Aromaticity and structure switching of cyclopropametallaindole to metallaquinolinium

Abstract

Skeletal reshaping between aromatic frameworks remains challenging. The rearrangements between azulene and naphthalene structures are some of the only examples to date. Metallaaromatics, an important class of aromatic compounds, attract broad attention for their fascinating structures and diverse reactivities. Herein, we report the reshaping of cyclopropametallaindole to metallaquinolinium, demonstrating both aromaticity and structure switching. The proposed electrophilic addition and elimination mechanism has been investigated by radical trapping experiment and theoretical study. This transformation achieves both the aromaticity and structure switching of a metallacycle, and demonstrates a new synthetic route for metallaaromatics.

Aromatic rings are stable, making it hard to reshape their frameworks. Here, the authors convert one metal-containing aromatic ring into another, offering a new route to metallaaromatic compounds.

Introduction

Electrophilic aromatic substitution (EAS) stands as one of the fundamental and widely applied transformations for molecular modification. This classical process generally involves the addition of an electrophile on an electron-rich aromatic carbon with the generation of σ complex, followed by the elimination of the ipso-proton, which regenerates the aromatic structure (Fig. 1)¹. In 1995, the Bleeke group pioneered the extension of EAS reaction to an iridathiophene complex². Since then, examples of EAS on other metallaaromatics³, such as iridabenzofurans^4–6, iridabenzenes⁷, osmabenzenes^8,9, osmabenzynes^10,11, osmapentalene derivative¹² and osma-dehydro[11] annulene¹³ have been documented. Other unique reaction modes of aromatic compounds continuously attract great interest from experimental and theoretical chemists. Reshaping one aromatic framework to another constitutes one of the challenges and has limited precedents. Only the azulene–naphthalene interconversions have been reported (Fig. 1)^14–16. Azulene can reshape to the thermodynamically more stable naphthalene at high temperatures with the maintaining of ring atoms and size (10-membered aromatic bicycles)^14,15.

Fig. 1. Some unique reaction modes of aromatics and metallaaromatics.

Electrophilic aromatic substitution and aromatic skeletal rearrangement reactions of aromatics and metallaaromatics.

With the involvement of transition metals, metallaaromatics have demonstrated fascinating structures and diverse reactivities^3,17–21. The Xia group has presented the transformations of metallaindenols (10-membered fused tricycles containing a 7-membered aromatic cyclopropametallabenzene) to metallapentalenes (9-membered aromatic metallatricycles), as the first skeletal reshaping from one metallaaromatic framework to another²². However, with the extension of the conjugation system, the ring size has changed. In this work, we report the skeletal rearrangement of cyclopropametallaindoles (10-membered aromatic metallatricycles) to metallaquinoliniums (10-membered aromatic metallabicycles) through an electrophilic addition and elimination process (Fig. 1). This reshaping represents an example of (i) aromaticity switching of (π + σ) aromatics to π aromatics, (ii) structural switching of aromatic fused bicycles to monocycles, (iii) ring expansion of metallaindoles to metallaquinoliniums. The reaction mechanism and the aromatic property are studied by theoretical calculations.

Results

Synthesis and characterization of cyclopropaosmaindole complexes

As shown in Fig. 2, complexes 1 were obtained from the reactions of OsCl₂(PPh₃)₃ with 2-ethynylanilines in tetrahydrofuran (THF) at 60 °C under argon atmosphere. Single-crystal X-ray diffraction analysis reveals that complex 1b adopts a slightly distorted octahedral structure (Fig. 3). The α-osmaindole is fused with a metallacycloprop-1-ene unit, and the tricyclic structure is nearly planar (the mean deviation from the least-squares plane is 0.0283 Å). The bond lengths Os-C1 (2.179(8) Å), Os-C2 (1.968(9) Å), and C1–C2 (1.370(13) Å) are all within the range of those reported osmacyclopropene structures (Os–C single bonds, 2.082–2.368 Å; Os = C double bonds, 1.926–2.046 Å; C–C bonds, 1.336–1.443 Å)^22–32. The Os-N bond distance (2.046(6) Å) is at the lower end of those reported data for osmapyrole fragment (2.047(4)–2.197(12) Å)^33–41. Its solution NMR data are consistent with the solid structure. For instance, the ³¹P {¹H} NMR spectrum of 1b exhibits a singlet at −15.0 ppm in CDCl₃. In the ¹H NMR spectrum, the metallacyclopropene CH₂ signal appears at 4.84 ppm as a triplet (J = 10.0 Hz), while the singlets of the NH and the CH₃ groups are observed at 8.61 and 2.29 ppm, respectively. Furthermore, the resonance of Os = C splits into a triplet at 233.6 ppm (²J(PC) = 1.0 Hz). Complexes 1a, 1c, and 1d show similar NMR spectra (Figs. S15–S25), and their structural formula have been characterized by the high-resolution mass spectrometry (HRMS, Figs. S37–40). Complexes 1 are proposedly formed through a similar pathway as that of osmabenzofuran-fused metallacyclopropene complexes²³. The alkyne- and amino-coordinated intermediate Int A is generated first, which subsequently undergoes proton transfer to afford 1 (Fig. 2). The UV–Vis absorption spectra of complexes 1 were recorded in the 300–800 nm range (Fig. S44). All complexes exhibited strong absorption bands in the UV region (300–450 nm), and the absorption intensity of these peaks varies among the derivatives (1a, log ε = 6.500 M⁻¹ cm⁻¹; 1b, log ε = 6.583 M⁻¹ cm⁻¹; 1c, log ε = 6.502 M⁻¹ cm⁻¹; 1d, log ε = 6.516 M⁻¹ cm⁻¹, where ε is the molar extinction coefficient). Only very weak absorptions are observed in the visible region above 500 nm.

Fig. 2. Formation of complexes 1.

Synthesis of cyclopropaosmaindole complexes.

Fig. 3. Structure of complex 1b.

ORTEP drawing of complex 1b with thermal ellipsoids set 50% probability (phenyl groups in PPh₃ and hydrogen atoms are omitted for clarity). Selected bond distances (Å) and angles (°): Os1-C1 2.179(8), Os1-C2 1.968(9), Os1-N1 2.046(6), Os1-Cl1 2.462(2), Os1-Cl2 2.443(2), Os1-P1 2.404(2), Os1-P2 2.4130(19), C1-C2 1.370(13), C2-C3 1.394(12), C3-C4 1.445(12), C4-N1 1.286(12), Cl1-Os1-Cl2 89.39(8), P1-Os1-P2 176.79(16), C1-Os1-C2 38.1(4), C2-Os1-N1 73.2(4), C1-C2-Os1 79.3(5), C2-C1-Os1 62.5(5), Os1-N1-C4 121.9(7), N1-C4-C3 113.5(8), C4-C3-C2 107.9(8), C3-C2-Os1 123.4(7).

Reshaping of cyclopropaosmaindoles to osmaquinoliniums

When we treated complexes 1 with N-chlorosuccinimide (NCS), an electrophile, in dichloromethane (DCM) at 40 °C, osmaquinolinium complexes 2 were obtained (Fig. 4). 1a and 1d gave the same product 2a. The metalla-bicyclic structures of 2a and 2c have been confirmed by single-crystal X-ray diffraction study (Figs. 5 and S9). The bond lengths of Os-C1 (2a, 1.962(4) Å; 2c, 1.974(8) Å) and Os-N1 (2a, 1.933(4) Å; 2c, 1.936(7) Å) are similar to those of the known osmapyridinium (osmapyridine) structures (Os–C, 1.922(4)–2.199(7) Å; Os-N, 1.937(3)–2.0924(6) Å)^42–44. In the ¹H NMR spectra, the characteristic Os−CH signals are found at 57.07 ppm (2a), 55.65 ppm (2b), and 75.20 ppm (2c), which are much lower compared to the reported ones^42–44. Their Os-NH signals appear at 29.89, 28.65, and 34.68 ppm, respectively. This observation is resulted from the paramagnetism of complexes 2, with the lowest singlet and triplet states close in energy^42,44. The energy differences ∆E_ST (E_singlet −E_triplet, in kcal/mol) between the singlet and triplet states of complexes 1a, 2a, 2b, and 2c, respectively, have been calculated at three levels of theory (B3LYP, OPBE, and M06)^45–52. The results show that the singlet and triplet energies of complexes 2a, 2b, and 2c are extremely close, with the ∆E_ST ranging from −3.55 to 0.43 kcal/mol. Meanwhile, diamagnetic complex 1a exhibits a significantly more stable singlet state, with large ∆E_ST (−26.93 to −29.09 kcal/mol). The electron paramagnetic resonance (EPR) spectroscopy measurements of 2 further support their paramagnetic property (Figs. S34–S36).

Fig. 4. Reshaping of complexes 1 to 2.

Aromatic skeletal rearrangement from 1 to 2 and mechanistic study.

Fig. 5. Structure of complex 2a.

ORTEP drawing of complex 2a with thermal ellipsoids set 50% probability (phenyl groups in PPh₃ and hydrogen atoms are omitted for clarity). Selected bond distances (Å) and angles (°): Os1-C1 1.962(4), Os1-N1 1.933(4), Os1-Cl1 2.4144(10), Os1-Cl2 2.5095(10), Os1-P1 2.4256(10), Os1-P2 2.4250(10), C1-C2 1.375(6), C2-C3 1.419(6), C3-C4 1.433(6), C4-N1 1.355(5), Cl1-Os1-Cl2 97.54(3), P1-Os1-P2 168.00(4), C1-Os1-N1 88.1(4), C2-C1-Os1 126.6(3), C1-C2-C3 127.2(4), Os1-N1-C4 133.5(3), N1-C4-C3 121.4(4), C4-C3-C2 120.9(4).

In the ³¹P{¹H} NMR spectra, the signals of OsPPh₃ appear at 16.5 ppm (2a), 15.1 ppm (2b), and 27.7 ppm (2c). In the ¹³C{¹H} NMR spectra, the signals of OsCH, OsC = CCl, and OsNHC of 2a are observed at 138.3, 166.5, and 199.7 ppm. The ¹³C{¹H} NMR spectra for 2b and 2c could not be obtained due to their poor stability. While 2a exhibited only slight degradation after 16 h in DCM, 2b changed to a complicated mixture after 8 h, and 2c completely decomposed after 13 h. The lack of phosphonium substituents on the metallacycle may results in this instability^53,54. The instability of complexes 2 produced some side products and the formations of several unidentified species were observed. During the purification by column chromatography on silica gel, a portion of 2 was adsorbed. Thus, the yields of osmaquinolinium complexes 2 were relatively low. Complexes 2a–c have been characterized by HRMS (Figs. S41–43) and elemental analysis (EA). The UV–Vis absorption spectra of 2 were measured (Fig. S45). The absorption maxima of 2a–2c appear around 350 nm (2a, log ε = 6.524 M⁻¹ cm⁻¹; 2b, log ε = 6.511 M⁻¹ cm⁻¹, 2c; log ε = 6.468 M⁻¹ cm⁻¹), with a broad shoulder extending into the visible region (450–600 nm).

Though the conversion of indole (pyrrole) to quinoline (pyridine) has been reported as early as 1881 as the Ciamician–Dennstedt reaction⁵⁵, such skeletal transformation remains unknown for metallaaromatics. The proposed reshaping mechanism has been studied by DFT calculations at the B3LYP-D3(BJ)/6-311 G(2 d,p) + def2-TZVP(Os)/SMD(dichloromethane) level. As shown in Fig. 6, electrophilic addition of Cl atom onto the fused Os = C bond of 1a generates the cationic intermediate Int B via transition state TS1. This step is the rate-determining one with an energy barrier of 20.4 kcal/mol and is endothermic by 3.6 kcal/mol (at 298.15 K). Subsequently, the succinimide anion abstracts a hydrogen from the CH₂ of metallacyclopropene unit with simultaneous cleavage of the inner Os–C bond, ultimately yielding 2a’ and 2,5-dipyrrolidine-one via transition state TS2. The bicycle of Int B merges into six-membered metallapyridinium monocycle with an energy barrier of 18.9 kcal/mol and is highly exergonic by 37.8 kcal/mol (the computational details are provided in Supporting Information). The key reaction steps of reshaping aromatic cyclopropametallaindoles to metallaquinoliniums possibly involve the aforementioned EAS-like electrophilic addition and elimination process. This also represents a unique reaction mode for fused metallacyclopropene complexes⁵⁶, which can undergo oxidation coupling reactions⁵⁷, [3 + 1]⁵⁸, [3 + 2 + 1]⁵⁹, [3 + 2]^60,61 and [3 + 3]²⁷ reactions. The chlorination of benzene moiety at the para-position of N atom could occur before the skeletal rearrangement, as 1d gives a similar reaction result with NCS and indoles are more reactive than quinoliniums for EAS reactions. The condensed Fukui function has been employed to quantify the regioselectivity of complexes 1 towards electrophilic addition⁶². As shown in Fig. S4, C3 of 1a displays the largest f_k⁻ value (0.042), which would be the most nucleophilic site and undergoes chlorination first. While C5 (Os = C) is the most nucleophilic site of 1d, indicating the occurrence of the following electrophilic addition of Cl atom. Furthermore, another electrophilic chlorination reagent Palau’Chlor (2-chloro-1,3-bis(methoxycarbonyl)guanidine)⁶³ was employed, affording 2a in 39% yield from 1a and 44% from 1d. In addition, in the presence of TEMPO (2,2,6,6-tetramethylpiperidinooxy, 5.0 equiv.), 2a was obtained in 29% isolated yield from 1a under the same reaction conditions, which further suggests the role of NCS as an electrophile but unlikely a radical reagent during this transformation.

Fig. 6. Computed Gibbs energy profile (in kcal/mol) for reshaping 1a to 2a’.

DFT theory level: B3LYP-D3(BJ)/6-311G(2d,p) + def2-TZVP(Os)/SMD(dichloromethane).

There was no reaction between the cyclopropaosmabenzofuran complex²³ and NCS under the same reaction conditions (in DCM at 40 °C) for one day, highlighting the unique reactivity of cyclopropaosmaindoles (complex 1). To quantitatively rationalize this observation, we also computed the activation energy for the electrophilic addition step, which was 42.0 kcal/mol (Fig. S2), much higher than that of 1. When higher reaction temperature was employed (in dichloroethane at 80 °C), only a complicated mixture was produced. It is noted that there is no report for electrophilic substitution/addition reactions occurring at the C3 position of organic benzofurans, while indoles can readily undergo such reactions⁶⁴.

Aromaticity and molecular orbital analysis

To explore the aromatic property of complexes 1 and 2, DFT calculations were performed on the simplified model 1’, 2’S (singlet state) and 2’T (triplet state) (PPh₃ was replaced with PH₃ for simplification). The optimized bond lengths of 1’ and 2’S are close to the crystallographic data, with deviations within the range of 0.001–0.04 Å for the metallacycles (Fig. S1). NICS is a commonly used indicator for assessing aromaticity^65–68, where distinctly negative values denote aromaticity and positive values indicate antiaromaticity. As illustrated in Fig. 7a, the computed NICS(1)_ZZ value of the five-membered ring (5MR) in 1’ is −13.1 ppm, that of the six-membered ring (6MR) in 2’S is −20.9 ppm, and the NICS(0)_ZZ value of the three-membered ring (3MR) in 1’ is −71.7 ppm, indicating that these metallacycles are all aromatic. The NICS(1)_ZZ value of 6MR in 2’T is −5.2 ppm, indicating its weak aromaticity. The contributions of the σ- and π-orbitals to the NICS values of 1’ and 2’S are analyzed. The results reveal that the NICS(0)_ZZ of 3MR in 1’ is dominated by the contributions from σ orbitals, whereas the NICS(1)_ZZ of 5MR in 1’ and 6MR in 2’S are mainly contributed by π orbitals (the selected orbitals are shown in Figs. S5–6).

Fig. 7. DFT calculations.

a The computed NICS values (in ppm) of 1’, 2’S, and 2’T; b AICD plots of 1’ with isosurface value of 0.025 a.u.; c AICD plots of 2’S with isosurface value of 0.025 a.u.; d Aromatic stabilization energies of 1’, 1”, 2”S, and 2”T (in kcal/mol). [Os]’ = OsCl₂(PH₃)₂.

The anisotropy of the induced current density (AICD)^69,70 was investigated to visually demonstrate the aromaticity. As shown in Figs. 7b, c and S6–12, regular clockwise diatropic ring currents are observed for 3MR and 5MR in 1’, as well as 6MR in 2’S and 2’T, supporting their aromatic property. To gain a more detailed understanding, we separated the AICD plots into σ-orbital and π-orbital contributions. For 1’, the AICD-σ shows the diatropic ring currents in 3MR, while 5MR exhibits such ring currents in AICD-π. In 2’S, no distinct ring currents is observed throughout the molecule in the σ-orbital contributions, whereas only the AICD-π demonstrates regular diatropic ring currents. We can conclude that the aromaticity of 3MR in 1’ primarily originates from σ orbital contributions, and that of 5MR mainly stems from π orbital contributions. While in 2’S, 6MR is π aromatic. Therefore, the reshaping of 1 to 2 represents the first aromaticity switching of (π + σ) aromatics to π aromatics. Isodesmic reactions were used to further support the aromaticity of 3MR in 1’, 5MR in 1”, 6MRs in 2”S and 2”T^71–73. As depicted in Fig. 7d, the aromatic stabilization energies (ASE) are all positive (24.7, 49.6, 22.9, and 13.9 kcal/mol), indicating the aromaticity in 3MR, 5MR, and 6MRs, as the breakage of aromatic systems are highly thermodynamically unfavorable.

From the perspective of electron delocalization, the multicenter bond index (MCI) is analyzed^74,75. The MCI values of 3MR and 5MR in 1’ are 0.435 and 0.337, respectively, while that of 6MR in 2’S is 0.426, suggesting that all of these metallacycles are aromatic (Fig. S3). The reported MCI values of the metallabenzenes of some early transition metals vary from 0.370 to 0.504⁷⁶. Furthermore, the normalized multicenter bond order (MCBO) has been studied^77–79, and the values are 0.360 and 0.437 for 3MR and 5MR in 1’, respectively, and those of 6MRs in 2’S and 2’T are 0.512 and 0.414, respectively, indicating their aromaticity (Fig. S3). For comparison, the normalized MCBO values of the aromatic four-membered rings in binuclear Cr complexes are between 0.438 and 0.472, and that of benzene is 0.665⁸⁰.

In summary, novel cyclopropametallaindole complexes are synthesized, and their skeletal rearrangement to metallaquinoliniums are demonstrated. The unique electrophilic addition and elimination mechanism and the aromaticity of these structures have been studied by DFT calculations. This transformation achieves the reshaping of one metallaaromatic framework to another, as well as (π + σ) aromatic bicycle to π aromatic monocycle. Our work provides new insights into aromatic chemistry and organometallic chemistry.

Methods

General information

Details of the synthesis and characterization of metallacycles (1a–d and 2a–c) can be found in the Supplementary Information, pp. 1–3. For X-ray data of all the described complexes (1b, 2a, and 2c), see supplementary, pp. 10–11. For ¹H, ³¹P{¹H} NMR, and ¹³C{¹H} NMR, EPR, HRMS, and UV–Vis absorption spectra of complexes in this article, see Supplementary Information, pp 12–25 and Supplementary Figs. S15–45.

Computational details

The optimizations were performed with the Gaussian 16 Rev. A.03 software package⁸¹ at the B3LYP level of density functional theory (DFT)^45–47. The def2-TZVP basis set was implemented for the Os atom at the B3LYP DFT level^82,83. The 6-311G(2d,p) basis set had been used for the rest of atoms⁴⁸. Nucleus-independent chemical shift (NICS) values and anisotropy of the current density (AICD) were calculated at the B3LYP//6-311G(2d,p)/def2-TZVP level^65–70. The wave function analysis was performed by the Multiwfn software^84,85. In mechanistic studies, all geometry optimizations of complexes were performed using the Gaussian 16 package⁸¹, at the B3LYP level of density functional theory (DFT). The Grimme D3(BJ) empirical dispersion correction was applied to account for dispersion effects. For the Os atom, the def2-TZVP basis set together with the corresponding Stuttgart relativistic energy-consistent pseudopotential was used^82,83, and calculations for C, N, P, O, Cl, and H atoms were carried out using the 6-311G(2d,p) basis set ⁴⁸. Frequency calculations at the same level of theory were performed to identify the number of imaginary frequencies n (n = 0 for local minima and n = 1 for transition states) and provide the thermal corrections to Gibbs free energies. In addition, intrinsic reaction coordinate (IRC) calculations were performed to confirm the assignment of transition states. Furthermore, all calculations were performed with the SMD implicit solvent model, and dichloromethane was employed as the solvent in mechanistic studies. The Calculated Cartesian Coordinates of complexes in this article see the file “Calculated Cartesian Coordinates.xyz”.

X-ray crystallographic study

Single crystals of complexes suitable for X-ray diffraction were grown from CDCl₃ solution layered with n-hexane for 1b, and CH₂Cl₂ solution layered with toluene and n-hexane for 2a and 2c. Intensity data of 1b (CCDC No. 2424053), 2a (CCDC No. 2493840), 2c (CCDC No. 2503668) were collected on a Bruker D8 Venture diffractometer at 127.0 K, 120.0 K, and 128.0 K using Mo-Kα radiation (λ = 0.71073 Å). Unit cell indexing was refined using SAINT, Absorption correction was applied by using multi-scan program SADABS. The structure was solved with OLEX2 software, and the SHELXT structure solution program using combined direct method^86–88. The crystal structure was refined by least squares using SHELXL. All non-hydrogen atoms were refined anisotropically. Hydrogen atoms bonded to carbon atoms were placed at calculated positions and refined using a riding model approximation, with C–H = 0.95(aromatic CH) and with Uiso(H) = 1.2 Ueq(C), C–H = 1.00(–CH) and with Uiso(H) = 1.2 Ueq(C), C–H = 0.99(-CH₂) and with Uiso(H) = 1.2 Ueq(C), C–H = 0.98 Å (–CH₃) and with Uiso(H) = 1.5 Ueq(C). The crystal data are listed in pp. 11 and supplementary Table S2.

Author contributions

W.B. and W.J. conceived the project. B.F. performed the experiments and conducted theoretical computations. Y.Z. solved the X-ray structures. W.B., B.F., and Y.L. drafted the paper. All authors discussed the results and contributed to the preparation of the manuscript.

Peer review

Peer review information

Nature Communications thanks Fabio Marchetti and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This research is supported by the National Natural Science Foundation of China (Grant No. 22001030) and the Liaoning Provincial Natural Science Foundation (Grant No. 2025-MS-001).

Data availability

Source Data are provided with this manuscript. All data that support the findings of this study are available within the paper, and its supplementary information files. Crystallographic data for the structures reported in this paper have been deposited at the Cambridge Crystallographic Data Center, under the deposition numbers CCDC 2424053 (1b), 2493840 (2a), 2503668 (2c). Copies of these data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif. All data are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74826-y.