Stable Mg²⁺ Dication Weakly Stabilized/Coordinated in Solution: Synthesis, Structure, Reactivity, and Use in Catalysis

Abstract

The first soluble and stable Mg²⁺ dication stabilized only by weakly coordinating and chemically robust carborate anions [HexCB₁₁Cl₁₁]⁻ is described. Mg[HexCB₁₁Cl₁₁]₂ (1), prepared by reaction of Mg( ⁿ Bu)₂ with 2 equiv of [Ph₃C][HexCB₁₁Cl₁₁], consists, in the solid state, of a central Mg²⁺ surrounded by two [HexCB₁₁Cl₁₁]⁻ anions. In solution, experimental and classical molecular dynamics simulations (cMD) agree with cation/anion association being retained, reflecting the high electrophilicity of the Mg center. Yet, reflecting only weak anion/cation interactions, species 1 polymerizes 1‐hexene and coordinates alkynes. However, 1 displays no reaction with HSiEt₃ at room temperature, consistent with a low hydridicity of the hard (HSAB) Mg²⁺ center. Contrasting with 1 (FIA = 264 kJ mol⁻¹; FIA: Fluoride Ion Affinity), salt Mg[(nBu)₃NB₁₂H₄Cl₇]₂ (2), incorporating the more basic ammoniododecaborate [(nBu)₃NB₁₂H₄Cl₇]⁻ anion, is significantly less Lewis acidic (FIA = 214.7 kJ mol⁻¹) and unreactive with alkenes and alkynes. Salt 1 effectively catalyzes alkene/alkyne hydrosilylation via an initial alkene/alkyne coordination/initiation, as suggested by experimental and computational data. It also efficiently catalyzes (with a catalyst loading down to 0.1 mol%) the hydrosilylation of CO₂ to CH₄ in the presence of HSiEt₃. Salt 1 smoothly promotes the catalytic transfer hydrogenation of 1,1‐diphenylethylene, and it is also an active imine hydrogenation catalyst in the presence of H₂.

Stable Mg²⁺ dication stabilized only by weakly coordinating carborates was first prepared and structurally characterized as a soluble salt in organic solvents. Combined experimental and theoretical data provided valuable insights into its behavior/reactivity in solution. All data agree with a strong Lewis acidic Mg²⁺ center, readily coordinating alkenes and alkynes and effectively catalyzing alkene/alkyne hydrosilylation and imine hydrogenation.

Introduction

Strong Lewis acids and electrophilic species are of crucial importance for the mediation of numerous stoichiometric and catalytic transformations.^{[ 1} , ² , ³ , ⁴ , ^{5 ]} In this area, well‐defined ligand‐supported Mg^II species are gaining importance for use in catalysis due to the potent electrophilicity of the Mg^II center, favoring the activation of various substrates, with earth abundance, low toxicity, and low cost being additional attractive features of magnesium.^{[ 6 ]} Neutral Mg^II species have mostly been shown to catalyze the heteroelementation of various polar/unsaturated substrates, with representative examples ranging from the hydro‐silylation/‐boration of alkenes/alkynes, nitriles, and carbonyl substrates (such as ketones, esters, and CO₂) to dehydrocoupling catalysis.^{[ 7} , ⁸ , ⁹ , ¹⁰ , ^{11 ]} Notably, a couple of examples of Mg‐catalyzed imine hydrogenation catalysis were also recently reported.^{[ 12} , ^{13 ]} Though less explored than their neutral analogues, discrete cationic Mg^II complexes are of particular interest because the cationic charge at the Mg^II center enhances metal electrophilicity and thus substrate activation. Representative examples of ligand‐supported Mg cations for use in catalysis include their application in cyclic esters ring‐opening polymerization,^{[ 14} , ^{15 ]} CO₂ hydro‐silylation/‐boration,^{[ 8} , ^{16 ]} carbonyl hydroboration.^{[ 17 ]} Noticeably, combining a cationic charge and low coordination at Mg^II, as in two‐coordinate cations of the type [(BDI)Mg]⁺ (BDI = β‐diketiminate‐type bidentate ligand), recently reported by Harder,^{[ 18} , ^{19 ]} results in a highly electrophilic Mg center able to form isolable adducts with internal alkynes and various arenes, and catalyze alkyne hydrophosphination.^{[ 20 ]}

Stabilization of Mg^II organocations by various N‐/O‐based ligand platforms as well as NHC ligands, has thus led to stable and soluble electrophilic entities enabling substantial substrate activation. However, ideally, maximal electrophilicity at Mg^II could be achieved with a simple “ligand‐free” Mg²⁺ dication, though such a “naked” dication does not exist in the condensed phase. Toward this goal, Mg²⁺ dication stabilized only by properly designed weakly coordinating anions (WCAs), i.e., as Mg[WCA]₂ salts, may lead to soluble and stable sources of “pseudo‐naked”/weakly coordinated Mg²⁺ dication. Thus far, such Mg salts are unknown and, more generally, there are only rare examples of divalent metal salts M[WCA]₂ exhibiting stability and solubility in organic media, including: i) a Zn[WCA]₂ salt consisting of a central Zn²⁺ dication only weakly stabilized by two [HexCB₁₁Cl₁₁]⁻ carborate anions,^{[ 21 ]} ii) the Sr[Al{OC(CF₃)₃}₄]₂ salt featuring a rather large Sr²⁺ dication partially associated with the alanate anions,^{[ 22 ]} and iii) a very recent Ca[Al{OC(CF₃)₃}₄]₂ salt, though not structurally authenticated and only stable in SO₂ solution.^{[ 23 ]} Taking the example of salt Zn[HexCB₁₁Cl₁₁]₂, it is the strongest Zn‐based Lewis acid in solution, validating the approach of associating a divalent M²⁺ center with halogenated carborates as WCAs for unprecedented electrophilicity. We now report on the synthesis and structural characterization of the first Mg[WCA]₂ salt, incorporating a central Mg²⁺ stabilized by two [HexCB₁₁Cl₁₁]⁻ carborate anions. The stability and solubility of Mg[HexCB₁₁Cl₁₁]₂ also allowed reactivity studies with various polar/unsaturated substrates, all in line with a highly electrophilic Mg^II center. Further, classical molecular dynamic simulations combined with DFT calculations provided critical insights into its solution structure and mechanistic aspects, rationalizing the observed reactivity.

Results and Discussion

Synthesis and Structure

To access Mg[WCA]₂ salts as potential sources of “pseudo‐free” or weakly stabilizing Mg²⁺ dication, the use of the classical [B(C₆F₅)₄]⁻ borate anion as a possible counterion was initially studied through an attempted synthesis of Mg[B(C₆F₅)₄]₂. To this end, Mg( ⁿ Bu)₂ was reacted with 2 equiv of [Ph₃C][B(C₆F₅)₄] (C₆D₅Br, 1 h, RT; eq. 1): both reagents are quantitatively consumed affording an intractable mixture of [Mg‐C₆F₅]‐incorporating species according to ¹H and ¹⁹F NMR data, consistent with anion degradation and indicating that a more chemically robust anion is required.^{[ 24 ]} Next, the much more robust [Al{OC(CF₃)₃}₄]⁻ anion, popularized by Krossing, was investigated as a possible WCA.^{[ 25 ]} The reaction of Mg( ⁿ Bu)₂ with 2 equiv [Ph₃C][Al{OC(CF₃)₃}₄]^{[ 26 ]} (PhBr or 1,2‐difluorobenzene, 1 h, RT; eq. 2) led to the formation of Mg[Al{OC(CF₃)₃}₄]₂ as the major product, according to ¹H, ¹⁹F and ²⁷Al NMR data, along with unidentified side species (see Supporting Information). Yet, in our hands, isolation of pure Mg[Al{OC(CF₃)₃}₄]₂ could not be achieved from the crude mixture despite several attempts, precluding further studies.

$$\begin{array}{ccc}\begin{array}{c}\text{Mg}{{(}^n\text{Bu})}_2\\ +\\ 2[\mathrm{P}{\mathrm{h}}_3\mathrm{C}][\mathrm{B}{({\mathrm{C}}_6{\mathrm{F}}_5)}_4]\end{array}\hfill & \underset{\begin{array}{c}\text{-P}{\mathrm{h}}_3\text{CH}\hfill \\ \text{-butene}\hfill \end{array}}{\overset{\text{RT},1\mathrm{h},\text{PhBr}}{\to }}& \begin{array}{c}mixtureof\\ Mg-C_6{F}_{5}species\end{array}\end{array}$$ (1)

$$\begin{array}{ccc}\begin{array}{c}\text{Mg}{{(}^n\text{Bu})}_2\\ +\\ 2[\mathrm{P}{\mathrm{h}}_3\mathrm{C}][\text{Al}{\{\text{OC}{(\mathrm{C}{\mathrm{F}}_3)}_3\}}_4]\end{array}\hfill & \underset{\begin{array}{c}\text{-P}{\mathrm{h}}_3\text{CH}\hfill \\ \text{-butene}\hfill \end{array}}{\overset{\text{RT},1\mathrm{h},\text{DFB}}{\to }}& \begin{array}{c}\text{Mg}{[\text{Al}{\{\text{OC}{(\mathrm{C}{\mathrm{F}}_3)}_3\}}_4]}_2\\ \begin{array}{c}crude\\ notisolated\end{array}\end{array}\end{array}$$ (2)

$$\begin{array}{ccc}\begin{array}{c}\text{Mg}{(}^n\text{Bu})2\\ +\\ 2[\mathrm{P}{\mathrm{h}}_3\mathrm{C}][\text{HexC}{\mathrm{B}}_{11}\mathrm{C}{\mathrm{l}}_{11}]\end{array}\hfill & \underset{\begin{array}{c}\text{-P}{\mathrm{h}}_3\text{CH}\hfill \\ \text{-butene}\hfill \end{array}}{\overset{\text{RT},1\mathrm{h},\text{DFB}}{\to }}& \begin{array}{c}\text{Mg}[\text{HexC}{\mathrm{B}}_{11}\mathrm{C}{\mathrm{l}}_{11}]2\\ 1\end{array}\end{array}$$ (3)

Equations (1), (2), (3). Synthesis attempts of Mg[WCA]₂ and synthesis of Mg[HexCB₁₁Cl₁₁]₂ (1).

Fortunately, the chemically robust carborate anion [HexCB₁₁Cl₁₁]⁻ proved suitable to access pure Mg[WCA]₂ species.^{[ 24} , ^{27 ]} Thus, the double ionization reaction of Mg( ⁿ Bu)₂ with 2 equiv [Ph₃C][HexCB₁₁Cl₁₁], prepared according to a literature procedure,^{[ 21 ]} proceeds smoothly (DFB, 1 h, RT) to quantitatively produce salt Mg[HexCB₁₁Cl₁₁]₂ (1, eq. 3) along with Ph₃CH and polybutene (due to polymerization of the generated 1‐butene). Salt 1 was isolated in 80% yield as an off‐white solid. It is soluble and stable for days at room temperature in aromatic solvents such as PhBr and DFB, while decomposing within minutes to unknown species in dichloromethane. Alternatively, a salt metathesis reaction between MgI₂ and 2 equiv of Ag[HexCB₁₁Cl₁₁] (MeOH, RT) was also attempted to prepare salt 1. Though the reaction proceeded as expected to afford a MeOH‐stabilized Mg²⁺ dication associated with two non‐coordinated [HexCB₁₁Cl₁₁]⁻ anions, crude 1 could not be further purified (see Supporting Information for additional details).

The molecular structure of the Mg salt 1, depicted in Figure 1, was established through X‐ray diffraction studies. It consists of a central Mg²⁺ dication in an octahedral environment due to the association of the two [HexCB₁₁Cl₁₁]⁻ anions in a κ³‐Cl₃ coordination mode. The Mg–Cl distances in 1 (2.534(1) Å average) are longer than those in the neutral complex [(THF)₄MgCl₂] (2.45 Å average)^{[ 28 ]} and the dinuclear Mg cation [(THF)₃Mg(μ‐Cl)₃Mg(THF)₃]⁺ (2.507 Å average).^{[ 29} , ^{30 ]} Due to the coordination to Mg²⁺, the B–Cl bond distances of the Mg‐coordinated chlorines (1.810 Å average) are lengthened relative to those in free [HexCB₁₁Cl₁₁]⁻ (1.76 Å average). A similar lengthening of the involved B─Cl bonds also occurs upon association of [HexCB₁₁Cl₁₁]⁻ with Zn^II mono‐ and di‐cations.^{[ 21} , ^{31 ]} As anticipated, the interactions between the central Mg²⁺ center and the surrounding chlorine atoms are purely electrostatic, as indicated by DFT‐estimated non‐covalent interactions (NCI, Figure 2).

Figure 1.

Molecular structure of Mg[HexCB₁₁Cl₁₁]₂ (1) with partial atom labeling for clarity. Selected bond distances (Å): Mg1–Cl1 = 2.5031(3), Mg1–Cl2 = 2.5276(2), Mg1–Cl3 = 2.5712(3), B1–Cl1 = 1.812(1), B2‐Cl2 = 1.809(1), B3–Cl3 = 1.811(1).

Figure 2.

Non‐covalent interactions (NCI) for a model species salt 1 as estimated by DFT (B3LYP/6–31 + G**). The central Mg²⁺ dication is in purple and the surrounding chlorine atoms are in green. Attractive electrostatic and dispersive interactions appear in blue and green.

In solution (C₆D₅Br, RT), ¹¹B NMR data of salt 1 agree with an effective coordination of the two [HexCB₁₁Cl₁₁]⁻ anions to the Mg²⁺ dicationic center, with the presence of two resonances (δ −5.3 and −10.8 ppm), contrasting with the ¹¹B NMR spectrum for unassociated/“free” [HexCB₁₁Cl₁₁]⁻ anion (as in [Ph₃C][HexCB₁₁Cl₁₁])^{[ 21 ]} containing three distinct signals (δ −2.9, −10.0 and −11.6 ppm) under identical conditions. Similar NMR data were obtained in DFB, a more polar solvent, also suggesting [HexCB₁₁Cl₁₁]⁻ coordination to Mg²⁺. On the other hand, DOSY NMR measurements for a C₆D₅Br solution of 1 (RT) led to an estimated hydrodynamic volume of 1160 Å³ (see Supporting Information). The latter value lies between the volume of salt 1 from solid state data (approximated to 1460 Å³ following a known methodology)^{[ 32 ]} and that of the carborate anion [HexCB₁₁Cl₁₁]⁻ (720 Å), which could be consistent with salt 1 undergoing a fast cation/anion dissociation on the NMR timescale under the studied conditions. Yet, given the presence of two hexyl chains in 1 (at both ends), which are fluxional groups in solution, the hydrodynamic volume of 1 may significantly differ from that in the solid state, limiting the accuracy/validity of such a comparison.

To gain further insight on the solution structure of the Mg salt 1, classical molecular dynamics (cMD) simulations were performed by modeling the structure/fluxional behavior of a model of 1 (I) within a solvent (PhBr) box (size of 70 x 70 x 70 Å, including roughly 1000 molecules of PhBr) using the Amber 18.0 software (see Supporting Information for further details). The trajectory consisted of a simulation of 2 µs at T = 298.15 K during which the system was let to evolve. A cumulative view of the trajectory is shown in Figure 3, where all snapshots of the structure of I (taken every 10 ns along the trajectory) are superposed, while the trajectory may be visualized (see video, Supporting Information). These data clearly indicate that both [HexCB₁₁Cl₁₁]⁻ anions essentially remain κ³ ‐Cl₃ coordinated to the Mg²⁺ center, with Mg‐Cl distances (2.47(1) Å average) similar to those observed in the solid state for 1 (2.533(1) Å average). Importantly, the cMD‐estimated diffusion coefficient of I under the simulation conditions (D = 2.5 ± 0.7 10⁻¹⁰ m² s⁻¹ at T = 298.15 K) matches well that experimentally estimated from DOSY NMR data for 1 in C₆D₅Br (D = 2.94 ± 0.58 10⁻¹⁰ m² s⁻¹ at T = 298 K), providing experimental support to the proposed cMD approach. Therefore, as observed in the solid state, salt 1 likely consists of associated Mg²⁺ and [HexCB₁₁Cl₁₁]⁻ ions in PhBr at RT, reflecting the strong Lewis acidity of the Mg²⁺ center. Similarly, salt 1 does not presumably dissociate in DFB, given the similar NMR data for 1 in both solvents.

Figure 3.

Cumulative view of the structures of modeled salt 1 (model I) in PhBr (70 × 70 × 70 ų box) along the cMD simulation. The central Mg²⁺ dication is in green. A snapshot was taken every 10 ns along the 2 µs‐long simulated trajectory. The color of the [HexCB₁₁Cl₁₁]⁻ anions evolves as a function of the simulation time according to the color bar represented at the bottom right.

For comparison to the Mg carborate species 1, salt Mg[(nBu)₃NB₁₂H₄Cl₇]₂ (2), incorporating the more coordinating/Lewis basic ammonio‐dodecaborate [(nBu)₃NB₁₂H₄Cl₇]⁻ anion, was also targeted and prepared following an aqueous route previously established for the synthesis of the Zn analogues.^{[ 33 ]} The reaction of MgCl₂ ^.H₂O with 2 equiv of the acid [H(H₂O)_n][(nBu)₃NB₁₂H₄Cl₇], prepared by following a literature procedure,^{[ 33 ]} led to the formation of salt 2 according to all characterization data (see Supporting Information for more details). The molecular structure of the Mg salt 2, as established through X‐ray diffraction studies (Figure S23, Supporting Information), features a central Mg²⁺ di‐cation in a similar octahedral environment to that observed for the carborate salt 1, with the κ³‐Cl₃ coordination of the two [(nBu)₃NB₁₂H₄Cl₇]⁻ anions to Mg²⁺. The Mg–Cl distances in 2 (2.513(2) Å average) are close, though a bit shorter, to those in 1 (2.533(1) Å average). In solution at RT, ¹H and ¹¹B NMR data agree with salt 2 retaining its solid‐state structure in both C₆D₅Br and DFB.

Lewis Acidity Estimation

To assess the Lewis acidity of salts 1 and 2, their fluoride ion affinity was evaluated by DFT (B3LYP/6–31 + G**, PhBr, see Supporting Information).^{[ 34} , ^{35 ]} Consistent with a highly electrophilic Mg^II center, the FIA value for salt 1 (264 kJ mol⁻¹) is significantly higher than that of B(C₆F₅)₃ (220.5 kJ mol⁻¹), reflecting the strong Lewis acidity of 1. Despite Mg²⁺ being a harder Lewis acid than Zn²⁺ (HSAB theory), the FIA of 1 is similar to that of its Zn analogue Zn[HexCB₁₁Cl₁₁]₂ (262.1 kJ mol⁻¹), which is probably due to a stronger anion/cation association of the Mg salt 1. In contrast, the FIA of the ammonio‐dodecaborate Mg salt 2 (214.7 kJ mol⁻¹) is in line with a significantly less electrophilic Mg^II due to stronger coordination of the [(nBu)₃NB₁₂H₄Cl₇]⁻ anions to the metal center.

To probe the potential super Lewis acidic character of salt 1,^{[ 36 ]} its reactivity with the anion [SbF₆]⁻ was studied. Salt 1 immediately and quantitatively reacts with 2 equiv of [(nBu)₄ N][SbF₆] (DFB, RT, 15 min) via a [HexCB₁₁Cl₁₁]⁻/[SbF₆]⁻ anion exchange to afford Mg[SbF₆]₂ and non‐coordinated carborate anions, as deduced from ¹H and ¹¹B NMR data. Remarkably, the resulting mixture is stable under the studied conditions with no observable change after 24 h at RT, suggesting that Mg[SbF₆]₂ is a viable and stable salt in DFB solution and that salt 1 is not super Lewis acidic in DFB. In contrast, performing the same reaction in a 4/1 DFB/C₆D₅Br solvent mixture led after 2 h at room temperature to the formation of a mixture of fluorinated aromatics resulting from C₆D₅Br fluorination along with [SbF₆]⁻ decomposition, according to NMR characterization including ¹H, ²H, ¹⁹F, ¹¹B, and 2D NMR data (see Supporting Information for additional details). Though the formation of a strong fluorinating agent is obvious, the formation of insoluble and not identified solids (possibly of type [Mg‐F] and related aggregates) as the reaction proceeded prevented any further characterization and thus any clear‐cut conclusion. However, in line with the behavior of salt 1 in DFB, the DFT‐estimated FIA of SbF₅ (337.5 kJ mol⁻¹, PhBr) lies well‐above that of salt 1 (264 kJ mol⁻¹, PhBr) suggesting that salt 1 is not a super Lewis acid.

Reactivity with Alkenes and Alkynes

Given its electrophilic nature, the reactivity of salt 1 was first probed for activation of unsaturated substrates such as alkenes and alkynes. As an initial test, the reaction of salt 1 with 1 equiv of 1‐hexene (C₆D₅Br or DFB, 5 min) led to the immediate formation of polyhexene. Further indicative of a highly Lewis acidic Mg^II center in 1, it quantitatively polymerized 100 equiv of 1‐hexene within 20 min (C₆D₅Br, RT) to produce ill‐defined polyhexene according to ¹H NMR and GPC data (see Supporting Information) with salt 1 retaining its integrity; a reaction probably proceeding through a Lewis acid‐assisted cationic polymerization. For comparison, salt 1 is more reactive than its Zn analogue toward 1‐hexene, which under identical conditions, converted 60% of 1‐hexene to polymer after 3 h at RT. No reaction between the less Lewis acidic Mg salt 2 and 1‐hexene (100 equiv) was observed after 24 h (C₆D₅Br, RT). Contrasting with its reactivity toward 1‐hexene, salt 1 showed no apparent reaction with cyclohexene (1 or 4 equiv, DFB) after 16 h at room temperature, though polymerization to polycyclohexene occurs at 80 °C (complete cyclohexene conversion within 1 h of a 1/4 1/cyclohexene mixture). From a 1/4 1/cyclohexene mixture, no evidence of cyclohexene coordination to the Mg cationic center could be detected by low‐temperature ¹H and ¹¹B NMR analysis (DFB, −30 °C). To promote alkene coordination to 1, possibly through chelate formation, the bis‐alkene 1,5‐cyclooctadiene (1,5‐COD) substrate was thus studied next, with no observable reaction at room temperature of a 1/2 1/1,5‐COD solution in DFB from NMR data. However, low temperature ¹¹B NMR data (DFB, −30 °C) of the mixture indicated a splitting of the major ¹¹B NMR resonance (δ −10.7 ppm for salt 1, DFB, −30 °C) to two resonances (δ −9.9 and −10.5 ppm), in line with a weaker cation/anion association upon addition of 1,5‐COD. For comparison, the non‐coordinated/“free” [HexCB₁₁Cl₁₁]⁻ anion also features two such resonances but with a larger chemical shift difference (δ −9.9 and −11.8 ppm; see Figures S44 and S45). These data suggest a labile association/coordination of 1,5‐COD to the Mg center under the studied conditions.

The reaction of salt 1 with alkynes such as 1‐octyne, 2‐hexyne, and 2,6‐octadiyne was also evaluated for coordination to the Mg center. The ¹¹B NMR spectrum of a 1/6 1/1‐octyne mixture (DFB, RT, 5 min) displays a significant broadening of the anion ¹¹B NMR signals, indicative of a lower cation/anion association, suggesting 1‐octyne coordination to the Mg²⁺ center. Similarly, the reaction of 1 with 1 equiv of 2‐hexyne provoked an immediate broadening of the ¹¹B NMR major peak (see Supporting Information). In addition, low temperature ¹H and ¹¹B NMR data of a 1/4 1/2‐hexyne mixture (DFB, −30 °C) agree with the formation of labile alkyne^…Mg adducts under the studied conditions, with the ¹¹B NMR signals in line with a lower cation/anion association as well as downfield shifted 2‐hexyne resonances (Δδ = 0.06 ppm average vs. free 2‐hexyne; see Figures S50 and S51). To further promote/stabilize alkyne coordination, bis‐alkyne 2,6‐octadiyne was reacted with salt 1. The reaction of salt 1 with 2 equiv 2,6‐octadiyne (DFB, RT, 15 min) led to an immediate broadening of the ¹¹B NMR resonances, suggesting a dynamic process. At −30 °C (Figures S52 and S53), the ¹¹B NMR spectrum is similar to that of a free or weakly interacting [HexCB₁₁Cl₁₁]⁻ anion, clearly indicating alkyne coordination to the Mg²⁺ dication. Although the formation of a [bis(bis‐alkyne)Mg]²⁺ dication seems possible, the rather broad ¹H NMR signals suggest labile coordination/chelation even under the studied conditions. To date, characterized alkyne‐Mg cationic adducts are rare, and only mono‐alkyne adducts are known as described by Harder and coworkers by coordination to a [(BDI)Mg]⁺ cation (BDI = β‐diketiminate).^{[ 19 ]} In the present case, despite their apparent stability over time (24 h, RT, PhBr or DFB), such cationic adducts could not be isolated and further characterized, possibly due to their labile nature.

To gain additional insight/support on alkene and alkyne reactivity with salt 1, DFT and cMD calculations were performed. Taking propene as model, the DFT‐computed Gibbs free energy for the reaction of I, model of salt 1, with a propene molecule to afford adduct [(HexCB₁₁Cl₁₁)Mg(propene)]⁺ along with a fully dissociated [HexCB₁₁Cl₁₁]⁻ anion is quite endergonic (ΔG = 61.5 kJ mol⁻¹, PhBr), inconsistent with the observed fast reaction of 1 with 1‐hexene.

On the other hand, cMD simulations gave a better understanding of salt 1 dynamic behavior and reactivity in PhBr solution. cMD simulation of I (PhBr box, 7.5 µs long simulation, RT) in the presence of 100 propene molecules indicates the ready coordination of propene to the Mg center with the formation of bis‐ and tris‐propene cationic adducts as major species (Figure 4; Figure S68 and video in the ESI). Importantly, as deduced from potential mean force (PMF) calculations (Table S7, ESI), the coordination of one propene to the Mg cationic center of I is thermodynamically feasible under the studied conditions (ΔG = 2.1 kJ mol⁻¹) and proceeds with a low energy barrier (26.4 kJ mol⁻¹), consistent with a fast and labile alkene coordination to the Mg center. Considering all tris‐propene Mg adducts [Mg(η ²‐propene)₃][HexCB₁₁Cl₁₁]₂ computed along the cMD trajectory, they consist of three dissymmetrically Mg‐bound propene ligands with Mg–C1_propene distances (2.12(1) Å average) significantly shorter than the Mg–C2_propene (2.68(1) Å average), with a representative example of such [Mg(η ²‐propene)₃][HexCB₁₁Cl₁₁]₂ adduct being depicted in Figure 4. The calculated (average) Mg–C1_propene distances in tris‐propene adducts are similar to the Mg–alkyl σ‐bond length,^{[ 37 ]} supporting substantial alkene activation upon coordination to the Mg^II center, and thus susceptible to undergo cationic polymerization as observed with 1‐hexene in the presence of 1. For comparison, a dissymmetrical alkene bonding/contact to Mg^II was also observed in a structurally characterized neutral Mg(η ²‐alkene)₂ dibenzoazepinate complex, albeit with longer and less dissymmetrical Mg–C contacts (2.740(3) and 2.848(3) Å average for the shorter and longer distances, respectively).^{[ 38 ]} Apart from propene activation by the Mg²⁺ cation, it is also noteworthy that the two [HexCB₁₁Cl₁₁]⁻ anions essentially remain η ¹‐coordinated, each through a chlorine atom, to the Mg^II center (Mg^…Cl distance = 2.69(1) Å average), thus readily adapting their coordination mode (from κ ³‐Cl₃ to η ¹‐Cl) in solution upon incoming alkene substrates.

Figure 4.

Molecular structure of model [Mg(propene)₃][HexCB₁₁Cl₁₁]₂ as calculated from the cMD trajectory (7.5 µs long) for a mixture of I and 100 propene molecules in PhBr (70 × 70 × 70 Å³ box). Snapshot taken along the trajectory at t = 7051 ns. Labels are omitted for clarity. The central Mg²⁺ dication is in yellow and the surrounding carbon and chlorine atoms are in black and green, respectively.

Similarly, cMD calculations of model I in the presence of 100 propyne molecules (in a PhBr box) also suggest the fast and labile coordination of propyne to the Mg^II center at room temperature, with the fast formation of bis‐, tris‐, and tetra‐propyne Mg^II adducts that readily interconvert through fast propyne coordination/decoordination (see Supporting Information for a video and Figure 5 for a representative cMD‐computed bis‐propyne Mg adduct). This is further confirmed by PMF calculations indicating that coordination of a propyne molecule to I is slightly exergonic (ΔG = −1.3 kJ mol⁻¹) with a low energy barrier (16.8 kJ mol⁻¹), in line with a labile propyne coordination. In the modeled bis‐propyne adducts, the alkyne molecules also coordinate dissymmetrically to Mg²⁺ with Mg^…C1_propyne distances (2.26(1) Å average) shorter than Mg–C2_propyne distances (2.67(1) Å average). The Mg–alkyne contact/bond lengths in such adducts are comparable to those in a mono‐cationic alkyne adduct, [(BDI)Mg(2‐butyne)]⁺, recently characterized.^{[ 19 ]}

Figure 5.

Molecular structure of a representative [Mg(propyne)₂][HexCB₁₁Cl₁₁]₂ model as calculated from the cMD trajectory (2 µs long) for a mixture of I and 100 propyne molecules in PhBr (70×70×70 ų box). Snapshot taken along the trajectory at t = 354 ns. Labels are omitted for clarity. The central Mg²⁺ dication is in yellow, and the surrounding carbon and chlorine atoms are in black and green, respectively.

Reactivity with HSiEt₃

The activation of the Si─H bond is well‐established for a number of strong Lewis acids, most notably for B(C₆F₅)₃ and Al(C₆F₅)₃,^{[ 39} , ⁴⁰ , ^{41 ]} and is typically considered as the key initial step for Lewis‐acid‐mediated hydrosilylation of various unsaturated substrates. Probing the reactivity of Lewis acidic salt 1 with HSiEt₃ as an Si–H source (1/100 1/HSiEt₃ mixture, C₆D₅Br, RT) led to no apparent reaction after 24 h according to ¹H and ²⁹Si{¹H} NMR data, contrasting with the rather fast HSiEt₃ redistribution observed with the Zn counterpart Zn[HexCB₁₁Cl₁₁]₂ under identical conditions.^{[ 21 ]} Consistent with the latter, the DFT‐estimated hydride ion affinity (HIA, PhBr) of salt 1 (HIA = 370 kJ mol⁻¹) is significantly lower than that of Zn[HexCB₁₁Cl₁₁]₂ (HIA = 496 kJ mol⁻¹), clearly indicating that salt 1 is a harder Lewis acid than its Zn analogue.^{[ 36 ]} Nevertheless, heating a 1/100 1/HSiEt₃ mixture (C₆D₅Br) at 90 °C promoted silane redistribution with a 40% HSiEt₃ conversion to 1/1 H₂SiEt₂/SiEt₄ mixture after 24 h as deduced from NMR data (Scheme 1).

Scheme 1.

Reaction of the Mg salt 1 with HSiEt_3.

As DFT‐estimated, the reaction of model I (model of 1) with one HSiEt₃ molecule to mono‐cation [(HexCB₁₁Cl₁₁)Mg(HSiEt₃)]⁺ (and free [HexCB₁₁Cl₁₁]⁻) is clearly endergonic (ΔG = 77.7 kJ mol⁻¹). Likewise, the reaction of such Mg^…HSiEt₃ adduct with an additional HSiEt₃ to yield [(HexCB₁₁Cl₁₁)MgH] and [Et₃Si–H–SiEt₃]⁺ is disfavored thermodynamically (ΔG = 69.7 kJ mol⁻¹). Yet, upon forcing conditions, the formation of trace amounts [Et₃Si–H–SiEt₃]⁺ seems possible, and would account for the observed HSiEt₃ redistribution reaction slowly proceeding at 90 °C.

cMD simulations of model I in the presence of 100 HSiEt₃ molecules (in a PhBr box) further indicate the low reactivity of HSiEt₃ with 1. As illustrated in Figure 6, the calculations agree with little to no Mg²⁺/HSiEt₃ interactions under simulation conditions (PhBr, RT) with most HSiEt₃ molecules being outer‐sphere, as indicated by long Si─H^…[Mg]²⁺ distances all above 3.5 Å (see also a video in the ESI). Accordingly, unlike propene and propyne, PMF‐estimated HSiEt₃ coordination to the Mg center in I is quite endergonic (39.8 kJ mol⁻¹) while no energy barrier could be computed.

Figure 6.

Cumulative view of HSiEt₃ molecules interactions with salt I, with distances between the Si‐H hydrogen and the Mg²⁺ center within a) 3.5, b) 4.0, and c) 4.5 Å, as computed from a cMD simulation of model I in the presence of 100 HSiEt₃ modelled molecules (70 × 70 × 70 Å³ box of PhBr molecules, 2 µs trajectory, room temperature.) HSiEt₃ are color‐coded as a function of the time step of the trajectory. The corresponding color bar (in µs) is given in the bottom right corner. The central Mg²⁺ dication is in green.

Alkene and Alkyne Hydrosilylation Catalysis

Mg‐catalyzed alkene and alkyne hydrosilylation catalysis has thus far been little investigated, and essentially includes contributions from the groups of Parkin and Hill, showing that well‐defined Mg–H species may mediate alkene/alkyne hydrosilylation catalysis through an insertion/σ‐bond metathesis sequence, albeit with low catalytic activity.^{[ 8} , ⁴² , ^{43 ]} Alternatively, in principle, Mg‐mediated alkene/alkyne hydrosilylation should also be achievable through Lewis acid‐promoted catalysis with a proper Mg‐based Lewis acid such as salt 1. Satisfyingly, the reaction of a 1/1 1‐hexene/HSiEt₃ mixture with salt 1 as catalyst (1 mol% of 1, C₆D₅Br) led to the immediate and quantitative formation of the anti‐Markovnikov hydrosilylation product 3 within 10 min at room temperature (Scheme 2), with catalyst 1 retaining its integrity thus agreeing with a Lewis acid type mechanism. For comparison, 1‐hexene hydrosilylation catalyzed by Mg–H complexes of the type [(BDI)Mg–H]₂ (BDI = β‐diketiminate) requires days for reaction completion at higher catalyst loading and reaction temperature.^{[ 42 ]}

Scheme 2.

1‐hexene hydrosilylation catalyzed by the Mg salt 1.

Likewise, the Mg salt 1 also effectively catalyzes alkyne hydrosilylation, with the reaction of a 1/1 1‐octyne/HSiEt₃ mixture in the presence of 1 (1 mol%) affording hydrosilylation products 4–6 in a 3/1/1 4/5/6 ratio (Scheme 3, 80% overall NMR yield) after 40 min at room temperature according to ¹H NMR and GC‐MS data analysis. Given the experimental and cMD data on the relative reactivity of alkenes, alkynes and HSiEt₃ toward 1 (vide supra), the present hydrosilylation catalysis proceeds via an initial alkene/alkyne activation through coordination of the Mg^II center, rather than an initial Si–H activation as typically observed for most main group Lewis acid catalysts.^{[ 40} , ⁴¹ , ^{44 ]} For instance and comparison, 1‐hexene hydrosilylation catalyzed by Zn[HexCB₁₁Cl₁₁]₂ first involves HSiEt₃ ionization to form [Et₃Si–H–SiEt₃]⁺ that then acts as the hydrosilylation catalyst.^{[ 21 ]}

Scheme 3.

1‐octyne hydrosilylation catalyzed by the Mg salt 1.

Additional mechanistic rationale and support were provided through cMD and DFT calculations (ωB97XD/6–31 + G**, PhBr as solvent through a PCM model; see Supporting Information for details) using modeled propene and propyne molecules for simplicity. Prior to DFT calculations and to obtain realistic starting models, cMD simulations were first conducted with a 100/100/1 propene (or propyne)/HSiEt₃/I mixture during a 2 µs trajectory (PhBr box, RT), from which representative models were extracted and used as starting points for DFT‐computed mechanisms (for their molecular structures, see Figures S73, S78 and S82; ESI). As depicted in Figure 7 for propene hydrosilylation, cMD‐estimated model A, consisting of a tris‐propene Mg adduct (along with two carborate anions) in the vicinity of a HSiEt₃ molecule (Figure S73), may readily react through a Si–H hydride transfer to the C _β carbon of a Mg‐coordinated propene moiety via transition state TS1 (ΔG = 72.0 kJ mol⁻¹) to yield the rather high energy intermediate B (ΔG = 71.1 kJ mol⁻¹). The latter is only a fleeting intermediate since it consists of a newly formed Mg‐propyl function that only weakly stabilizes a [SiEt₃]⁺ cation through C _β –H interaction. It then reacts intramolecularly through a [SiEt₃]⁺ electrophilic attack of the Mg‐propyl group (TS2, ΔG = 107.5 kJ mol⁻¹) to afford hydrosilylation product C (ΔG = −36.0 kJ mol⁻¹) and re‐generate the Mg²⁺ dication.

Figure 7.

Left: DFT‐estimated (ωB97XD/6–31 + G**) for the hydrosilylation of propene catalyzed by I, a model of Mg salt 1. The molecular structure of the starting model A is depicted in Figure S73, as computed from cMD simulations. Right: molecular structure of TS1 with the central Mg²⁺ dication in yellow while the surrounding carbon and chlorine atoms are in black and green, respectively.

Regarding propyne hydrosilylation, a one‐step mechanism could be DFT‐computed to account for the formation of the major product, 4 (Scheme 3), and is depicted in Figure 8. Thus, from the cMD‐simulated model D, which incorporates a bis‐propyne Mg adduct (along with two carborate anions) with an HSiEt₃ molecule in close proximity (Figure S78), the formation of the anti‐hydrosilylation product may directly occur through a low‐energy transition state TS3 (ΔG = 46.4 kJ mol⁻¹) involving a Si–H transfer to a Mg‐activated propyne (at C _β ) having an apparent vinylic character (Figure 8, right). TS3 then directly collapses to the exclusive formation of the Z‐isomer hydrosilylation product (E, ΔG = −115.1 kJ mol⁻¹).

Figure 8.

Left: DFT‐estimated (ωB97XD/6–31 + G**) for the hydrosilylation of propyne catalyzed by model Mg salt I to form the major product E. The molecular structure of the starting model D is depicted in Figure S78, as computed from cMD simulations. Right: molecular structure of TS3 with the central Mg²⁺ dication in yellow while the surrounding carbon and chlorine atoms are in black and green, respectively.

Interestingly, the nature of the propyne Mg adduct undergoing hydrosilylation influences product selectivity according to calculations. Thus, computation of a propyne hydrosilylation pathway starting from a cMD‐simulated model incorporating a tetra‐propyne Mg adduct (Figure S82), which is significantly more hindered than the bis‐propyne adduct analogue, led the formation of E as well as the α‐isomer (Me)(SiEt₃)C═CH₂ (model H, Figure S81), thus rationalizing the formation of minor product 6 in 1‐octyne hydrosilylation catalyzed by 1. The formation of the minor E‐isomer (simplified model of 5) could not be DFT‐rationalized with a Mg‐catalyzed process. However, trace amounts of [SiEt₃]⁺ may form as the catalysis proceeds, which could account for the formation of 5.^{[ 45 ]} In this regard, an independent 1‐octyne hydrosilylation catalysis run using in situ generated [SiEt₃][HexCB₁₁Cl₁₁] as catalyst (1 mol%) led to a quantitative 1‐octyne consumption within 20 min (C₆D₅Br, RT, ¹H NMR monitoring), with the predominant formation of polymeric material (not further characterized) along with 10%–15% formation of hydrosilylation products 4, 5 and 6 in a roughly 1/1/1 ratio.

Carbonyl and CO₂ Hydrosilylation by Species 1

The potent Lewis acidity of Mg salt 1 also prompted its evaluation in carbonyl hydrosilylation catalysis, given the known activity of some strong Lewis acids, such as B(C₆F₅)₃ and various main group metal organocations,^{[ 40} , ⁴⁶ , ^{47 ]} for efficient hydrosilylation catalysis under mild conditions. Yet, no reaction was observed upon combining a 1/1 benzophenone/HSiEt₃ mixture with salt 1 (1mol) after 24 h at RT (and then 24 h at 90 °C), reflecting the strong oxophilicity/low hydridicity of the Mg^II center. ¹H and ¹¹B NMR monitoring of the mixture (C₆D₅Br) indicated the immediate formation of a benzophenone di‐cationic adduct of the type [Mg(O═CPh₂)_x]²⁺ (x = 4 to 6) along with free carborate anions. Under the studied conditions, benzophenone coordination apparently precludes any Mg^II…HSiEt₃ interaction even upon heating, and thus silane activation that would permit/initiate catalysis.^{[ 40 ]} An identical outcome was observed with benzaldehyde as a carbonyl source. In comparison, under identical conditions, the softer Lewis acid Zn[HexCB₁₁Cl₁₁]₂ is an efficient benzophenone hydrosilylation catalyst at room temperature (TOF up to 200 h⁻¹).^{[ 21 ]}

In contrast, the Mg salt 1 was found to be more reactive with the less Lewis basic CO₂ under hydrosilylation conditions, though CO₂ is typically a more challenging C═O substrate to reduce.^{[ 48 ]} Thus, species 1 (1 mol% vs. HSiEt₃) efficiently catalyzes CO₂ hydrosilylation (1.5 atm, CO₂, DFB, 90 °C) to the quantitative formation of methane along with siloxanes by‐products, according to ¹H and ¹³C NMR data. The present Mg‐mediated CO₂ hydrosilylation catalysis,^{[ 8} , ¹⁶ , ^{49 ]} is roughly twice as fast as that with Zn[HexCB₁₁Cl₁₁]₂ as catalyst and even proceeded effectively at lower catalyst loading (0.1 mol%) with complete HSiEt₃ conversion to methane in the presence of CO₂ (1.5 atm) within 4 h at 90 °C (Scheme 4). The present catalysis probably proceeds through a Lewis acid‐type activation similar to Al‐ and Zn‐based Lewis acidic catalysts.^{[ 46} , ⁵⁰ , ^{51 ]}

Scheme 4.

CO₂ hydrosilylation catalyzed by Mg salt 1.

Hydrogenation Catalysis with Mg Salt 1

Catalytic hydrogenation of unsaturated substrates mediated by Mg^II metal complexes remains rare, and notably includes the use of Mg amido catalysts for imine hydrogenation catalysis^{[ 12 ]} and Mg–H pincer complexes for the catalytic semi‐hydrogenation of alkynes and alkene hydrogenation.^{[ 52 ]} To probe the potential of Mg salt 1 for hydrogenation, transfer hydrogenation catalysis of 1,1‐diphenylethylene was first tested using 1,4‐cyclohexadiene (CHD) as the hydrogen source. Thus, salt 1 efficiently catalyzes such transformation with 80% conversion of 1,1′diphenylethylene to hydrogenation product Ph₂CH–Me (70% NMR yield, Scheme 5) along with a dimer of 1,1′‐diphenylethylene (2,3‐dihydro‐1‐methyl‐1, 3,3‐triphenyl‐indene) as a minor product, within 30 min at room temperature.^{[ 53 ]}

Scheme 5.

Transfer hydrogenation catalysis mediated by salt 1.

Species 1 was next evaluated for imine hydrogenation using molecular hydrogen. Though no reaction was observed at room temperature, salt 1 (10 mol%) satisfyingly catalyzes hydrogenation of (Ph)(H)C═N(tBu) at 100 °C (PhBr) in the presence of H₂ (6 bar) with the quantitative formation of the corresponding amine product within 16 h (not optimized, Scheme 6). In contrast, salt 1 is inactive under hydrosilylation conditions, with no reaction between a 1/1 (Ph)(H)C═N(tBu)/HSiEt₃ mixture and salt 1 (10 mol%) after 24 h at 100 °C (PhBr). The higher reactivity toward H₂ versus HSiEt₃ suggests a hydrogenation catalysis going through a FLP‐type operative mechanism.^{[ 54} , ^{55 ]} With 1/(Ph)(H)C═N(tBu) acting as the Lewis pair, an initial H₂ cleavage to a transient Mg–H of the type [(HexCB₁₁Cl₁₁)Mg–H][HexCB₁₁Cl₁₁] along with iminium [(Ph)(H)C═NH(tBu)]⁺ likely occurs with a subsequent Mg–H transfer to produce the amine product and re‐generate catalyst 1.

Scheme 6.

Imine hydrogenation catalyzed by salt 1.

Conclusion

Stable and soluble Mg²⁺ stabilized only by weakly coordinating [HexCB₁₁Cl₁₁]⁻ carborate anions was shown to be accessible. All experimental and cMD data on salt 1 are consistent with anion coordination being retained in solution, reflecting the strong Lewis acidity of the central [Mg]²⁺ dication. Nevertheless, in line with rather weak Mg^…carborate interactions, unsaturated substrates such as alkene/alkynes readily coordinate the [Mg]²⁺, with a terminal alkene like 1‐hexene being immediately polymerized upon coordination. In contrast and reflecting to the low hydridicity of [Mg]²⁺, salt 1 only exhibits low reactivity with hydrosilanes such as HSiEt₃. Overall, salt 1 stands as a strong Mg‐based Lewis acid and, as expected, is harder (HSAB) than Zn[HexCB₁₁Cl₁₁]₂.

Salt 1 was also exploited in hydrosilylation and hydrogenation catalysis. It is a highly effective catalyst for alkene/alkyne hydrosilylation under mild conditions, thus the first efficient Mg‐based alkene/alkyne hydrosilylation catalyst. All data strongly suggest a catalyst proceeding through an alkene/alkyne coordination/initiation, rather than initial silane activation. In contrast, and again reflecting the hardness of the Mg²⁺, salt 1 showed no hydrosilylation catalysis activity with imines and ketones, but it is effective in the catalytic hydrosilylation of CO₂, a less coordinative and Lewis basic substrate, with catalyst loading down to 0.1 mol%. Finally, salt 1 is also able to undergo FLP‐type catalysis, as illustrated in the case of imine hydrogenation with molecular hydrogen.

From a methodological point of view, the present approach combining experimental characterization, cMD simulations, along with DFT calculations revealed crucial to understand the structure/reactivity trends of salt 1 in solution and should be useful for future developments and reactivity understanding of novel “pseudo‐free” or weakly coordinated metal cations in the condensed phase.

Author Contributions

X.X., S.T., S.M.: experimental work. A.C.: cMD simulations and analysis. C.G.: DFT calculations and analysis. B.J.: supervision. R.W., S.D.: supervision, conceptualization, funding acquisition, data curation, formal analysis, writing—review & editing.

Supporting Information

NMR spectra, crystallographic details, DFT calculation details. CCDC 2431141 and CCDC 2431142 contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe (http://www.ccdc.cam.ac.uk/structures).

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Xu X., Chaumont A., Gourlaouen C., Tongdee S., Munshi S., Jacques B., Wehmschulte R., Dagorne S., Angew. Chem. Int. Ed. 2025, 64, e202506266. 10.1002/anie.202506266

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.