, to some lesser extent, phosphinosilanes. Examples of the heavier group 15 elements bearing oligosilanyl substituents, however, are quite scarce. For antimony in particular, only four compounds are known at all. 5−7 These molecules were prepared by two different synthesis routes. Hassler and Seidl utilized reactions of alkali metal stibides with chlorosilanes to obtain dendrimeric 5 or bicyclic 6 compounds. Alternatively, Hopkins and co-workers utilized the reaction of (Me 3 Si) 3 SiK with SbCl 3 to obtain [(Me 3 Si) 3 SiSb] 4 . 7 It should be noted that exactly the same type of chemistry was also reported with bismuth. 5,6,8 The oligosilanyl anion chemistry developed in our group over recent years 9 has proven to be very useful for the synthesis of heteroatom-substituted oligosilanes. 2 With respect to group 15 compounds, however, this was limited to a single study demonstrating the availability of phosphacyclo-and bicyclosilanes. 10 The current account intends to show how oligosilanyl anion chemistry can be used to establish some foundations of oligosilanylstibine chemistry.

RESULTS AND DISCUSSION
Our good experience of using oligosilanyl dianions for the synthesis of heterocyclosilanes 10−16 encouraged us to attempt the synthesis of stibacyclosilanes by reaction of oligosilanyl dianions 1a and 2a 12 with antimony trichloride. However, the reactions were not successful because the major reaction pathway in these cases was metal−halogen exchange leading to cyclosilane formation. To suppress this unwanted course of reaction, we switched to analogous magnesium silanediides 1b and 2b, which are readily available from the respective potassium silanediides by metathesis with magnesium bromide. 13 Upon reaction of 1b with SbCl 3 , formation of a 1-halo-1stibacyclotetrasilane was expected. However, the respective distibine 3 was obtained instead in a modest yield of 28% (Scheme 1). The fact that 3 is formed instead of the expected halostibine can be explained with a metal−halogen exchange reaction of the initially formed halostibine with another equivalent of 1b, leading to a magnesium stibide that can then react further with another halostibine to distibine 3. Alternatively, the reduction of the halostibine to a radical followed by recombination to a distibine may also be a possibility, especially in light of the recent findings of stable antimony-centered radicals by Iwamoto and co-workers. 17 In any case, the formation of distibines can only be imagined at the expense of some silanide.
It is important to note that using a 1:1 stoichiometry of the oligosilanyl dianion and SbCl 3 , a reaction sequence as proposed above limits the theoretically possible yield to 66% with respect to the used oligosilanyl dianion. The assumptions outlined are supported by the reaction of magnesium 1,4-cyclohexasilanediide 5 18 with SbCl 3 . Again a distibine (6) was formed, and in addition, the known bicyclo[2.2.0]hexasilane, 7, 18 was detected as the sole byproduct (Scheme 2). The latter is known to form from the respective dianion upon addition of 1,2-dibromo-ethane, which causes a potassium bromine exchange reaction as the first step followed by cyclization. 18 The formation of distibines is likely related to what occurs in the formation of [(Me 3 Si) 3 SiSb] 4 (9), as described by Hopkins and co-workers. 7 This is supported by the large amount of (Me 3 Si) 3 SiSi(SiMe 3 ) 3 formed as a byproduct in this reaction. To confirm this, we used an adapted protocol for the formation of [(Me 3 Si) 3 SiSb] 4 (9), which utilizes [(Me 3 Si) 3 Si] 2 Mg (8) instead of (Me 3 Si) 3 SiK and uses only 2 / 3 equiv of SbCl 3 . Although the yield of the modified reaction conditions was still poor (13%) the purity of the obtained product allowed reassignment of the 29 Si NMR signals reported in the original study 7 (vide inf ra) (Scheme 3). Repeating the reaction with an equimolar amount of SbCl 3 gave rise to the formation of bis[tris(trimethylsilyl)silyl]antimony bromide (10) (Scheme 3).
In an attempt to obtain a five-membered stibacyclosilane, SbCl 3 was reacted with magnesium tetrasilanediide 2b. Comparison with the reaction of 1b revealed a different outcome, and eventually the expected halostibine 4 was obtained in 70% yield. Compound 4 is, however, not the anticipated chlorostibine but rather a bromostibine, which was likely formed in a Finkelstein-type reaction with the magnesium bromide present in solution (Scheme 1). The bromostibyl unit is interesting as it allows further reactions at the antimony atom with nucleophiles and reducing agents.
To obtain a distibine analogous to 3 and 6, compound 4 was treated with potassium graphite (C 8 K) to give rise to the formation of 11 (Scheme 4). When 4 was treated with potassium tert-butoxide the alkoxystibine 13 was obtained in excellent yield (Scheme 4). Eventually reaction of 4 with half an equivalent of magnesium silanide 8 gave trisilylated stibine 12 (Scheme 4).
For compound 11, a diffuse reflectance spectrum in the UV− vis−NIR range was obtained ( Figure 2).
Quite similar to the solid state spectra of (Me 3 Si) 2 SbSb-(SiMe 3 ) 2 , 19 (Me 3 Ge) 2 SbSb(GeMe 3 ) 2 19 and (Me 3 Sn) 2 SbSb-(SnMe 3 ) 2 20 the spectrum of 11 is dominated by a broad unstructured absorption band in the visible range. In addition, a series of combinations and overtones of molecular vibrations is visible in the NIR part of the spectrum. The broad band accounts for the dark-purple color of 11 in that it shows slightly higher reflectance in the red and blue regions (see inset of Figure 2). To assign the bands in the NIR region, an infrared spectrum of the solid compound was recorded, which is shown in Figure 3.
A tentative assignment of the observed bands in the IR spectrum according to the literature 22 is given in Figure 3 and ν as (CH 3 ), region C (ca. 6840−7400 cm −1 ) is a combination of the two stretching vibrations and one bending vibration, and region D (ca. 8250−8700 cm −1 ) corresponds combinations of the second overtone of a stretching vibration with one of the bending vibrations.
2.2. NMR Spectroscopy. Because antimony does not have stable-spin 1/2 isotopes, NMR spectroscopic characterization of the obtained compounds has to concentrate on 1 H, 13 C, and in particular on 29 Si NMR spectra ( Table 1). 1 H, 13 C, and 29 Si NMR spectra of 3 exhibit a molecular symmetry that is consistent with configurational stability of the antimony atom. Accordingly, side differentiation of the stibacyclotetrasilane is observed. The spectra of distibine 3 also reveal some influence of the antimony atom on the chemical shifts of the molecule. Interestingly, it is neither the neighboring silicon atoms nor the trimethylsilyl groups attached to those that is most affected but rather the SiMe 2 group. Comparison of 3 with 1,1,3,3tetrakis(trimethylsilyl)tetramethylcyclotetrasilane 23 shows that 1 H and 13 C shifts of the methyl groups are within the expected range. The same is true for the 29 Si shift of trimethylsilyl groups and the quaternary silicon atom. The latter displays the typical downfield shift characteristic for cyclotetrasilanes. The 29 Si NMR resonance of the SiMe 2 group is unexpected because its value of −11.3 ppm is considerably downfield to the −25.5 ppm found for 1,1,3,3-tetrakis(trimethylsilyl)tetramethylcyclotetrasilane. 23 If, however, the SiMe 2 shift of 3 is compared to 1,1,2,2-tetrakis(trimethylsilyl)tetramethylcyclotetrasilane, 24   18 In contrast to the spectra of distibines 3 and 6, the molecular symmetry of the third distibine, 11, exhibits no configurational stability at antimony. Only one type of SiMe 3 and SiMe 2 groups are present, and additionally, the methyl groups of the SiMe 2 units are magnetically equivalent. An alternative explanation for this magnetic equivalence would be a rotation around the Sb− Sb bond of 11. As pointed out below, this process is, however, energetically not feasible. All chemical shifts of 11 are in the expected ranges.
The spectra of the precursor to 11, bromostibine 4, show two different trimethylsilyl resonances in the 1 H, 13 C, and 29 Si spectra, indicating that the antimony atom exhibits configurational stability. As a consequence of this, two different resonances for the methyl groups of the SiMe 2 unit are also expected. In the 1 H NMR spectrum, one of these SiMe 2 signals overlaps with that of a SiMe 3 resonance at δ = 0.42 ppm. In the 13 C spectrum, however, the two SiMe 2 signals exhibit the same chemical shift, as confirmed by 2D 1 H− 13 C correlation spectroscopy (gHSQC).
The structurally similar alkoxystibine, 13, which also features a configurationally stable antimony atom, shows all the required signals in expected ranges. Tris(trimethylsilyl)silylated stibacyclopentasilane 12 does not display side-differentiation of the ring. One may conclude that the reason for this is the steric bulk of the tris(trimethylsilyl)silyl group, but the effect is most likely electronic in nature because a trimethylsilylated stibacyclopentasilane (16) is also configurationally unstable. 25 All chemical shifts of 12 are within the expected ranges ( Table  1). The NMR spectra of cyclotetrastibine 9 are completely as expected. The only reason to mention these spectra at all is that in the original report on the synthesis of this compound 7 a much more complicated spectroscopic picture was described that was probably caused by the presence of some oligosilane side products.
2.3. Crystal Structure Analysis. The structures of compounds 3, 4, 6, 9, and 11−13 in this study were characterized by single-crystal X-ray diffraction (Tables S1 and S2). A compilation of the obtained structural data is given in Table 2 26 The structures of distibines 3 (Figure 4), 6 ( Figure 6), and 11 ( Figure 8) are surprisingly different. Although no extraordinary differences with respect to bond distances can be observed, the conformational situations of the three compounds are quite diverse. Distibine 3 presents itself as a molecule with an inversion center in the middle of the Sb−Sb bond. The two four-membered rings are parallel, and the angle of the Sb−Sb bond crossing the plane Si1−Sb−Si3 is 69.40(9)°, the largest of all three compounds. A comparison of structurally characterized distibines reveals the expected strong preference for a conformation where the two Sb-lone pairs are in trans position. This conformation is found perfectly for 3, and it is also present in a slightly distorted form in 6. Although the quality of the structure of distibine 11 is not as good as the ones of the compounds 3 and 6, the difference in conformational behavior is nevertheless clearly visible. The Si− Sb−Si bond angle of the five-membered ring is considerably larger (approximately 99°) than those for 3 (80.5°) and 6 (approximately 85°), which brings the trimethylsilyl groups attached to the neighboring silicon atoms closer together. To avoid steric interaction, the angle of the Sb−Sb bond crossing the Si−Sb−Si plane is diminished to 50°, and in addition, the two rings are twisted along the Sb−Sb bond. A similar conformation was also observed for a recently reported tetraalkyl distibine that is in equilibrium with stibinyl radicals. 17 It is likely that the particular conformation of 11 facilitates the inversion process of the antimony atom that was observed in the NMR spectra. This is in agreement with the fact that the Sb atoms of 3 and 6 are highly pyramidalized as indicated by the sums of bond angles around Sb of approximately 291° (Table  2). To minimize steric interaction, the degree of pyramidalization of Sb in distibine 11 is much less, displaying angle sums of 322−324°. It seems also likely that the strained distibine conformation is responsible for the fact that bromostibine is formed at all and also for the distinctly different UV−vis absorption behavior of 11 compared to that of 3 and 6. Compared to the short intermolecular Sb···Sb distances 21 found for the simpler distibines on the type (Me 3 E) 2 SbSb- Bromostibine 4 ( Figure 5) was found to crystallize in the monoclinic space group C2/c. The Sb−Br unit was found to be disordered over two positions, with the bromine atom being either above or below the ring plane. Cyclotetrastibine 9 was reported to crystallize in the orthorhombic space group Pbcn. 7 Our crystals of 9 (Figure 7) contain THF and crystallized in the monoclinic space group P2(1)/c. Nevertheless, the metrical parameters observed are very similar to those reported earlier. Tris(trimethylsilyl)silylated stibine 12 (Figure 9) again shows that a larger substituent on the Sb atom causes steric interactions with the trimethylsilyl groups of the five-membered ring. To avoid these interactions, the angle of the Sb− Si(SiMe 3 ) 3 bond with the Si−Sb−Si plane of the ring    diminished to 49.29(6)°, which is similar to that found in distibine 11. Also, for 12, the degree of pyramidalization is rather low, which is reflected by a sum of bond angles of 329.61(5)°and is consistent with a lack of configurational stability. In contrast to this, alkoxystibine 13, with the less sterically demanding tert-butoxy substituent (Figure 10), exhibits an angle for the Sb−O bond of 75.62(6)°and a sum of bond angles around Sb of 296.12(5)°.

Computational Study.
Given the structural similarity between recently reported stibinyl radical 14 17 ( Figure 11) and monomer of distibine 11, the question of whether radicals are also an issue in our study led us to study computationally the dissociation processes of distibines R 2 Sb−SbR 2 for different R substituents. The calculated dissociation energies, D E , for distibines Me 2 Sb−SbMe 2 , (H 3 Si) 2 Sb−Sb(SiH 3 ) 2 , and 3 are substantial and almost identical (193,198, and 190 kJ mol −1 , at M06-2X/6-31G(d) (Si,C,H) SDD (Sb); see Table 3). 8,28,29 Although D E calculated for distibine 11 is markedly smaller (D E = 164 kJ mol −1 ), it is still significantly higher than that predicted for the dimer, 15, of stibinyl radical 14 (D E = 75 kJ mol −1 ). 30 The high dissociation energies predicted for the Sb− Sb bonds in distibines 3 and 11 clearly rule out homolytic cleavage of the Sb−Sb bonds to a significant degree at ambient temperature. In contrast, the steric congestion brought about by the comparatively short Sb−C bonds in distibine 15 considerably weakens the Sb−Sb bond, which makes stibinyl radical formation at ambient conditions feasible in this case. The longer Sb−Si bonds in persilylated distibine 11 release part of the strain energy, which results in a stronger Sb−Sb bond. The four-membered rings in distibine 3 exert a back-pulling effect on the trimethylsilyl substituents in α position to the    antimony atom, thereby releasing some strain imposed on the Sb−Sb linkage and leading to additional strengthening. Tetramethyldistibine, Me 2 Sb−SbMe 2 , and most of the other structurally characterized distibines, for example, distibine 3, adopt an anti-periplanar (or simply anti) conformation (see Figure 12) around the Sb−Sb linkage in the solid state. In the gas phase, the situation for Me 2 Sb−SbMe 2 is less clear, as both anti and clinal conformations exist at 75°C. In this respect, it is interesting to note that sterically more congested distibines 11 and 15 adopt an unusual anticlinal conformation around the Sb−Sb bond in the solid state. A conformational analysis for tetramethyldistibine and for tetrasilyldistibine reveals a double minimum potential, with local minima for the anti-periplanar and the synclinal conformations (see Figure 12 and the Supporting Information). It is worth mentioning that for both compounds, the barriers for rotation are relatively small (13 kJ mol −1 for Me 2 Sb−SbMe 2 and 23 kJ mol −1 for (H 3 Si) 2 Sb− Sb(SiH 3 ) 2 , at MP2/def2tzvpd) and the located minima with synclinal and anti-periplanar conformations are relatively broad. In particular, the global anti-periplanar minima are expanded to anticlinal conformations. At the density functional M06-2X/6-31G(d),SDD level, which was actually used in this study for the computations concerning experimentally observed distibines 3, 11, and 15, the potential energy surface along the rotational coordinate even shows a very flat maximum for the antiperiplanar conformation (see Figure S9). The situation for heavily substituted distibines 3 and 11 is somewhat altered. The energy differences between the individual conformers are significantly more pronounced and the synperiplanar conformation of distibine 11 is actually higher in energy than for the two isolated stibinyl radicals (see Table 3). This suggests that a simple rotation process around the Sb−Sb bond in distibine 11 is not responsible for the equivalence of its trimethyl-and dimethylsilyl groups on the NMR time scale. Information regarding pyramidal inversion barriers of stibines is rather scarce. In addition to theoretical work on the inversion  of SbH 3 31,32 and an early NMR study on diisopropyl-ptolylstibine, 33 only a report on the isomerization of 2,3,7,8tetramethyl-5,10-di(p-tolyl)-5,10-dihydrostibanthrene 34 is available. To get an estimate of the expected inversion barriers, E inv , we investigated the inversion process of a number of model compounds such as SbH 3 , SbMe 3 , and Sb(SiH 3 ) 3 ( Figure 13). The results of these calculations indicate the influence of both substituent electronegativity and sterics. Although SbH 3 and, in particular, SbMe 3 are clearly configurationally stable, the calculated inversion barrier for Sb(SiH 3 ) 3 is comparably low (92 kJ mol −1 ) because of stabilization of the planar transition state by electrostatic and hyperconjugational effects. 35 Studying the pyramidal inversion processes of antimony atoms incorporated into the 1,4-and 1,3-silandiylene substituents present in 11 and 3 provides further insight. The inversion barrier calculated for the five-membered trimethylsilyl-substituted model compound 16 is as low as 47 kJ mol −1 . This small barrier indicates that for this compound the inversion process is expected to be fast at ambient temperature, which is consistent with 29 Si NMR data obtained for compound 16 25 and the tris(trimethylsilyl)silyl substituted stibacyclopentasilane 12. A closer model for distibines 3 and 11 is provided by the dimethylstibinyl-substituted compounds 17 and 18. Replacement of the silyl substituent by the stibinyl group in 17 results in an increase of the barrier by 24 kJ mol −1 , and the less flexible four-membered ring in 18 increases the barrier to a similar extent (by 18 kJ mol −1 ). The tendency revealed by these model calculations suggests that the observed magnetic equivalence of the silyl groups in distibine 11 is a result of fast inversion of the antimony atoms. Such processes are however less favored for distibine 3 because of the less flexible four-membered ring.

CONCLUSIONS
The chemistry of silylated antimony compounds is a field that has not received too much attention so far. The number of oligosilanylated antimony compounds reported so far is four. 5−7 Employing the chemistry of oligosilanyl dianions developed in our group over recent years, we decided to synthesize a number stibacyclosilanes. Although the reactions of oligosilanides with SbCl 3 indeed provided access to oligosilany-lated antimony compounds instead of the expected cyclic disilylated halostibines, formation of the respective distibines 3 and 6 was observed for the cases of magnesium 1,3-disilanide 1b and cyclic magnesium 1,4-disilanide 5. Reaction of SbCl 3 with magnesium 1,4-disilanide 2b, however, provided cyclic bromostibine 4. The reason for this different reaction behavior is likely a better steric shielding of the antimony atom of 4. This assumption is supported by the different conformational properties of distibine 11, which could be formed by reductive coupling of bromostibine 4 with C 8 K. Although for distibines 3 and 6 the Sb−Sb bond exhibits the typical conformational behavior of distibines, the two five-membered rings of 11 are strongly twisted. The steric interactions causing this twist are similar to what was observed recently by Iwamoto and coworkers for their solid-state dimer of a stibinyl radical. 17 Despite the structural similarity of Iwamoto's distibine and compound 11, no radical formation can be expected by dissoziation of 11. This can clearly be deduced from a computational evaluation of Sb−Sb bond energies. Compared to short Sb−C bonds, the longer Si−Sb distances of 11 diminish the steric strain between the two five-membered rings, which is mainly responsible for an easy stibinyl radical formation. Theoretical studies also explain the difference between distibines 3 and 6, both of which exhibit configurational stability of antimony, and 11, which is lacking this configurational stability.

EXPERIMENTAL SECTION
4.1. General Remarks. All reactions involving air-sensitive compounds were carried out under an atmosphere of dry nitrogen or argon using either Schlenk techniques or a glovebox. All solvents were dried using a column-based solvent purification system. 36 C 8 K 37 as well as compounds 2a,b, 13 5, 13,18 and 8 13,38 were prepared according to published procedures. All other chemicals were obtained from different suppliers and used without further purification. 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. If not noted otherwise, for all samples benzened 6 was used or, in the case of reaction samples, they were   29 Si, the INEPT pulse sequence was used for the amplification of the signal. 39,40 Elementary analysis was carried out using a Heraeus VARIO ELEMENTAR. UV−vis spectra were measured on a PerkinElmer Lambda 35 spectrometer using spectroscopy-grade pentane as solvent. The diffuse reflectance spectrum of compound 11 in the UV− vis−NIR range was obtained using a PerkinElmer Lambda 950 spectrometer equipped with an integrating Spectralon sphere and an R950 photomultiplier for the UV−vis and an InGaAs detector for the NIR range. The solid crystalline sample was transferred to the sample holder in a glovebox and then quickly measured under ambient atmosphere. Several scans were made in succession to make sure that the sensitive sample had not decomposed during the time of measurement. IR spectra were obtained using a Bruker Alpha P FT-IR-spectrometer with ATR module. Raman spectra were obtained using a PerkinElmer RamanStation 400F instrument with the solid sample in a sealed capillary under nitrogen atmosphere.
4.2. X-ray Structure Determination. For X-ray structure analyses, the crystals were mounted onto the tip of glass fibers, and data collection was performed with a BRUKER-AXS SMART APEX CCD diffractometer using graphite-monochromated Mo K α radiation (0.71073 Å). The data were reduced to F 2 o and corrected for absorption effects with SAINT 41 and SADABS, 42 respectively. The structures were solved by direct methods and refined by full-matrix least-squares method (SHELXL97). 43 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. Crystallographic data (excluding structure factors) for the structures of compounds 3, 4, 6, 9, and 11−13 reported in this paper have been deposited with the Cambridge Crystallographic Data Center as CCDC-1008595 (3), -1008596 (4), -1008598 (6), -1008597 (9), -1008599 (11), -1010065 (12), and -1008600