Introduction

Carbenes and their heavier group 14 analogues are at the forefront of modern main group chemistry because their unique electron configuration, featuring a vacant p-orbital and a lone pair of electrons, allows them to act as both electron donors and electron acceptors (Fig. 1B)1,2,3,4,5,6,7,8,9,10,11. These amphiphilic species have the potential to mimic the behavior of transition metals, thus offering an effective alternative to transition metal catalysis12. In the past two decades, research on the carbene analogues has primarily focused on their synthesis and reactivity5,6,7,8,9,10,11. However, their application in redox catalysis remains elusive. Unlike transition metals that readily engage in redox catalysis by shuttling between different oxidation states (Fig. 1A), carbene analogues face challenges in the subsequent reductive elimination from the high valence state after oxidative addition, thereby impeding the completion of the catalytic cycle (Fig. 1B)13,14,15,16,17. Despite significant achievements in bridging the gap between bond activation and catalysis, achieving simple redox catalysis using carbene analogues as catalysts seems still within sight but beyond reach.

Fig. 1: Redox catalysis with transition metal complexes or carbene analogues.
figure 1

A Well-studied transition-metal mediated redox catalysis. B TM-behavior of heavier group 14 carbene analogues and their potential application in redox catalysis. C Carbodiphosphorane ligated stannylene and its application as redox catalyst in the hydrodefluorination of fluoroarenes. TM transition metal, OA oxidative addition, LM ligand metathesis, RE reductive elimination.

Full size image

Over the last decade, it has been recognized that geometric perturbation serves as an effective strategy for fine-tuning the electronic configuration of the main group elements18,19. This approach has unlocked the unique reactivity and catalytic potential in main group compounds. Recent studies indicate that the utilization of multidentate pincer ligands leads to the formation of the deformed pnictogen compounds, thereby imparting them with characteristics of transition metals18,19,20,21,22. These geometrically constrained pnictogen systems have demonstrated their ability to act as catalysts in redox catalysis via PIII/PV or BiI/BiIII redox cycles20,21. Remarkably, these catalytic systems involving group 15 elements exhibit all three fundamental steps consisting of oxidative addition (OA), ligand metathesis (LM) and reductive elimination (RE), which are pivotal to the transition metal catalytic mechanisms (Fig. 1A). The impressive advancements in geometrically constrained organopnictogen redox catalysis have refined the framework of main group catalysis. Cognizant of this, we propose that the ligand-enforcement strategy can endow heavier group 14 systems with distinctive properties, thereby facilitating the development of group 14 redox catalysts.

Herein, we report a stannylene 1 with nucleophilicity supported by the rigid carbodiphosphorane ligand (Fig. 1C). Reactivity studies demonstrate that stannylene 1 exhibits capability in activating σ-bonds via oxidative addition, while the corresponding adduct undergoes ligand metathesis in the presence of an additive. More importantly, stannylene 1 presents catalytic activity for facilitating the hydrodefluorination of fluoroarenes via C–F oxidative addition, F/H ligand metathesis, and C–H reductive elimination processes, akin to the classical transition metal catalysis. This represents an example of metallomimetic catalysis involving a SnII/SnIV redox cycle.

Results and discussion

Synthesis and characterization of the stannylene

Our study commenced with the preparation of carbodiphosphorane ligated stannylene 1. The salt metathesis reaction between dilithio-carbodiphosphorane (Li2CDP)23,24,25,26,27 and stannous chloride in tetrahydrofuran (THF) at −30 °C resulted in the CDP stannylene (CDPSn) 1 as main product, which could be isolated in 71% yield (Fig. 2A). The identity of CDPSn 1 was initially demonstrated by nuclear magnetic resonance (NMR) spectroscopy. The product presents a low field 31P NMR resonance (δ31P = 24.1 ppm) compared to what is found in the starting material Li2CDP (δ31P = 17.0 ppm) (Table S5)23. A significantly high field signal in the 119Sn NMR spectrum at 135.9 ppm is observed for the central tin atom verifying the product formation. Comparison with 119Sn NMR chemical shifts of the general dicoordinated stannylene (δ119Sn = ca. 2000–3000 ppm) indicates the coordination number of the central Sn atom in 1 is larger than two28,29,30,31,32.

Fig. 2: The synthesis and computational analysis of stannylene 1.
figure 2

A Synthetic approach of stannylene 1. B Molecular structure of 1 in the solid state (Hydrogen atoms and solvent molecules are omitted for clarity. Thermal ellipsoids are at the 50% probability level). C Selected frontier molecular orbital of 1. D Electron localization function analysis of C1–Sn–C2 plane in 1. E The contour plot of the Laplacian of electron density (∇2ρ(rc)) in the C1–Sn–C2 plane of 1. BCP: bond critical point; ρ(rc): electron density; ∇2ρ(rc): Laplacian of electron density; H(rc): total electron energy density; ε(rc): ellipticity of electron density. All computational analysis was calculated at the BP86-D3(BJ)/def2-TZVP level.

Full size image

Yellow crystals of 1, suitable for single crystal X-ray diffraction (sc-XRD) analysis, were obtained by slow diffusion of n-hexane onto a concentrated toluene solution of 1 at room temperature. The crystallographic analyses show that the tin atom adopts a tricoordinated environment with the sum of the bond angles around the tin atom Σα(Sn) = 269.0°, implying the presence of a lone pair at the tin atom (Fig. 2B). The two five-membered Sn-heterocycles form a butterfly confirmation with a dihedral angle of θ  = 71°. Attributable to the rigidity of the trident pincer-type CDP ligand, stannylene 1 is forced into a Cs type geometry, which is distinct from the reported carbodiphosphorane-GeCl2 complex ((Ph3P)2C–GeCl2) with a regular C3V geometry33. The inner-cyclic Sn–C1 bond length (2.278(2) Å) is somewhat longer than the Sn–C2 and Sn–C3 bonds (2.254(2) Å, 2.219(2) Å) and those in diaryl/dialkyl stannylene (2.20–2.27 Å)28,29,30,31. However, it is shorter than the CNHC → Sn dative bond in the N-heterocyclic carbene (NHC) stabilized stannylene (Sn–CNHC: 2.29–2.33 Å) (Table S4)34,35. These results confirm the expected bonding situation of σ-donation of the C1 atom into the vacant 5p-orbital of the central tin atom. Overall, the Sn–C1 bond in 1 may be described with a resonance between dative bond and electronsharing σ-bond (Fig. 2A).

To further elucidate the electronic situation of CDPSn 1, density functional theory (DFT) calculations were performed at the BP86-D3(BJ)/def2-TZVP level36. The highest occupied molecular orbital (HOMO) in 1 consists of a lone pair (sp0.27 hybrid), indicating its nucleophilic character, which is consistent with the results of sc-XRD analysis (Fig. 2C). Electron localization function (ELF) calculations were carried out for 1 (Fig. 2D)37. The presence of the red region at the bottom-right of the Sn atom again validates the presence of a lone pair on Sn atom. Furthermore, we performed the quantum theory of atoms in molecules (QTAIM) analysis to gain further details for the bonding nature between Sn atom and C1 atom in 138. The contour plot of the Laplacian of electron density (∇2ρ(r)) in the C1–Sn–C2 plane is shown in Fig. 2E. The results of QTAIM analysis disclose that the Sn–C1 bond in 1 can be classified as typical donor–acceptor bond with its bond critical point (BCP) located proximally to a nodal surface of the Laplacian ∇2ρ(r) and a negative value for the total energy density H(rc). For standard single bond, the ellipticity value (ε(rc)) should be zero due to the cylindrical contour of electron density. In the case of 1, the ellipticity value (ε(rc) = 0.03) for Sn–C1 bond is close to zero, pointing out the absence of donation of π lone pair of C1 atom to Sn atom. These observations were further supported by natural bond orbital (NBO) analysis39, where the Sn–C1 σ-bond and the π-lone pair on C1 atom were identified and the Wiberg bond index (WBI)40 value of 0.51 for Sn–C1 bond fell into the typical range of C→Sn dative bond (Fig. S90)34,35. On the basis of these computational findings, two major resonance structures of 1 are described in Fig. 2A, consistent with the results of the structural and spectroscopic analyses.

Reactivity studies of the stannylene

Stannylenes have been extensively studied in transition metal coordination and bond activation chemistry5,6,7,8,9,10,11,14,15,16,17. To examine the reactivity of stannylene 1, we next investigated its reactions with Fe2(CO)9, methyl iodide and pentafluoropyridine (Fig. 3). 1 reacted smoothly with Fe2(CO)9 in THF at room temperature to give complex 2 in 81% yield, which was fully characterized by NMR, sc-XRD analysis and infrared (IR) spectroscopy. The shielded 119Sn resonance (342.7 ppm) is observed in 119Sn NMR spectrum of iron complex 2 in comparison with that in 1 (135.9 ppm) (Table S5). The Sn–Fe bond (2.530(7) Å) in 2 is characteristic for a typical Sn–Fe single bond (2.50–2.65 Å)32. The infrared bands of the CO groups in 2 are important parameters for evaluating the donor ability of stannylene 141. The A1 CO stretching frequency at νCO = 2004 cm−1 of 2 is smaller than the isostructural Fe(CO)4 complexes of tBu3P (νCO = 2045 cm−1), and N-heterocyclic carebenes (NHC) (νCO = 2035–2037 cm−1). This demonstrates the σ-donating property of CDPSn 1 is stronger than the tBu3P and NHC ligands. The significantly strong donor property was re-corroborated by calculating the νCO of the model [CpIr(CO)L] complex described by Gusev (see Supplementary Information, Table S7)42.

Fig. 3: Reactivity studies of 1.
figure 3

A Reaction of 1 with Fe2(CO)9 gave complex 2. Oxidative addition reaction of 1 with CH3I and pentafluoropyridine afforded adducts 3 and 4. Compound 5 was formed via ligand metathesis of 4 with LiCl, or via a one-pot reaction by combining 1, pentafluoropyridine and LiCl. B Molecular structures of 2–5 in the solid state (Hydrogen atoms and solvent molecules are omitted for clarity. Thermal ellipsoids are at the 50% probability level).

Full size image

Studies in σ-bond activation were subsequently investigated by the reaction of 1 with CH3I or pentafluoropyridine. After workup, adducts 3 and 4 were isolated in yields of 80% and 17% respectively. The 119Sn NMR signals (−36.1 and −172.1 ppm) for tin atoms of 3 and 4 shift to the up-field in comparison with those in 1 (135.9 ppm) and 2 (342.7 ppm). The C1–Sn bond distances in 3 (2.215(2) Å) and 4 (2.240(3) Å) are approximately the same as those found in 1 (2.278(2) Å) and 2 (2.200(3) Å) (Fig. 3B and Table S4). Interestingly, when the LiCl was added as an additive into the THF solution containing 4, the white participation 5 was obtained via Cl/F ligand exchange. A simple one-pot preparation of 5 could be achieved via the treatment of 1 with pentafluoropyridine and LiCl. The structural parameters of 5 are very close to what were found for 4 (for details, see Table S4). Under the condition that oxidative addition and ligand metathesis can proceed smoothly, 1 shows the possibility to break through the boundaries between small molecular activation and chemical bond formation. Therefore, we contemplated the feasibility of employing 1 as catalyst for redox catalytic reactions.

Redox catalysis of the stannylene

Inspired by the pioneering work of Radosevich, Cornella, et al.27,43,44,45 on organopnictogen-mediated catalytic hydrodefluorination (HDF), this unsophisticated yet meaningful reaction was chosen as a template reaction to shed light on the application of 1 in redox catalysis. The HDF of pentafluoropyridine (6a) with PhSiH3 is selected as our prototypical system. We were surprised to discover that CDPSn 1 did catalyze the HDF of 6a (Fig. 4A). A variety of reaction conditions were examined to explore the effects of stannylene, silane, solvent and temperature on the HDF of 6a (For detail, see Table S1–2). Systematic screening of reaction conditions led to the identification of optimal conditions as follows: 0.4 equiv. of PhSiH3 and 10 mol% of CDPSn 1 in THF at 60 °C for 2 h. Under the conditions, the desired product 2,3,5,6-tetrafluoropyridine 7a was obtained in almost quantitative yield. To the best of our knowledge, this is catalytic HDF reaction using a low-valent stannylene as the catalyst.

Fig. 4: Mechanism studies for SnII-catalyzed hydrodefluorination of pentafluoropyridine.
figure 4

A Proposed catalytic cycle of HDF reaction. B Overlay of 31P NMR spectra (THF-d8, 162 MHz, argon atmosphere, room temperature.) of the reaction of 4 with PhSiH3 (1.0 equiv.) recorded at variable time. C The calculated potential energy surface of the catalytic cycle of HDF reaction (in kcal/mol). The free energy was obtained at the theoretical level of BP86-D3(BJ)/def2-TZVP/THF(IEFPCM)//BP86-D3(BJ)/def2-SVP/THF(IEFPCM).

Full size image

The studies then segued to investigate the reaction mechanism after confirming the feasibility of catalytic HDF using 1 as catalyst. The plausible mechanism is depicted in Fig. 4A. The reaction initiates with the oxidative addition between nucleophile 1 with 6a through a stepwise pathway intermediated by Meisenheimer complex 846, resulting in the SnIV intermediate 4. Then the hydrosilane A reacts smoothly with 4 to afford Sn–H species 9 via ligand metathesis along with the formation of fluorosilane B. The subsequent reduction elimination of 9 leads to catalyst turnover and finally delivers product 7a.

The process for the formation of intermediate 4 via oxidative addition of 1 with 6a was well-defined, as mentioned earlier for the reactivity studies of 1 (Fig. 3). To shed light on the ligand metathesis and reductive elimination processes, the in-situ reaction between SnIV compound 4 and PhSiH3 (1.0 equiv.) in THF-d8 was investigated by NMR spectroscopy at different time intervals (Figs. S2–4). After 5 min, a high field signal at 10.8 ppm (blue labeled) was observed in the 31P NMR spectrum along with the distinctively diminished signal at 13.4 ppm (green labeled) of 4, which unambiguously validated the formation of SnIV intermediate 9 via F/H ligand metathesis (Fig. 4B). The green labeled signal disappeared completely 30 min later, demonstrating that the F/H ligand metathesis reaction could finish in a short time. The subsequent reductive elimination process resulted in the generation of the catalyst CDPSn 1 detected by the 31P NMR spectroscopy (δ31P = 24.1 ppm, red labeled). The complete transformation from 9 to 1 requires approximately 12 h, which is considerably longer than oxidative addition (within 5 min) and ligand metathesis (within 30 min) processes, indicating that reductive elimination is the rate-limiting step in HDF of 6a. Further evidence supporting the formation of 9 via ligand exchange was obtained by high-resolution mass spectrometry (HRMS) (Fig. S5) and 19F NMR spectroscopic analysis (Fig. S3). The triplet 19F NMR resonance at δ19F = −171.6 ppm, assigned to the F atom bonding with Sn atom in 4, disappeared within 5 min after the addition of PhSiH3, revealing that the reaction between the 4 and PhSiH3 was conspicuous (Fig. S3). The generation of 9 was also detected via 19F NMR in which F atoms resonate at −97.4 and −124.5 ppm respectively.

To further understand the reaction mechanism, DFT calculations were performed on the potential energy surface (PES) of the reaction at the theoretical level of BP86-D3(BJ)/def2-TZVP (Fig. 4C). The computational results indicate that the oxidative addition of SnII in 1 to the C–F bond can be accomplished with thermodynamic favorableness, involving energy barriers of ΔG‡ = 9.0 kcal/mol for the formation of a Meisenheimer-type intermediate 8 and facile fluoride migration to afford intermediate 4 (ΔG‡ = 0.3 kcal/mol). This process is a stepwise nucleophilic aromatic substitution (SNAr) reaction leading to the initial Sn–F formation46. Subsequently, PhSiH3 participates in the ligand metathesis of the Sn–F bond, proceeding through a four-membered ring transition state TS3 and overcoming an energy barrier of ΔG‡ = 16.9 kcal/mol to reach the hydride intermediate 9. Finally, a reductive elimination step leads to the release of the product and regeneration of catalyst 1. It is noteworthy that reductive elimination is the rate-determining step of the reaction (ΔG‡ = 25.3 kcal/mol), which is consistent with experimental results. Other possible mechanistic pathways (e.g. single electron transfer47) were also investigated and discussed in view of the computed energy profile (see Supplementary Section 2.4.7). The energy barriers for the alternative mechanisms are all higher than the proposed mechanism (shown in Fig. 4C) involving C–F oxidative addition, F/H ligand metathesis, and C–H reductive elimination.

With an in-depth comprehension of the mechanism of CDPSn 1-catalyzed HDF reaction, the substrate scope of the HDF was tested under the optimized reaction conditions (Fig. 5). The scope of the catalytic HDF reaction includes pentafluoropyridine 6a and other fluoroarenes with distinct electronic properties. Pentafluorobenzenes bearing electron-withdrawing or electron-neutral groups were all well-tolerated, affording the desired para-hydrogenated products 7b–7g in high yields (84%–97%) and with high selectivity. The HDF of pentafluorobenzenes with formyl group or electron-donating group could also proceed under the standard condition (6h–6k), delivering the hydrodefluorinated products in diminished yields (28–52%). For pentafluorotoluene 6 l and pentafluoroanisole 6 m with stronger electron-donating groups, however, only trace amounts of desired products were detected. Remarkably, the catalytic HDF required more time as the decrease of the electron-withdrawing abilities of the substituents, confirming the significant effect of the electronic properties of the substrates on the reaction. Reactants with multiple active sites were also within the scope of the investigation. Excessive PhSiH3 would lead to the diHDF of perfluorobiphenyl 6n, 2,4,6-trifluorobenzonitrile 6o, 2,3,4,5,6-pentafluorobenzonitrile 6p and hexafluorobenzene 6q, giving the corresponding mono- or double-hydrogenated products in moderate to good yields. Partially fluorinated substrate 6o proved to be a reliable substrate, producing the mixture of 2,4-difluorobenzonitrile 7o (46%) and 2-fluorobenzonitrile 7o’ (48%). Furthermore, the selectivity of the products in the HDF reaction of 6p could be controlled by altering the amount of PhSiH3. Compared to previous pnictogen systems27,44,45, catalyst 1 showed better performance in the case of substrates with weak electron-donating groups (6i, 6j) and non-perfluorine substitutes (6o). These findings disclose that the functional groups are generally tolerated in SnII-catalyzed HDF reaction, and again highlight the feasibility of employing stannylene 1 as surrogates of transition metal catalyst in redox catalysis.

Fig. 5: Scope of the catalytic hydrodefluorination of fluoroarenes.
figure 5

Reactions were performed on a 0.25 mmol scale; Reaction time and yields were reported in parentheses. Yields were calculated by quantitative 19F NMR using 4-fluorotoluene as internal standard, unless stated otherwise. aIn the presence of 0.4 equiv. of PhSiH3. bIsolated yields. cIn the presence of 2.0 equiv. of PhSiH3.

Full size image

Ever since Power’s groundbreaking review on the concept of “Main-group elements as transition metals”, the group 14 carbene analogues have been central for main group chemistry12. However, the application of these low-valent species in redox catalysis remains elusive due to the inherent challenges associated with achieving the EII/EIV (E = Si, Ge or Sn) catalytic cycle. Herein, we have reported the synthesis and isolation of a nucleophilic stannylene 1 supported by the tridentate pincer carbodiphosphorane ligand. Stannylene 1 can function as a strong Sn-donor ligand, as demonstrated by the synthesis of iron complex. The observation of the oxidative addition and ligand metathesis processes during the reaction of 1 with pentafluoropyridine 6a is particularly noteworthy, as it resembles elementary steps in transition metal redox catalysis. Furthermore, we have successfully achieved the catalytic HDF reaction using stannylene 1 as a redox catalyst, which undergoes sequential C–F oxidative addition, F/H ligand metathesis, and C–H reductive elimination processes. As yet, this is an example of main group catalysis involving SnII/SnIV redox couple. Currently, we are exploring further possibilities for applying the group 14 carbene analogues in various redox catalytic reactions.

Methods

Synthesis of stannylene 1

Stannous chloride (183 mg, 1.0 mmol, 1.0 equiv.) was added to a THF (15 mL) solution of dilithio-carbodiphosphorane (548 mg, 1.0 mmol) at −30 °C. The reaction mixture was allowed to warm to room temperature after 10 min and then stirred for 12 h to form a yellow solution. Subsequently, all the volatiles were removed under reduced pressure, and the resulting solid was washed with n-hexane (10 mL). The solid was re-dissolved in toluene (30 mL) and the solution was filtered by using a PTFE syringe filter to remove LiCl. The solvent was removed in vacuum and the product 1 was obtained as yellow powder. Yield: 464 mg (71%). Bright yellow crystals, suitable for X-ray analysis, were obtained by slow diffusion of n-hexane onto a solution of 1 in toluene at room temperature.

General procedure for HDF

In a glovebox, stannylene 1 (14.0 mg, 0.02 mmol, 10 mol%) and THF (1.25 mL) were added to a Schlenk tube. The tube was taken out of the glovebox and charged with polyfluoroarenes (0.25 mmol), PhSiH3 (0.4–2.0 equiv.) upon stirring. The reaction was carried out for different time at 60 oC. After completion of the reaction, the mixture was cooled to room temperature and exposed to air to quench the reaction. The product was isolated by flash chromatography. If product was volatile, 4-fluorotoluene (27.5 mg, 0.25 mmol, 1.0 equiv.) was used as internal standard and the product was analyzed by quantitative 19F NMR.