Selective One‐Pot Syntheses of Mixed Silicon‐Germanium Heteroadamantane Clusters

Abstract

Si_xGe_y alloys are emerging materials for modern semiconductor technology. Well‐defined model systems of the bulk structures aid in understanding their intrinsic characteristics. Three such model clusters have now been realized in the form of the Si _x Ge _y heteroadamantanes [0], [1], and [2] through selective one‐pot syntheses starting from Me₂GeCl₂, Si₂Cl₆, and [nBu₄N]Cl. Compound [0] contains six GeMe₂ and four SiSiCl₃ vertices, whereas one and two of the GeMe₂ groups are replaced by SiCl₂ moieties in compounds [1] and [2], respectively. Chloride‐ion‐mediated rearrangement quantitatively converts [2] into [1] at room temperature and finally into [0] at 60 °C, which is not only remarkable in view of the rigidity of these cage structures but also sheds light on the assembly mechanism.

Toward novel nanoclusters: The fourfold SiCl₃‐substituted Si₄Ge₆ ([0]), Si₅Ge₅ ([1]), and Si₆Ge₄ ([2]) heteroadamantane cores are accessible directly from Me₂GeCl₂ and Si₂Cl₆/Cl⁻ in a time‐ and cost‐efficient synthesis. In a chloride ion‐mediated rearrangement, [2] is converted into [1] at room temperature and into [0] at 60 °C.

Introduction

Bulk silicon is the materials basis of semiconductor technology. For the deposition of silicon thin films, oligosilanes have been intensively studied and used as volatile precursors. ^[1] Marschner's sila‐adamantane [A] is a substructure of bulk cubic silicon and a particularly fine example of a large, monodisperse oligosilane (Figure 1a). ^[2] The incorporation of Ge atoms into bulk silicon can lead to Si _x Ge _y alloys with unprecedented optoelectronic properties of exceptional promise. ^[3] To fully exploit the potential of this class of materials, deeper insight into fundamentally important phenomena, such as σ‐electron conjugation, ^[4] would be desirable and can best be gained by studying well‐defined molecular model systems. Apart from Kouvetakis’ perhydrogenated single‐source Si _x Ge _y precursors, which have been successfully used for the CVD of corresponding mixed semiconductors,[ ⁵ , ⁶ ] only few examples of complex (polycyclic) Si _x Ge _y oligomers are known to‐date, making a systematic assessment of their properties difficult.[ ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ ] In this regard, the neo‐Si₃Ge₂ structure [B] (Figure 1b) is noteworthy, which was obtained by the du Mont group from Me₃GeCl and HSiCl₃/NEt₃ (Benkeser reagent).[ ¹⁵ , ¹⁶ ] The analogous reaction with Me₂GeCl₂ led to the double silylation product Me₂Ge(SiCl₃)₂. Recently, our group succeeded in synthesizing germanide [C] ⁻ from GeCl₄ and the alternative trichlorosilylation system Si₂Cl₆/Cl⁻,[ ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ ] which disproportionates into SiCl₄ and the actual reactive intermediate [SiCl₃]⁻; treatment of [C] ⁻ with AlCl₃ gave the neo‐Si₄Ge species [D]. ^[21]

Figure 1.

a) Solid‐state structure of Marschner's sila‐adamantane [A] (Si: blue, CH₃: black). b) Schematic representations of the Si _x Ge _y oligomers [B], [C] ⁻, and [D].

Herein, we describe reactions of Me₂GeCl₂ with the Si₂Cl₆/Cl⁻ system and show that, in striking contrast to du Mont's results with the Benkeser reagent, three structurally defined Si _x Ge _y heteroadamantanes, [0], [1], and [2], become accessible in good yields, which can be regarded as long‐sought model systems of Si _x Ge _y alloys (see Figure 2 for the molecular structures and an explanation of the numbering scheme).

Figure 2.

Syntheses of the Si _x Ge _y heteroadamantanes [0], [1], and [2] from Me₂GeCl₂ and the Si₂Cl₆/Cl⁻ system in CH₂Cl₂. Note that the reactions are catalytic in [nBu₄N]Cl; in practice ca. 1 equiv. was used. The compound numbers [X] refer to the number, X, of SiCl₂ vertices incorporated into the cluster core instead of GeMe₂ vertices (ideal number of the latter: 6 in [0]). (i) 1: room temperature, 4 h; 2: removal of SiCl₄; 3: 60 °C, 6 d. (ii) room temperature, 13 d, in‐situ crystallization from an unstirred mixture. (iii) room temperature, 6 d, with stirring. (iv) 60 °C, 2 d. (v) 60 °C, 4.5 d.

Results and Discussion

Syntheses and reactivities of the heteroadamantanes [0], [1], and [2]

All reactions were carried out in CH₂Cl₂ or CD₂Cl₂. Our initial experiments with Me₂GeCl₂, Si₂Cl₆, and cat. [nBu₄N]Cl using the theoretically required stoichiometry for the formation of the Si₈Ge₆ heteroadamantane [0] (i. e., Me₂GeCl₂/Si₂Cl₆ 6 : 16; see Figure S1 in the Supporting Information for the atom and electron count) gave the target compound in 20 % yield. Further optimization of the reaction conditions led to the following protocol for the synthesis of [0]: in a one‐pot procedure, a Me₂GeCl₂/Si₂Cl₆ 6 : 24 mixture was first stored at room temperature for 4 h, evaporated to remove the released SiCl₄, re‐dissolved, and heated to 60 °C for 6 d.

Heteroadamantane [0] then crystallized from the solution and was isolated in 40–45 % yield (see below for a rationale of the modified stoichiometry). Single crystals of a second heteroadamantane, the Si₁₀Ge₄ species [2], grew and were isolated after 13 d in 25–30 % yield, when a 6 : 24 mixture of Me₂GeCl₂/Si₂Cl₆ was stored at room temperature without stirring. Compared to [0], in [2] two GeMe₂ vertices at opposite positions of the cluster are replaced by SiCl₂ groups. A further increase in the crystallization time caused an increasing contamination of the crystal crop by a third heteroadamantane, the Si₉Ge₅ derivative [1], in which only one of the six GeMe₂ vertices of [0] is exchanged for SiCl₂. Marschner prepared sila‐adamantane [A], the close molecular relative to [0], [1], and [2], in a fundamentally different way by a reaction inspired by Schleyer's adamantane synthesis (8 steps from SiCl₄ and Me₃SiLi, 17 % overall yield).[ ² , ⁹ ]

How are the formations of [0], [1], and [2] interrelated? While pure [2] is stable over weeks in CH₂Cl₂ at room temperature, NMR monitoring proved that a continuous conversion [2]→[1]→[0] is possible in the presence of [nBu₄N]Cl (quantitative with respect to GeMe₂ fragments; Figure S2): Reaction [2]→[1] already takes place at room temperature, whereas reaction [1]→[0] requires prolonged heating at 60 °C. This temperature dependence allows the selective synthesis of [1] from [2] in 50–55 % yield after workup (Figure 2). Isolation of [2] is thus only possible if [2] is allowed to escape rearrangement by crystallization. Upon going from [2] to [1] and [0], dichlorosilylenes (SiCl₂) are extruded from the cluster cores and dimethylgermylenes (GeMe₂) are incorporated. Formal cyclocondensation of 6 SiCl₂ moieties would give perchlorinated cyclohexasilane, which was indeed detected by ²⁹Si NMR spectroscopy in the form of [cyclo‐Si₆Cl₁₂ ⋅ 2Cl]²⁻.[ ²² , ²³ , ²⁴ , ²⁵ , ²⁶ ] The GeMe₂ fragments, in turn, must originate from cannibalized heteroadamantanes [2] and [1].

The following conclusions can be drawn: i) the assembly of [0] most likely involves Si‐enriched [2] as a key intermediate, which explains why the best yields of [0] are obtained when the starting materials are combined in the stoichiometry theoretically required for the synthesis of [2] (Figure S1). ii) Since the sequence [2]→[1]→[0] cannot be reversed by heating of [0] with [nBu₄N]Cl and SiCl₄ or Si₂Cl₆, it apparently represents the downhill pathway to the thermodynamically most favorable species. iii) The reaction critically depends on certain properties peculiar to Ge, because the use of Me₂SiCl₂ instead of Me₂GeCl₂ does not lead to the corresponding Si₁₄ heteroadamantane (Me₂SiCl₂ rather behaved as an innocent bystander of the Cl⁻‐induced Si₂Cl₆ disproportionation ^[17] ).

X‐ray crystal structure analysis of the heteroadamantanes [0], [1], and [2]

Compound [0] crystallizes from CH₂Cl₂ as C ₁‐symmetric solvate [0] ⋅ CH₂Cl₂. ^[27] The heteroadamantane cluster core is built of six Ge and four Si vertices, arranged in a perfectly alternating manner (Figure 3a). The valences of each Ge or Si vertex are saturated by two Me groups or one SiCl₃ substituent, respectively. Thus, [0] combines the structural motifs of neo‐pentatetrelanes and (fused) cyclohexatetrelanes, both of which are frequently encountered in products of Si₂Cl₆ disproportionation (e. g., [cyclo‐Si₆Cl₁₂ ⋅ 2Cl]²⁻ and Si(SiCl₃)₄).[ ²³ , ²⁸ , ²⁹ ] The average Si−Ge bond length of [0] (2.395 Å) is essentially the same as that determined for SiGe alloy in the bulk phase (2.398 Å). ^[30]

Figure 3.

a) Solid‐state structure of [0] (Si: blue, Ge: purple, Cl: yellow‐green, CH₃: black); Si₈Ge₆ core (left), complete structure (right). b) Schematic representations of [0], [1], and [2] with ²⁹Si NMR chemical shift values given for comparison. c) Sections of the ²⁹Si{¹H} NMR spectra of [0], [1], and [2] in CD₂Cl₂. d) ²⁹Si/¹H HMBC NMR spectrum of [1] to prove the proposed structure containing 3 (2) magnetically inequivalent kinds of Me groups (quaternary Si vertices) in the molecule (CD₂Cl₂).

In the solid state, the molecules of [2] and [1] are located on a threefold rotation axis and a mirror plane, respectively. The GeMe₂ groups are disordered with SiCl₂ moieties. Structure refinement gave the best figures‐of‐merit when the sum of site occupation factors of all GeMe₂ groups was constrained to 4 (rather than 5 or 6) in the case of [2] and 5 (rather than 4 or 6) in the case of [1] (see the Supporting Information for more details). X‐ray analysis thus supports the proposed molecular structures of [1] and [2], but the proof could only be gained in combination with NMR spectroscopy.

NMR spectroscopic characterization of the heteroadamantanes [0], [1], and [2]

The Me groups of [0] give rise to one ¹H (0.91 ppm) and one ¹³C NMR signal (2.6 ppm); the ²⁹Si{¹H} NMR spectrum is characterized by two resonances at −83.4 (SiSiCl₃) and 16.2 ppm (SiSiCl₃; Figure 3b,c). The number of signals is in line with an average T_d symmetry of [0] in solution, and the ²⁹Si chemical shift values agree with those of the reference compounds [B] (−84.2 ppm, SiSiCl₃; 17.2 ppm, SiSiCl₃) ^[15] and Si(SiCl₃)₄ (−80.9 ppm, SiSiCl₃; 3.5 ppm, SiSiCl₃). ^[28] A ²⁹Si/¹H HMBC NMR experiment on [0] gave a pronounced crosspeak between the signals at −83.4 ppm (²⁹Si) and 0.91 ppm (¹H), in line with the direct Si(quart)−GeMe₂ bond that is the principal interaction within the heteroadamantane scaffold.

Compound [2] (point group D _2d ) retains high symmetry and thus chemically equivalent Me groups, but the corresponding shift values, δ(¹H)=1.03 ppm and δ(¹³C)=1.6 ppm, differ slightly from those of [0]. Three signals are detectable in the ²⁹Si{¹H} NMR spectrum, two of them (−81.0 ppm, SiSiCl₃; 11.9 ppm, SiSiCl₃) appear in the same ranges as the two resonances of [0], the third one is assignable to the SiCl₂ centers (29.6 ppm; Figure 3b, c). Crosspeaks are observed between the GeMe₂ and SiSiCl₃ as well as SiCl₂ signals in the ²⁹Si/¹H HMBC NMR spectrum of [2].

Compound [1] (point group C _2v ) shows three ¹H NMR resonances with integral values of 6H, 12H, and 12H. The five signals visible in the ²⁹Si{¹H} NMR spectrum can be assigned to two chemically inequivalent SiSiCl₃ units (−80.7, −83.3 ppm), two inequivalent SiSiCl₃ moieties (16.2, 12.1 ppm), and one SiCl₂ vertex (31.0 ppm; Figure 3b,c). Importantly, the SiSiCl₃ signal at −80.7 ppm shows only crosspeaks to the two more intense proton resonances in the ²⁹Si/¹H HMBC NMR spectrum, whereas the signal at −83.3 ppm couples to all Me groups present in [1] and consequently corresponds to the two quaternary Si atoms that are linked to the unique GeMe₂ group (Figure 3d).

Conclusion

In summary, time‐ and cost‐efficient one‐pot syntheses of Si₈Ge₆, Si₉Ge₅, and Si₁₀Ge₄ heteroadamantanes [0], [1], and [2] from the simple, commercially available building blocks Me₂GeCl₂, Si₂Cl₆, and [nBu₄N]Cl have been disclosed. The clusters obtained are subunits of bulk cubic Si _x Ge _y alloys with the advantage of containing the two elements in different stoichiometries. Theory predicts that a Si₁₀ cluster is already large enough to exhibit representative features of Si nanoparticles. ^[31] We therefore conclude that our Si _x Ge _y heteroadamantanes help to bridge the gap between small Si _x Ge _y molecules, such as [B] and [D], and more‐extended Si _x Ge _y nanoclusters. The effects of doping the adamantane scaffold with varying numbers of Si and Ge atoms have so far been studied only theoretically. Considerable consequences for the optoelectronic properties of the individual compounds have been predicted[ ³² , ³³ , ³⁴ ] and can now be experimentally confirmed (cf. optical band gaps of 4.35, 4.43, and 4.56 eV for [0], [1], and [2], respectively; Table S2). According to works of Tamao et al., ^[35] the all‐anti conformation of pairs of SiCl₃ substituents in [0], [1], and [2] should result in pronounced σ‐conjugation, thereby rendering our heteroadamantanes suitable building blocks for the fabrication of Si _x Ge _y ‐based molecular wires. ^[36] The fact that [0], [1], and [2] carry exohedral SiCl₃ substituents should be of benefit in this context. Thus, similar to the discovery of carbonaceous diamondoids and the elaboration of their remarkably high application potential,[ ³⁷ , ³⁸ , ³⁹ ] the successful synthesis of [0], [1], and [2] is expected to pave the way to novel and useful Si _x Ge _y nanostructures.

Conflict of interest

B.K., H.‐W.L., and M.W. are inventors on patent application PCT/DE2021 100470 submitted by Johann Wolfgang Goethe‐Universität, which covers the synthesis and use of [0], [1], and [2].

Supporting information

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.

Supporting Information

B. Köstler, M. Bolte, H.-W. Lerner, M. Wagner, Chem. Eur. J. 2021, 27, 14401.