Atomically Precise Expansion of Unsaturated Silicon Clusters

Abstract

Small‐ to medium‐sized clusters occur in various areas of chemistry, for example, as active species of heterogeneous catalysis or as transient intermediates during chemical vapor deposition. The manipulation of stable representatives is mostly limited to the stabilizing ligand periphery, virtually excluding the systematic variation of the property‐determining cluster scaffold. We now report the deliberate expansion of a stable unsaturated silicon cluster from six to seven and finally eight vertices. The consecutive application of lithium/naphthalene as the reducing agent and decamethylsilicocene as the electrophilic source of silicon results in the expansion of the core by precisely one atom with the potential of infinite repetition.

Small‐ to medium‐sized clusters assume key roles in various technologically important areas. In heterogeneous reactions, metal and metal compound clusters are powerful catalysts for chemical transformations on an industrial scale.1 Metal and semi‐metal clusters constitute key intermediates during the chemical vapor deposition of thin films and bulk2 as well as during the bottom‐up synthesis of nanoparticles.3 Although a remarkable number of stable molecular clusters have been reported, for example, Si₃₂R₄₅ ⁻,4 (CdSe)₃₄,5 R₂₀Al₇₇ ²⁻,6 and R₄₄Au₁₀₂ 7 (R=ligand or functional group), they are typically obtained in a non‐systematic or even serendipitous manner, often entailing difficulties with yield and reproducibility.

Unsaturated silicon clusters with partial substitution (siliconoids) have attracted attention for their role as proposed intermediates in gas‐phase deposition processes. With their unsubstituted vertices, they share important surface features of both nanoparticles and the elemental bulk.8 The preference of silicon for single bonds as opposed to the competitive strength of multiple bonds in the corresponding carbon systems leads to a growing dominance of cluster motifs when the number of silicon core atoms is increased.9 For example, the first stable siliconoid, Si₅R₆ 10a (1; Scheme 1), adopts a tricyclic framework with one unsubstituted silicon vertex, and the isomeric pentasilapropellane 2 even features two “naked” cluster atoms.10b The low‐energy Si₆R₆ isomers 3 10c and 4 10d are both characterized by a cluster framework with two unsaturated vertices, in marked contrast to the iconic benzene, which is by far the lowest energy structure on the C₆H₆ potential energy surface.11

Scheme 1.

Selected saturated and unsaturated silicon clusters. 1, 3, 4, 5 a: R=Tip=2,4,6‐iPr₃C₆H₂; 2: R=2,4,6‐Me₃C₆H₂; 5 b: R=2,6‐iPr₂C₆H₃; 6 a, 7 a, 8: R=tBu₃Si; 6 b, 7 b: R=((Me₃Si)₂CH)₂(Me)Si; 9, 11: (R)₂=C(SiMe₃)₂CH₂CH₂C(SiMe₃)₂, R′=tBu; 10: (R)₂=C(SiMe₃)₂C(Me)₂C(SiMe₃)₂; 11: D=p‐(dimethylamino)pyridine. a: KC₈; b: ICl.

While solvated Zintl ions—charged, polyhedral clusters—of germanium and tin can be used as well‐defined precursors for the synthesis of partially substituted molecular clusters,12 the strongly reducing silicon congeners have only been employed as ligands towards a few redox‐stable transition‐metal fragments,13 and were only very recently protonated to the partially hydrogen‐substituted clusters [HSi₉]³⁻ and [H₂Si₉]²⁻.14 Therefore, siliconoids are typically accessed by reductive cleavage of competent leaving groups from classical saturated precursors. This approach was initially demonstrated by the preparation of saturated polyhedral silicon clusters such as hexasilaprismane R₆Si₆ 5 a, b 15 and tetrasilatetrahedrane R₄Si₄ 6 a, b 16 by reductive dimerization of the corresponding linear R₂Si₂X₄ or cyclic R₃Si₃X₃ building blocks. Functionalization of the saturated Si₄ core was achieved with KC₈ to yield isolable tetrasilatetrahedranides K⁺[R₃Si₄ ^‐] 7 a, b.16b,16c Oxidation of 7 a with iodine monochloride gave siliconoid R₆Si₈ 8,17 which had previously been synthesized from a different R₃Si₄ precursor by reduction.18 Further examples of isolable siliconoids were reported by Iwamoto19 (9, 11) and Kyushin20 (10). Despite the isolated reports of core expansion of siliconoids,15b, 19 the number of atoms in the scaffold, once assembled, cannot be increased in a systematic way.

Based on our report on the first anionic siliconoid with partial substitution Si₆R₅ ⁻ (12; R=Tip=2,4,6‐triisopropylphenyl),21 we sought to develop an atomically precise cluster expansion procedure. As the leaving group properties of the pentamethylcyclopentadienyl ligand (Cp*) are well‐established in main group chemistry,22 we speculated that Jutzi's decamethylsilicocene (SiCp*₂;23 despite the decacoordinate silicon atom) might react with 12 as an electrophilic source of precisely one silicon atom. Previously, only the corresponding cation Cp*Si⁺BAr₄ ⁻ (Ar=pentafluorophenyl)24 had been employed as electrophile.25 As proof of principle, we carried out the reaction of R₂Si=Si(R)Li with SiCp*₂. Indeed, the literature‐known Cp*‐substituted cyclotrisilene, cyclo‐Si₃Tip₃Cp*, was obtained in 90 % yield (determined by NMR spectroscopy).26

More importantly, treatment of the anionic siliconoid 12 with one equivalent of SiCp*₂ led to a uniform product 13 accompanied by precipitation of Cp*Li (Scheme 2). ¹H NMR analysis of 13 in C₆D₆ confirmed the presence of five nonequivalent Tip groups. The additional singlet resonance at 1.69 ppm corresponds to 15 hydrogen atoms and is assigned to the remaining Cp* substituent. The characteristic wide distribution of ²⁹Si NMR chemical shifts of siliconoid 4 and its derivatives (typically from about −270 ppm for the “naked” vertices to +175 ppm for the adjacent R₂Si bridge)8 was retained in the new product. A third strongly shielded resonance at −138.4 ppm without a cross‐peak in the ²⁹Si–¹H correlation NMR spectrum suggested the presence of an additional silicon vertex without a substituent. As the predicted global‐minimum isomer of the Si₇H₆ potential energy surface 13_H 27 has precisely three “naked” vertices, we calculated the ²⁹Si NMR spectrum of a B3LYP/def2‐TZVP‐optimized model system for 13 corresponding to the experimental case (13_Tip) at the OLYP/def2‐TZVP level of theory.28 The excellent agreement of the calculated chemical shifts of 13_Tip with those experimentally observed for 13 suggests very similar constitutions (Table 1; much less satisfactory agreement was achieved with the B3LYP functional; see Table S9, Supporting Information).

Scheme 2.

Cluster expansion from the anionic Si₆ siliconoid 12 to the Si₇ siliconoids 13 and 14 and finally the Si₈ siliconoids 15 and 16. R=Tip=2,4,6‐iPr₃C₆H₂, Cp*=1,2,3,4,5‐pentamethylcyclopentadienyl.

Table 1.

Experimental (13, 14⋅Li⁺(thf)₃, 15) and calculated (13_Tip, 14_Tip⋅Li⁺(thf)₃, 15_Tip) ²⁹Si NMR chemical shifts [ppm] at the OLYP/def2‐TZVP level of theory. Atoms of 15 and 15_Tip are numbered as in 16.

	13	13Tip	14	14Tip	15	15Tip
Si1	−229.6	−237.6	−191.9	−197.0	−138.9	−151.9
Si2	−241.9	−255.5	−195.7	−204.8	−165.1	−160.5
Si3	181.9	180.3	284.3	292.1	18.7	5.9
Si4	2.6	10.6	10.8	18.1	33.5	41.0
Si5	−138.4	−139.4	−66.1	−73.1	−81.7	−82.2
Si6	15.0	20.2	60.1	78.4	111.8	115.5
Si7	156.0	154.6	148.0	150.7	60.3	80.2
Si8					206.9	201.0

The structure of siliconoid Si₇R₅Cp* in the solid state was finally confirmed as the first stable example of a neutral siliconoid with three adjacent unsubstituted vertices 13 by X‐ray diffraction on orange‐red crystals obtained in 53 % yield (Figure 1).29 The cluster scaffold is almost indistinguishable from that predicted for 13_H and 13_Tip.27 The three unsubstituted vertices form an isosceles triangle, with the base and apex bridged on both sides by electron‐precise disilanyl units.

Figure 1.

Molecular structure of siliconoid 13 in the solid state. Hydrogen atoms are omitted for clarity. Thermal ellipsoids set at 50 % probability. Selected bond lengths [pm] and angles [°]: Si1–Si2 264.8(1), Si1–Si5 234.41(9), Si1–Si3 231.63(9), Si2–Si5 234.19(9), Si3–Si4 241.81(9), Si4–Si5 238.86(9), Si5–Si6 238.42(9), Si6–Si7 241.55(9); Si1‐Si5‐Si2 68.82(3), Si1‐Si3‐Si2 69.45(3), Si4‐Si5‐Si6 173.77(4).

The Si1–Si2 distance of 264.8(1) pm is relatively long, while all Si−Si bonds of the cluster core (231.8–241.8 pm) show typical Si−Si single bond lengths. Alternatively, the Si7 scaffold of siliconoid 13 can be described as a persilapropellane with twofold R₂Si bridging of the propeller blades and an unprecedented seesaw‐type tetracoordination of Si5 as shown by the close‐to‐linear bond angle Si4‐Si5‐Si6 of 173.88(4)°.

The remaining Cp* group in 13 should allow for the regeneration of the anionic functionality after the first core expansion step. The Si₇ siliconoid 13 was therefore treated with two equivalents of lithium/naphthalene in tetrahydrofuran (thf) at −100 °C in order to obtain the anion 14. ¹H NMR spectroscopy of the uniform reaction mixture confirmed the presence of five Tip substituents, while the resonance of the Cp* group had indeed completely vanished. Instead, characteristic signals for three equivalents of coordinated thf suggested the incorporation of a lithium cation into the molecule. The ²⁹Si NMR spectrum exhibited a similar distribution of the seven signals as the starting material 13. The most remarkable difference is the downfield shift of the formerly Cp*‐bonded silicon vertex by about Δδ=100 ppm, which was accompanied by substantial line broadening due to coupling to the quadrupolar ⁷Li nucleus (δ(⁷Li)=−0.15 ppm). In contrast, the unsubstituted vertices are only shifted by Δδ=40 to 70 ppm. A similar relative deshielding had been observed between the neutral Si₆ siliconoid 4 and the corresponding lithiated species 12.21 Applying the OLYP functional, the calculated ²⁹Si chemical shifts of the B3LYP‐optimized model Tip₅Si₇(Li)thf₃ (14_Tip⋅Li⁺(thf)₃) nicely reproduce the experimental data of 14⋅Li⁺(thf)₃ (Table 1).

A crystalline sample of 14 was obtained from pentane in 42 % yield. As shown by single‐crystal X‐ray diffraction, the lithiated siliconoid 14⋅Li⁺(thf)₃ crystallizes as a contact ion pair from pentane solution (see the Supporting Information). The structural parameters of the cluster scaffold of 14⋅Li⁺(thf)₃ are similar to those of the neutral precursor 13 except for the noticeably shortened base of the central triangle (Si1–Si2=254.48(5) pm; Δ=10.3 pm). In order to assess the influence of the close contact of the lithium cation to the anionic Si₇ siliconoid, the solvent‐separated ion pair 14[Li⁺(dme)₃] was prepared by crystallization from a mixture of hexane and 1,2‐dimethoxyethane in 73 % yield. Except for significant sharpening of the ²⁹Si NMR resonance of Si3 due to the absence of a direct interaction with the quadrupolar ⁷Li nucleus, all other resonances remained essentially unchanged. Similarly, the solid‐state structure of 14[Li⁺(dme)₃] is hardly influenced by the spatially distant Li⁺ countercation (Figure 2).

Figure 2.

Molecular structure of the solvent‐separated anionic siliconoid 14 in the solid state. Hydrogen atoms and the [Li⁺(dme)₃] countercation are omitted for clarity. Thermal ellipsoids set at 50 % probability. Selected bond lengths [pm] and angles [°]: Si1–Si2 255.96(8), Si1–Si5 233.77(8), Si1–Si3 239.37(8), Si2–Si5 231.96(8), Si3–Si4 244.82(8), Si4–Si5 235.68(8), Si5–Si6 237.59(8), Si6–Si7 241.12(7); Si1‐Si5‐Si2 66.68(2), Si1‐Si3‐Si2 64.16(2), Si4‐Si5‐Si6 174.00(3).

With anionic siliconoid 14 in hand, the obvious next step was the incorporation of an eighth silicon vertex. Similar to the synthesis of neutral Si₇ cluster 13 from anionic 12, the reaction of 14 with SiCp*₂ yielded a uniform product 15 (90 % spectroscopic purity). In this case, the observation of only three of the eight resonances at high field (δ=−165.1, −138.9, −81.7 ppm) suggested the migration of an aryl group from a disubstituted SiR₂ unit to a formerly “naked” vertex. Indeed, a ¹H–²⁹Si correlation spectrum revealed only one SiR₂ unit (δ=33.5 ppm), while four resonances (δ=111.8, 60.3, 35.5, 18.7 ppm) are due to SiR vertices bonded to just one substituent.

In one crystallization attempt, a few orange single crystals were collected and then investigated by X‐ray diffraction (Figure 3). Instead of the Si₈R₅Cp* siliconoid 15, however, its water adduct 16 with a central unit of three unsubstituted silicon atoms was obtained. The formerly almost linear bond angle of the Si₇ siliconoids 13 and 14 at Si5 is with 149.53(5)° much more acute in 16. Apparently, the release of strain results in an approximation of regular tetrahedral coordination at Si5.

Figure 3.

Molecular structure of siliconoid 16 in the solid state. Hydrogen atoms (except OH and SiH) and co‐crystallized solvent molecules are omitted for clarity. Thermal ellipsoids set at 50 % probability. Selected bond lengths [pm] and angles [°]: Si1–Si2 263.8(1), Si1–Si5 234.2(1), Si1–Si8 235.70(13), Si2–Si5 233.97(13), Si2–Si8 232.7(1), Si7–Si8 236.9(1), Si6–Si7 237.6(1), Si5–Si6 232.3(1), Si6–O 169.0(3); Si1‐Si5‐Si2 68.60(4), Si4‐Si5‐Si6 149.53(5).

The deliberate addition of one equivalent of H₂O to 15 led to the near quantitative formation of 16 (85 % spectroscopic yield). Using the structure of 16 (minus H₂O) as a reasonable starting point, we optimized the geometry of 15_Tip at the B3LYP/def2‐TZVP level of theory. The resulting R₆Si₈ cluster is related to the Si₇ species 13 inasmuch as one of the SiR₂ moieties is formally replaced by an exohedral unsaturated Si=Si bridge (Si6 and Si7). The calculated HOMO and LUMO of 15 are predominantly constituted by the π and π* components at this Si=Si moiety, which readily explains the selective H₂O addition across this bond (Figure 4). The experimental ²⁹Si NMR signals of 15 at δ=111.8 (Si6) and 60.3 ppm (Si7) can be assigned to the exohedral Si=Si unit on the basis of their similarity to those reported for 1,2‐disilyl‐1,2‐diaryl disilenes with chemical shifts between 85 and 130 ppm.30 Unsymmetrical substitution is well‐known to lead to polarized Si=Si bonds with sometimes extremely differing chemical shifts for the tricoordinate Si atoms.31 All calculated ²⁹Si chemical shifts of 15_Tip (obtained with the OLYP functional) show a convincing agreement with the experimental data. Those of the exohedral Si=Si moiety at δ=115.5 (Si6) and 80.2 ppm (Si7) are (as usually observed for atoms with bonding including a pronounced π‐component) slightly overestimated, but otherwise nicely reproduce this trend (Table 1). Indeed, the calculated structure of 15_Tip confirms appreciable pyramidalization at the relatively upfield‐shifted silicon atom Si7 with a sum of bond angles of Σ=333.4°, while the downfield‐shifted silicon atom Si6 is much closer to planarity (Σ=358.1°). Consistent with the electrophilic nature of Si6 in 15, the OH group is attached to the corresponding atom of the hydrolysis product 16.

Figure 4.

Calculated HOMO (left) and LUMO (right) of Si₈ siliconoid 15_Tip.

In conclusion, with the systematic transformation of the anionic Si₆ siliconoid 12 (R=Tip) into species with seven and even eight vertices 13, 14, and 15, we have provided proof‐of‐concept results for the stepwise and repeated expansion of clusters with atomic precision. Key to this method is the use of decamethylsilicocene as an electrophilic source of divalent silicon with pentamethylcyclopentadienide ligands (Cp*) as anionic leaving groups. As our process allows for the regeneration of the anionic functionality after the expansion step, a rapid increase in the number of available siliconoids (including heteroatom‐doped variants) can be expected. The ubiquitous use of Cp* throughout the Periodic Table may allow for application well beyond the realm of silicon clusters.

Conflict of interest

The authors declare no conflict of interest.

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.

Supplementary

K. I. Leszczyńska, V. Huch, C. Präsang, J. Schwabedissen, R. J. F. Berger, D. Scheschkewitz, Angew. Chem. Int. Ed. 2019, 58, 5124.