Introduction

Polycyclic aromatic hydrocarbons (PAHs) with unique optical and electrical properties are commonly used in organic light-emitting diodes (OLEDs)1,2,3, organic field effect transistors (OFETs)4, and organic solar cells (OSCs)5,6. Embedding heteroatoms (such as B, N, S, O, and P) into the π-conjugated frameworks of PAHs produces unexpected opportunities for optoelectronics advancement by providing an adjustable band gap, high electrochemical activity, and desirable charge polarization7,8,9. The replacement of the C=C bond with an isoelectronic B–N unit improved the stability, produced tunable energy levels, and amplified photoluminescence quantum yields (PLQYs)10,11. PVI=N-doped phenanthrenes with enhanced PLQYs were achieved via the introduction of PVI chirality, showing significant potential in circularly polarized luminescent materials (Fig. 1a)12. Sulfur-decorated PAHs, characterized by soft-atom and electron-rich properties, serve as valuable n-type semiconductors13 and can readily oxidize into p-type SVI (sulfonyl) semiconductors14, providing intriguing ambipolar transport characteristics and high resonance energy. The incorporation of the SVI motif enables the solubility and intermolecular interactions of PAHs and increases the processability of the materials15. Sulfoximines16 and sulfondiimines17, as aza-replacement of sulfones, provide customizable N-handles for fine-tuning the reactivity and functionality, concurrently bestowing chirality upon sulfur (Fig. 1b). These pioneering studies show the great potential of unexploited chiral SVI=N-embedded PAHs with attractive photophysical and chiroptical properties. Great contributions have been made to the construction of sulfur stereogenic centers18,19,20,21,22,23,24. However, a difficult challenge remains in differentiating similar aromatic compounds stereoselectively, and a π-extension system is rare in chiral naphthalene-type heterocycles. Powerful C–H activation25,26,27,28,29,30 is a straightforward strategy for heteroatom doping with polycyclic aromatics7,8,9. In contrast to C-31,32, Si-33,34, and P-chirality35,36,37, the construction of S(VI)-chirality poses notable challenges due to the presence of S=N double bond (Fig. 1c and Fig. S7). First, the spatial configuration of S=N bond leads to a longer distance between the coupled carbon atoms, particularly in rigid frameworks (4.87 Å), enabling C–H activation more difficult. Second, the polarizability of S=N bond, especially in the second S=N bond, imparts a high electron density to the nitrogen (−1.177), which can potentially coordinate with palladium and disrupt aryl coordination38. Along with our exploration of sulfoximine chemistry39,40,41, we herein present a Pd-catalyzed desymmetrization for the construction of chiral SVI=N-doped polyaromatic heterocycles, producing an (R)-configuration with the assistance of C–H···π interactions of TMCPA in split aromatics and an (S)-configuration originating from impeded C–S bond rotation in combined aromatics; moreover, promising photophysical and chiroptical properties were achieved from these polyaromatic heterocycles (Fig. 1d).

Fig. 1: Heteroatom doping of PAHs.
figure 1

a Incorporation of B–N, PV=N, and SVI=N bonds. b Sulfoximine and sulfondiimine as aza-analogs of sulfone. c Challenges in the Pd-catalyzed C–H activation of S=N-containing compounds. d Pd-catalyzed enantioselective desymmetrization of diaryl sulfoximines and sulfondiimines involving split (acyclic) and combined (cyclic) aromatics (this work).

Results

We began enantioselective C–H coupling with symmetric diaryl sulfoximine (1a) for the chiral dibenzothiazine oxide 2a with Pd(OAc)2 as a catalyst (Table 1). Chiral ligands play a crucial role in controlling the sulfur(VI) stereogenic center. Compared to L1, an electron-rich monodentate phosphine ligand L2 produced the cyclization product in 9% yield and 38% enantiomeric excess (ee) (Table 1, entries 1–2). A series of electron-donating bidentate ligands, including N/P- and P/P-based motifs, were investigated (entries 3–7). Chiraphos (L5) provided a moderate yield and a diminished enantiomeric excess of 2a, while a rigid ligand (L7) caused improved enantioselectivity. However, increasing the steric hindrance in the coordinating environment (L8) substantially attenuated the ee (entry 8). The impact of solvents was equally important (entries 9-13). Dimethylformamide (DMF) enabled the complete conversion of 1a but produced an almost racemic product (entry 9). The protonated byproduct from 1a was exclusively observed in a less bulky protonic solvent (isopropylalcohol (iPrOH), entry 10). Notably, tert-amyl alcohol (tAmylOH) and 1,2-dichloroethane (DCE) enhanced the stereoselectivity (entries 11 and 12). The yield of product 2a was significantly increased without compromising the optical purity of these mixed solvents (entry 13). Interestingly, replacing pivalic acid (PivOH) with tetramethylcyclopropylcarboxylic acid (TMCPA) substantially promoted both the reaction efficiency and enantioselectivity (entry 16). In the absence of the additive, a large decrease in the yield was observed, but the ee only slightly decreased (entry 17). These results demonstrated the pivotal role of the acid additives in facilitating the concerted metalation-deprotonation (CMD) process.

Table 1 Optimization of the reaction conditionsa

A comprehensive investigation was conducted with the optimized reaction conditions for the enantioselective synthesis of chiral SVI=N-heterocycles via the desymmetrization of the diaryl sulfoximines (Fig. 2). Electron-rich and electron-poor substituents at diverse aromatic positions tethered to sulfur(VI) were well tolerated, producing annulated sulfoximines in excellent yields and ee values (2a-2g). The absolute (R)-configuration of 2a was further confirmed via X-ray diffraction analysis. Heteroaryl sulfoximines containing thienyl and pyridinyl moieties were shown to be efficient in this enantioselective desymmetrization process (2 h and 2i). When the cyclic substrate dibenzothiophene oxide 1j was used, exceptional enantioselectivity was obtained for the corresponding product (2j, >99% ee); here, the structural rigidity of the substrate reversed the stereoselectivity (2a vs. 2j) with the (S)-configuration, which was confirmed via X-ray diffraction analysis. Highly enantioenriched pentacyclic sulfoximines were produced when the five-membered adaptor was replaced with a six-membered ring with the oxygen/nitrogen/carbonyl groups in dibenzothiophene oxide (2k-2m). The expanding of the conjugated system successfully increased the enantiomeric purity of the sulfur(VI)-chiral center (2n). In this cyclization strategy, excellent compatibility was exhibited with the N-substituted aromatic of sulfoximines possessing electronically diverse groups at different positions (2o-2u). Condensed and heteroaryl-derived sulfoximines efficiently underwent enantioselective cyclization (2v and 2w). Notably, a double-site carbazole-based substrate produced corresponding product 2x in >99% ee with excellent diastereoselectivity. Furthermore, the current desymmetrization strategy was successfully extended to the diaryl sulfondiimines, producing the challenging dibenzothiazine imides (4a-4o) in excellent yields and enantiomeric excesses (up to 99% ee). In comparison to sulfoximine 1g, sulfondiimine 3e with an electron-donating group (–OMe) contained a mixture of para- and ortho-positions (4e1 and 4e2). The diverse protecting groups on nitrogen, such as –Ns (4-nitrobenzene sulfonyl), –Cs (4-cyanobenzene sulfonyl), and –Boc (t-butyloxy carbonyl), displayed excellent reaction efficiency and stereoselectivity (4f-4h). The absolute configuration of 4o was confirmed via X-ray diffraction analysis as a representative example of a sulfondiimine product.

Fig. 2: Library of the chiral sulfoximines and sulfondiimines.
figure 2

aConditions: 1 or 3 (0.15 mmol), Pd(OAc)2 (6 mol%), L7 (6.6 mol%), TMCPA (30 mol%), K2CO3 (1.1 equiv), tAmylOH/DCE (6/1), 100 °C, N2, 8 h (for sulfoximines 1) or 12 h (for sulfondiimines 3). bPd(OAc)2 (8 mol%), L7 (8.8 mol%). cPd(OAc)2 (10 mol%), L7 (11 mol%), TMCPA (40 mol%). dK2CO3 (3 equiv). eK2CO3 (2 equiv.). Ns = 4-nitrobenzene sulfonyl. Cs = 4-cyanobenzene sulfonyl.

The parallel kinetic resolution of the diaryl sulfoximines and sulfondiimines bearing diverse para-substituents was also evaluated under standard conditions (Fig. 3a). Excellent stereoselectivities were achieved with ee values greater than 97% in most cases. When opposing electronic groups were present on either side of the aromatic ring, the enantioselective annulation slightly favored the electron-deficient substituents (2y and 4p). To highlight the practical synthetic utility of the structurally novel chiral SVI=N-doped heterocycles, the derivatization of sulfoximine and sulfondiimine was further performed (Fig. 3b). The process began with the lithium-halogen exchange of 2-bromobiphenyl 5, followed by the nucleophilic addition of 2m to form the corresponding alcohol intermediate 6a. Then, an intramolecular Friedel−Crafts reaction produced sulfoximine 6b with a fluorenyl-based spiro skeleton; here, the absolute (S)-configuration of 6b was confirmed via X-ray diffraction analysis. Notably, the gram-scale synthesis of cyclic sulfondiimine 4h was accomplished with an enhanced yield of 95% while retaining stereochemical fidelity. Subsequently, the Suzuki coupling of 4h with phenylanthracenyl boronic acid 7 provided extended conjugated product 8a, in which free sulfoximine 8b was liberated via trimethyl silyl iodide (TMSI)/triethylamine (Et3N) treatment.

Fig. 3: Synthetic applications.
figure 3

a Parallel kinetic resolution. b further derivatization of the S = N-doped heterocycles.

With synthesized S=N-doped heterocycles in hand, the photophysical characteristics of 2c, 2l, 2x, and 4m were investigated. The UV‒Vis spectra revealed a maximum absorption peak at approximately 280 nm for 2c and 2x and a higher absorption peak at approximately 365 nm for 2l and 4m. Due to the expansive conjugate system, 2x displayed an additional absorption feature at 388 nm (Fig. 4a). At the maximum excitation wavelength, sulfoximines 2c, 2l, and 2x exhibited broadened fluorescence emission bands, with 2x showing the farthest emission wavelength at 532 nm. However, 4m with an N=S=N motif displayed a noticeable reduction in fluorescence intensity (Fig. 4b). The absolute quantum yield was determined to be 0.12 for 2x in CH2Cl2. Moreover, the chiroptical properties of 2x with two chiral sulfur (VI) centers were studied via electronic circular dichroism (ECD) and circularly polarized luminescence (CPL) spectroscopy; here, the enantiomers of (R, R)-2x and (S, S)-2x showed evident mirror images from 250 nm to 490 nm in their CD spectra (Fig. 4c). The mirror-image spectrum in Fig. 4d showed the CPL activity of compound 2x, and the luminescence dissymmetry factor (glum) of (R, R)-2x was measured to be -0.0022 at 563 nm (Fig. 4e, f). These results highlighted the significant application potential of the SVI=N-heterocycles in advancing enantiomeric photoelectric materials, such as light-emitting diodes (LEDs), organic photovoltaics and chiroptical devices12,42.

Fig. 4: Photophysical and chiroptical properties of SVI = N-doped heterocycles.
figure 4

a Absorption spectra of 2c, 2 l, 2x, and 4 m in DCM (1.0 × 10−5 M). b Emission spectra of 2c, 2l, 2x, and 4m in DCM (1.0 × 10−5 M). c CD spectra of (R, R/S, S)-2x in DCM (1.0 × 10−5 M). d CPL spectra of (R, R/S, S)-2x in DCM (1.0 × 10−5 M) excited at 388 nm. e glum values wavelength curves for (R, R/S, S)-2x. f Structure, fluorescence image of CH2Cl2 irradiated at 365 nm, and optical values for (R, R)-2x.

To clarify the reaction mechanism, several mechanistic experiments were conducted. When the mono-oxide and dioxide derivatives of ligand L7 were tested43,44, products 2a and 2j were obtained in low yield or not observed at all (Fig. 5a). These results suggest that the diphosphine is the active ligand in the current transformation, rather than its oxidized forms. Parallel kinetic isotope effect (KIE) experiments45 with substrates 1aa and 1aa’ revealed a kH/kD ratio of 2.7 (Fig. 5b), indicating that C–H activation is involved in the rate-determining step.

Fig. 5: Mechanistic studies.
figure 5

a Control experiments for mono-oxide and dioxide derivatives of ligand L7. b Parallel kinetic isotope effect experiments.

Based on the experimental results and the reaction model for a bidentate ligand46, we next investigated the origins of the enantioselectivity in substrate 1a using density functional theory (DFT) calculations. The chiral Pd(0) catalyst undergoes oxidative addition with 1a, followed by ligand exchange with TMCPA to form intermediate INT1 (Fig. 6a). Two possible modes of C–H activation were considered due to the square planar geometry of Pd(II). In the first mode, bidentate coordination of the phosphine ligand leads to outer-sphere C–H bond activation. In the second mode, monodentate coordination of the phosphine ligand leaves a coordination site for the carboxylate anion, enabling inner-sphere C–H activation. Computational results indicate that the energy barrier for inner-sphere C–H activation is at least 9.1 kcal/mol lower than that for outer-sphere C–H activation (Fig. 6b). In the outer-sphere C–H bond activation transition states, steric hindrance exists between the ligand and the S-aryl group, regardless of which aryl C–H bond is activated. To mitigate the steric repulsion, the resulting C–Pd bond is elongated. In TS3-(R)-A and TS3-(R)-B, the C–Pd bond lengths are 2.50 Å and 2.52 Å, respectively, compared to 2.30 Å in TS3-(R). In contrast, the inner-sphere C–H bond activation transition state TS3-(R) not only lacks steric hindrance but also benefits from favorable π···π, C–H···O, and C–H···π non-covalent interactions, confirmed by independent gradient model based on Hirshfeld partition (IGMH) analysis47. The bidentate phosphorus-coordinated INT1 isomerizes to INT2 with the aryl fragment of sulfoximine coordinated to palladium. Subsequent concerted metalation-deprotonation via TS3 generates the palladacycle intermediate INT4, which undergoes reductive elimination (TS5, 12.3 kcal/mol), irreversibly producing the reduced palladium(0) complex INT6 with product coordination. The DFT-computed free energy profile suggests that C–H activation has the highest energetic barrier, with an overall barrier of 29.8 kcal/mol, which is consistent with the results of parallel KIE experiments (Fig. 5b).

Fig. 6: Calculations for the reaction mechanism.
figure 6

a The enantioselective C–H annulation of substrate 1a were investigated at the level of ωB97M-V/def2-TZVPP-SMD(2-methyl-2-propanol)//B3LYP-D3(BJ)/6-31G(d)-LANL2DZ. b The optimized structures and their relative Gibbs free energies of inner-/outer-sphere C–H activation transition states.

To elucidate the origin of the opposite enantioselectivity observed in split and combined substrates, we conducted a detailed study on the enantiodetermining step (C–H activation). For split substrate 1a, INT1 can rotate around the S–C and N–C bonds prior to C–H activation, generating two sets of diastereomeric transition states (Fig. 7a). In TS3-(R)/TS3-(S)-iso1, the oxygen atom of sulfoximine is distal to the Pd plane, whereas it is proximal in TS3-(S)/TS3-(R)-iso1. TS3-(R)/TS3-(S)-iso1 can isomerize to TS3-(S)/TS3-(R)-iso1 via C–S bond rotation. Among the transition states producing the R-enantiomer, TS3-(R)-iso1 is 6.7 kcal/mol higher in energy than TS3-(R) for two reasons. First, in TS3-(R)-iso1, the sulfoximine undergoes substantial distortion to maintain the Pd’s square planar geometry. This distortion is reflected in the highlighted dihedral angle ϕ(OSNC), which measures 105.9° in TS3-(R)-iso1 compared to 69.2° in the DFT-optimized geometry. Second, TS3-(R) benefits from multiple favorable noncovalent interactions between the phosphine ligand, substrate, and TMCPA anion (Fig. 6b). Among the transition states that produces the S-enantiomer, TS3-(S) is 4.5 kcal/mol lower in energy than TS3-(S)-iso1, stabilized by favorable π···π and C–H···O noncovalent interactions, despite increased distortion (99.4° vs 81.0° in TS3-(S)-iso1). The energy difference is also attributed to steric hindrance in TS3-(S)-iso1 involving TMCPA, the tolyl fragment of sulfoximine, and the methoxy group of ligand. Therefore, TS3-(R) and TS3-(S) are the key enantioselectivity-determining transition states, with TS3-(R) being 1.7 kcal/mol lower in energy, which is consistent with experimental observations. The additional C–H···π interactions between TMCPA and aryl groups of the ligand and substrate facilitate enantiodiscrimination in TS3-(R) (Fig. S8 and S9), explaining why different acid additives affect enantio-selectivity experimentally (Table 1, Entries 13-17). When C–S bond rotation is restricted, only TS3-(S) and TS3-(R)-iso1 are accessible, leading predominantly to form S-enantiomer, as TS3-(S) is 5 kcal/mol lower in energy than TS3-(R)-iso1. When combined substrates are used, the conformational restriction efficiently inhibits the C–S bond rotation feasible in split molecules, thus reversing the enantioselectivity to favor the S-enantiomer. Computational studies on combined substrate 2j shows consistent results with split substrates, confirming that restricting C–S bond rotation leads to S-enantiomer as the major product (ΔΔGǂ = 2.9 kcal/mol, Fig. 7b). Interestingly, analysis of the TS7-(R) and TS7-(S) structures reveals that enantioselectivity is driven by π···π stacking between the ligand and the N-aryl fragment, independent of acid additives. To further validate this chiral induction model, we conducted the reaction under standard conditions without TMCPA, relying on the acetate anion from palladium acetate for C–H activation. The reaction produced the target product with >99% ee, fully supporting the reliability of the computational model.

Fig. 7: Enantioselectivity investigations.
figure 7

a Enantioselectivity-determining transition states and their relative Free energies for split substrate 1a. b origins of enantioselectivity for combined substrate 1j.

Discussion

In conclusion, a Pd-catalyzed enantioselective C–H annulation of sulfoximines and sulfondiimines was successfully established with the assistance of TMCPA. This strategy provides a robust pathway for accessing chiral SVI = N-embedded heterocycles with excellent yields and enantioselectivities. The π-extension of the sulfoximines and sulfondiimines was efficiently achieved via the current strategy. Modulating the rigidity of the diaryl sulfoximines enabled the formation of an SVI = N-heterocycle with an inverted stereoconfiguration. The concerted Pd-catalyzed deprotonation was identified as the rate- and enantio-determining step. DFT studies revealed that distinct enantioselectivities emerged from the C–H···π interactions of TMCPA in split sulfoximines and from the π-π stacking between the ligand and the N-aryl motif in combined sulfoximines. Further study of the chiral SVI = N-doped π-conjugated materials is in progress in our laboratory.

Methods

General reaction procedure

A 25 mL oven-dried Schlenk tube was charged with diaryl sulfoximines 1 or sulfondiimines 3 (0.15 mmol, 1.0 equiv), Pd(OAc)2 (6 mol%), (R,R,R,R)-MeO-BIBOP L7 (6.6 mol%), TMCPA (30 mol%), K2CO3 (1.1 equiv), and a stirring bar. The tube was fitted with a N2 balloon, and a mixture of 1 mL tAmylOH/DCE (6/1) was added. Subsequently, the reaction mixture was transferred to 100 oC and stirred for 8–12 h. Upon cooling to room temperature, the reaction mixture was diluted with DCM, filtered, and concentrated under reduced pressure. The resulting residue was subjected to purification by column chromatography, affording the corresponding product.