Isolation and Versatile Derivatization of an Unsaturated Anionic Silicon Cluster (Siliconoid).

The characteristic features of bulk silicon surfaces are echoed in the related partially substituted-and thus unsaturated-neutral silicon clusters (siliconoids). The incorporation of siliconoids into more-extended frameworks is promising owing to their unique electronic features, but further developments in this regard are limited by the notable absence of functionalized siliconoid derivatives until now. Herein we report the isolation and full characterization of the lithium salt of an anionic R5 Si6 -siliconoid, thus providing the missing link between silicon-based Zintl anions and siliconoid clusters. Proof-of-principle for the high potential of this species for the efficient transfer of the intact unsaturated R5 Si6 moiety is demonstrated by clean reactions with representative electrophiles of Groups 13, 14, and 15.

The industrial‐scale preparation of silicon is one of the most far‐reaching scientific developments of the 20th century; the progress in microprocessor performance and thus the information technology‐centered society of today would be difficult to imagine without silicon as a readily available key component.1 The understanding and the manipulation of silicon surfaces2 are key aspects of silicon‐based technologies.3 Stable siliconoids4 are unsaturated neutral silicon clusters5 that reproduce the characteristic structural features of silicon nanoparticles and surfaces in the molecular regime, in particular the presence of one or more unsubstituted vertices.6 Stable examples were reported by us4, 7 and the groups of Wiberg,8 Breher,9 Kyushin,10 and Iwamoto.11 The pronounced electronic anisotropy of molecular siliconoids—as shown by an unprecedentedly wide distribution of 29Si NMR chemical shifts12—would be an attractive feature for the construction of structures on the subnano scale. Evidently, functionalized vertices are a prerequisite for the embedding of siliconoids in extended supramolecular assemblies. However, only unfunctionalized siliconoids have been reported so far,4, 7, 8, 9, 10, 11 severely limiting further developments in this regard.

Notably, the structural and NMR spectroscopic properties of siliconoids are reminiscent of Zintl anions of the Group 14 elements.13, 14 Spurred by novel methods for the extraction from solid‐state phases into polar solvents, such as liquid ammonia,15 ethylenediamine,16 or dimethylformamide,17 the potential of Zintl anions as building blocks for extended systems has been recognized in the last few years. Thus, for that Group 14 elements heavier than silicon, rod‐like polyanions,18 nanoparticles,19 or mesostructured elemental allotropes,20 are accessible. The multiple negative charges of Zintl anions—which in the widest sense can be considered as nucleophilic functionalities—could in principle be expected to allow for the attachment of substituents and/or functional groups under retention of unsubstituted vertices, thus establishing a conceptual bridge to the partially substituted and neutral siliconoids. In practice, however, the strongly reducing nature conferred by the multiple charges is detrimental to the reaction with electrophiles that would convert a Zintl anion E into the corresponding neutral siliconoid ER (E=Group 14 element).14a For germanium21 and tin22 a few experimental methods exist nonetheless, although low yields and poor selectivity tend to be observed possibly because of the involvement of competing radical mechanisms. In the case of Zintl anions of silicon, the only rational chemical transformations reported so far result in the complexation of the silicide anions to transition‐metal centers without any compensation of negative charges.23

Herein we report the synthesis and isolation of an intermediate species containing both an anionic and two neutral unsubstituted vertex atoms, which is accessible by the reverse approach of employing the neutral siliconoid 1 as the starting material rather than a Zintl anion. As demonstrated by reactions with representative electrophiles of Groups 13 to 15, this anionic siliconoid can be efficiently functionalized with various electrophiles, while maintaining the integrity of the unsaturated cluster scaffold.

As the reductive cleavage of aryl24 and silyl25 substituents in silicon compounds is known to generate anionic species (most notably a tetrasilatetrahedranide, a silicon cluster anion without any additional unsubstituted vertex25a), we investigated the reduction of the dismutational isomer 1. Treatment of 1 with two equivalents of lithium/naphthalene in a mixture of diethylether and tetrahydrofuran led to a color change from green to dark orange.26 Complete and uniform conversion into a new Si6‐species was detected by 29Si NMR spectroscopy. Six resonance signals with a large dispersion of 29Si chemical shifts, similar to those observed for the global‐minimum hexasilabenzene isomer 2 (Tip=2,4,6‐triisopropylphenyl, Scheme 1),7c were assigned on the basis of a two‐dimensional 1H–29Si correlation experiment. The 29Si signal of one SiTip2 moiety appears a long way downfield at δ=159.6 ppm, while resonances for another SiTip2 and a mono‐substituted SiTip unit appear in the usual region at δ=12.8 and 27.6 ppm, respectively (in 2: δ=174.6, 14.8,and 27.6 ppm). Remarkably, the unsubstituted vertices in the product are apparently not compromised by the reduction as witnessed by two upfield signals at δ=−237.3 and −238.2 ppm (in 2: δ=−274.2 ppm). A resonance at δ=66.8 ppm is assigned to a lithium‐bonded silicon atom on grounds of the apparent broadening through the coupling to the quadrupolar 7Li nucleus. These data are consistent with the reductive elimination of one of the Tip substituents and the formation of the lithiated Si6 siliconoid 3Li (Scheme 1).

Scheme 1

Synthesis of 3Li from either dismutational hexasilabenzene isomer 1 or 1,1,2‐trichlorocyclotrisilane 4 (magenta dots=Si; R=2,4,6‐triisopropylphenyl).

The lithium salt of the anionic siliconoid crystallizes as 3Li⋅(THF)2 in 62 % yield from n‐hexane, albeit with variable amounts of co‐crystallized THF and residual naphthalene. Under argon, 3Li⋅(THF)2 is stable in solution and in the solid state, but readily decomposes in the presence of air and moisture. Single crystals of 3Li⋅(THF)2 were analyzed by X‐ray diffraction and the constitution as a contact ion pair unambiguously confirmed despite the low quality of the data set which is due to pronounced disorder. To obtain higher quality X‐ray data, cation‐sequestering agents were employed. 3[Li(DME)3] (70 % yield) and 3[Li(12‐crown‐4)2] (72 % yield) were isolated as solvent‐separated ion pairs. The 29Si NMR spectra of 3[Li(DME)3] and 3[Li(12‐crown‐4)2] are very similar to that of 3Li⋅(THF)2, except for the significantly sharpened 29Si signal of the anionic silicon atom (3[Li(DME)3]: δ=−73.8, 3[Li(12‐crown‐4)2]: −72.0 ppm). However, only the structure of 3[Li(12‐crown‐4)2] could be refined without major disorder problems (Figure 1) and will therefore be exclusively referred to in the following: The distance between the unsubstituted bridgehead atoms in 3 (Si1–Si2 2.5506(9) Å) is much shorter than in 2 (2.7076(8) Å)7c or in Breher's pentasilapropellane Si5Mes6 (2.636 Å, Mes=2,4,6‐trimethylphenyl),9 but longer than in a Si11 siliconoid with a staggered arrangement of the substituent‐free vertices.4 The release of steric strain by the elimination of one of the Tip groups is reflected in a less‐acute angle between the two bridged “propeller blades” in 3 (Si1‐Si2‐Si3/Si1‐Si2‐Si4: 107.61°) versus the corresponding angle in 2 (96.7°).

Figure 1

The molecular structure of the anionic moiety in 3Li[12‐crown‐4]2 in the solid state. Countercation, hydrogen atoms, and co‐crystallized solvent molecules omitted for clarity. Thermal ellipsoids set at 50 % probability. Selected bond lengths [Å] and angles [°]: Si1–Si2 2.5506(9) Å, Si1–Si3 2.3871(9) Si1–Si4 2.3490(9), Si1–Si6 2.3575(9), Si2–Si3 2.4070(10), Si3–Si5 2.4202(9); Si1‐Si6‐Si2 65.35(3), Si6‐Si1‐Si2 57.50(3), Si4‐Si1‐Si6 98.01(3), Si1‐Si3‐Si5 83.50(3), Si4‐Si5‐Si3 85.06(3).

The similarity of reaction conditions during the synthesis of 3Li and of its precursor, the dismutational isomer 1, prompted us to also explore the direct reduction of 4 to 3Li without isolation of the intermediate 1. Indeed, treatment of a solution of 4 in Et2O with four equivalents of lithium/naphthalene in THF afforded 3Li in 51 % yield after crystallization.

With its anionic (and thus nucleophilic) silicon vertex, 3Li should be a valuable synthon for the further functionalization and transfer of the Si6R5‐scaffold. Based on the pronounced thermodynamic stability of 2 7c (representing the alleged global minimum on the Si6H6 potential‐energy surface27), 3Li can be regarded as the silicon analogue of phenyl lithium in carbon chemistry and might turn out to be equally useful for the transfer of the Si6R5 group.

Consequently, we investigated the reactions of 3Li with representative electrophiles (borane dimethyl sulfide complex BH3⋅SMe2, pivaloyl chloride tBuC(O)Cl, silicon tetrachloride SiCl4, and bis(dimethylamino)chlorophosphane ClP(NMe2)2). Indeed, the corresponding borate‐, carbonyl‐, silyl‐, and phosphanyl‐substituted siliconoids 5 a–d (Scheme 2) were obtained by straightforward combination of the reagents in toluene.26 All the reactions proceeded quantitatively according to NMR spectroscopy. The presence of intact Si6 cluster motifs akin to 2 was unequivocally confirmed by the 29Si NMR spectra of 5 a–d through the characteristic dispersions of chemical shifts (Table 1). Note that the two unsubstituted vertices, Si1 and Si2, give rise to two different signals in 29Si NMR spectra in all cases, despite an apparent chemical equivalence. We tentatively attributed this to hindered rotation of the functional groups. For 5 a and 5 c, broadened signals at room temperature suggested that coalescence was within reach. This assumption was verified by VT‐NMR in case of 5 c: cooling to 223 K resulted in significantly sharpened 29Si NMR signals at δ=−255.8 and −266.2 ppm. Coalescence is observed at 315 K, finally leading to a still somewhat broadened signal at δ=−256.7 ppm at 343 K. By application of the Eyring equation, the barrier of rotation for the SiCl3 group is estimated to ΔG ≠=58 kJ mol−1 (see Supporting Information). In 5 a, additionally, a broadening of the signal for Si3 (δ=−4.8 ppm) is observed as a result of the coupling with the quadrupolar 11B nucleus. In the 11B NMR spectrum of 5 a, the resonance at δ=−41.3 ppm is split into a quartet by coupling with the three chemically equivalent hydrogen atoms of the BH3 moiety (1 J B‐H=75 Hz). The 13C shift of the carbonyl carbon of 5 b is observed at δ=239.0 ppm in the typical range for silyl‐substituted ketones.28 In line with reported CO stretching frequencies,29 the two characteristic absorptions in the infrared spectrum of 5 b at ν=1637 and 1602 cm−1 are assigned to the asymmetric and symmetric C=O stretches of the carbonyl moiety.

Scheme 2

Syntheses of functionalized siliconoids 5 a–d. Reagents: a) BH3⋅SMe2, b) tBuC(O)Cl, c) SiCl4, d) ClP(NMe2)2. 5 a: E=BH3 −; 5 b: E=C(O)tBu; 5 c: E=SiCl3; 5 d: E=P(NMe2)2.

Table 1

Selected analytical data of functionalized siliconoids 3Li and 5 a–d (data for homoleptic siliconoid 2 7c for comparison).

Compound	Functional group E	29Si Si1, Si2  [ppm]	29Si Si6 [ppm]	Si1‐Si2 [Å]
3[Li(12‐crown‐4)2]	Li	−230.9, −232.6	152.2	2.5506(9)
5 a	BH3 −	−257.3, −265.0	161.2	2.620(1)
5 b	C(O)tBu	−264.7, −271.1	171.8	2.6430(6)
5 c	SiCl3	−252.3, −264.2	175.4	2.635(1)
5 d	P(NMe2)2	−256.0, −261.4	168.7	2.6231(5)
2 7c	Tip[a]	−274.2[a]	174.6[a]	2.7076(8)[a]

[a] Atom numbering differs in Ref. 7c.

For 5 c, an additional 29Si NMR resonance was observed at δ=−38.3 ppm for the SiCl3 substituent. In the phosphanyl‐substituted siliconoid 5 d, the 31P NMR resonance (δ=148.0 ppm) is strongly deshielded for a silicon‐substituted dialkylamino phosphane30 which we tentatively attribute to the cluster current in the Si6 core, also responsible for the unusual deshielding of one of the SiTip2 bridges.7c The silicon–phosphorus coupling constant of 1 J Si‐P=187 Hz in 5 d is fairly large for Si−P single‐bond coupling indicating high s character in this exohedral bond.31

Single crystals of 5 a–d were obtained in moderate to excellent yields (5 a: 60 %, 5 b: 38 %, 5 c: 68 %, and 5 d: 91 %) and X‐ray diffraction studies confirmed the formation of the corresponding functionalized Si6 siliconoids (Figure 2). The distances between the unsubstituted bridgehead atoms in 5 a–d are considerably shorter than in the all Tip‐substituted species 2, but longer than in 3Li (Table 1). The longest wavelength absorptions in the UV/Vis spectra of 5 a–d (5 a,b,d: λ max=475 nm, 5 c: λ max=460 nm) are in good agreement with that observed for 2 (λ max=473 nm) and are thus assigned to the HOMO–LUMO vertical singlet excitation.7c The intensity of the absorptions varies with the type of substituent linked to the Si6 backbone (ɛ=220 (5 a); 545 (5 b); 834 (5 c), 1212 m −1 cm−1 (5 d)).

Figure 2

Molecular structures of functionalized Si6‐siliconoids in the solid state: a) borate 5 a; b) ketone 5 b; c) trichlorosilane 5 c; d) bis(dimethylamino)phosphane 5 d. Hydrogen atoms, disordered Tip‐iPr groups, and co‐crystallized solvent omitted for clarity. Thermal ellipsoids set at 50 % probability.

In summary, by reduction of dismutational hexasilabenzene isomer 1, we obtained the first example of an anionic unsaturated silicon cluster 3, which closes the conceptual gap between siliconoids and Zintl anions of silicon. As the previously reported neutral Si6‐siliconoid 2 7c represents the likely global minimum on the Si6H6 potential‐energy surface,27 the structurally related anionic siliconoid 3 could be seen as a silicon analogue of phenyl lithium, if only in terms of thermodynamic stability. Indeed, it turned out to be a powerful synthon for the transfer the intact Si6 cluster motif (with both unsubstituted vertices untouched) to Group 13, 14, and 15 electrophiles as shown by the preparation of the representative set of functionalized Si6‐siliconoids 5 a–d.

Dedicated to Prof. Armin Berndt on occasion of his 79th birthday

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.