Chemists Crack a 50-Year Problem by Taming Silicon’s Most Reactive Cation

Silylium ions — the positively charged, three-coordinate cousins of the familiar carbocation — have long been prized by chemists for their ferocious electrophilicity. These silicon-centred cations can activate otherwise inert bonds, making them powerful reagents and catalysts in modern synthesis. Yet for half a century, one glaring gap has persisted in the toolkit used to make them: nobody could reliably prepare silylium ions bearing three different substituents on silicon. Now, a team led by Martin Oestreich at the Technische Universität Berlin, together with computational chemists around Mu-Hyun Baik at the Institute for Basic Science and KAIST in South Korea, reports in Nature Synthesis a general protocol that finally closes this gap, using nothing more exotic than ordinary halosilanes and a dash of Lewis superacidic trickery.

The problem, it turns out, is a phenomenon chemists call substituent scrambling. The classical route to silylium ions, introduced by J. Y. Corey in 1975, starts from hydrosilanes and removes the hydride with a trityl cation paired with a weakly coordinating anion. Oestreich’s laboratory later added a complementary protolysis strategy, exploiting the Brønsted superacidity of Reed’s benzenium carborate salt to cleave even inert silicon–carbon bonds. Both methods work beautifully — but only if you want a silylium ion with three identical substituents. Feed either method a mixed silane such as dimethylphenylsilane, and the acidic conditions trigger a rapid exchange of alkyl and aryl groups at silicon, ultimately delivering the homoleptic trimethylsilylium ion along with a tangle of by-products rather than the desired mixed cation.

Scrambling is not a minor nuisance; it is deeply rooted in the mechanism. Computational work from the Berlin group had previously shown that once the first silylium ion forms, it latches onto unreacted silane molecules to create stable Lewis adducts. In the case of alkylsilanes, a hydride-bridged bissilylhydronium ion forms and allows alkyl groups to swap positions. For arylsilanes, geminally bissilylated arenium ions take centre stage, and a calculated 1,2-silyl migration through a favourable five-membered transition state lowers the energy barrier for aryl/alkyl exchange. Both key intermediates have been isolated and characterized independently, so the scrambling pathway is no theoretical curiosity — it is a well-documented chemical reality that has defeated every attempt at generality.

Some researchers have turned scrambling to their advantage. The laboratory of Thomas Müller harnessed the unselective exchange to forge homoleptic triarylsilylium ions that would otherwise demand a synthetically demanding allyl-leaving-group approach, and the Berlin team showed computationally that redistribution can be steered toward trialkylsilylium carborates. But heteroleptic silylium ions — those with at least two different substituents on silicon — remained accessible only in the rare cases where bulky tert-butyl, isopropyl, triisopropylphenyl or terphenyl substituents kinetically freeze the exchange. For the vast majority of substituent combinations, the chemistry simply refused to cooperate.

The Berlin team’s insight came from an old observation: halosilanes, bearing one or more silicon–halogen bonds, are notoriously reluctant to redistribute their substituents under Lewis acidic conditions. The electron-withdrawing halogen deactivates the remaining alkyl groups toward electrophilic attack, raising the bar for the exchange reactions. If halosilanes could be converted directly into silylium ions by halide abstraction, the scrambling problem might be sidestepped entirely. There was a catch, however. Simple treatment of halosilanes with conventional electrophiles — sodium, potassium, silver or even trialkylsilylium cations paired with weakly coordinating anions — usually stops at the Lewis adduct stage. The rate-determining step, heterolytic cleavage of the silicon–halogen bond, only proceeds when the emerging cation is sufficiently stabilized, and most weakly coordinating anions are simply not nucleophilic enough to finish the job.

The missing ingredient was a stronger halophile. In 2025, the same Berlin laboratory isolated and crystallographically characterized halogen-substituted silylium ions, which rank among the strongest isolable Lewis superacids known, with calculated fluoride ion affinities reaching an extraordinary 1,004 kilojoules per mole. The researchers reasoned that such a superacidic fluorosilylium ion could wrest fluoride from a fluorosilane. The mechanism they envisioned is elegant: the fluorosilylium ion and the fluorosilane combine to form a bissilylated fluoronium ion, in which a fluoride bridges the two silicon centres in a three-centre, four-electron bond. In that adduct, the fluorosilylium fragment rivals the fluoride affinity of the target cation, tipping the balance toward silicon–halogen bond cleavage with help from the carborate anion.

The hypothesis held. Treating dimethylphenylfluorosilane with the preformed fluorosilylium carborate in n-hexane at room temperature delivered the heteroleptic dimethyl(phenyl)silylium salt in 81 percent isolated yield, with no detectable scrambling. Better still, the team developed a one-pot variant that sidesteps the need to isolate the shelf-unstable fluorosilylium reagent. Adding half an equivalent of the bench-stable benzenium carborate to the fluorosilane first triggers chemoselective dephenylation, generating the superacidic fluorosilylium ion in situ, which then abstracts fluoride from a second equivalent of substrate. This in situ protocol furnished the model silylium ion in 84 percent yield and worked equally well with the corresponding chloro-, bromo- and iodosilanes.

The scope of the method is strikingly broad. Trialkylsilylium ions bearing primary ethyl, n-propyl, n-butyl and benzyl substituents — combinations that would scramble catastrophically under established routes — were prepared using the preformed fluorosilylium reagent. Dialkyl(aryl), alkyldiaryl and triaryl congeners followed, including an example with three different substituents, n-butyl(methyl)phenylsilylium. Partially and fully fluorinated aryl substituents were tolerated in yields of 63 to 87 percent, allowing the team to rank the Lewis acidities of the products using Müller’s fluorobenzonitrile method; meta-fluorination and higher fluorination both increase acidity, the former because fluorine’s mesomeric donation is weaker at the meta position. Even strained five- and six-membered cyclic silylium ions were forged from cyclic chlorosilanes, and the protocol even improves access to homoleptic ions: liquid trimethylsilyl chloride replaces the awkward gaseous trimethylsilane, and clean triphenylsilylium emerges where hydride abstraction gives a complex mixture. Four of the new salts were characterized by X-ray diffraction, revealing the expected zwitterionic structures in which a bromine atom of the carborate cluster stabilizes the cationic silicon.

Density functional theory calculations at the B3LYP-D3(BJ)/ma-def2-TZVP level with a hexane solvation model explained why the method works and why scrambling vanishes. The in situ generation of the fluorosilylium ion proceeds through a stepwise protonation: the carborate acid, featuring a proton bridging two bromine atoms of the cluster, protonates the phenyl group with barriers of just 2.7 and 5.8 kilocalories per mole, after which the carborate anion attacks the silicon centre in a barrierless step. Crucially, when the fluorosilylium ion meets the fluorosilane, the fluorine-bridged adduct is far more stable than the methyl- or aryl-bridged intermediates that would lead to scrambling — by 15.7 and 7.9 kilocalories per mole, respectively — because the three-centre, four-electron Si–F–Si bond involves minimal electronic distortion. The carborate anion then migrates around the delocalized cation and delivers a backside nucleophilic attack with a barrier of only 1.8 kilocalories per mole, completing the halide abstraction with a thermodynamic driving force of 7.8 kilocalories per mole.

The implications reach well beyond one reaction class. Just as alkyl halides serve as the standard precursors in carbocation chemistry, halides bound to silicon can now act as nucleofugal groups for silylium generation — a conceptual unification the authors describe as closing a major gap in silicon chemistry. Because heteroleptic silylium ions are prochiral, the route opens the door to asymmetric applications of these potent Lewis acids, and essentially the entire family of silylium ions, acyclic and cyclic, secondary and tertiary, is now within synthetic reach. Notably, the carborate anion sheds its passive reputation as a weakly coordinating bystander: it actively cooperates in protonation, dephenylation and halide abstraction alike. For a field that has wrestled with scrambling since the 1970s, the ability to assemble silicon cations substituent by substituent, at ambient temperature and from commodity halosilanes, marks a genuine turning point.

Subject of Research: Scrambling-free synthesis of heteroleptic silylium ions by halide abstraction from halosilanes using Lewis superacids

Article Title: Scrambling-free synthesis of heteroleptic silylium ions by halide abstraction from halosilanes

Article References: Scrambling-free synthesis of heteroleptic silylium ions by halide abstraction from halosilanes. (n.d.). https://doi.org/10.1038/s44160-026-01176-4

Image Credits: AI Generated

DOI: 10.1038/s44160-026-01176-4

Keywords: silylium ions, halosilanes, halide abstraction, Lewis superacids, substituent scrambling, carborate anions, silicon chemistry, heteroleptic cations, DFT calculations, organosilicon, Nature Synthesis, reactive intermediates