The anti-aromatic dianion and aromatic tetraanion of [18]annulene

[18]Annulene (C18H18) has long served as a benchmark for testing concepts of aromaticity in large monocyclic π-systems. Hückel’s 4n + 2 rule predicts aromaticity for 22 π electrons and anti-aromaticity for 20 π electrons. A 1973 study claimed that [18]annulene could be reduced to an anti-aromatic dianion, proposing specific structural assignments from low-field 1H NMR. This work re-examines the dianion using modern high-field NMR methods, corrects its structural assignment, and demonstrates that [18]annulene can be further reduced to a stable, aromatic tetraanion. The study addresses how reduction changes conformation and aromatic character, and how these states interact with alkali metal cations.

Sondheimer’s seminal synthesis and studies in the 1960s extended Hückel aromaticity beyond benzene, including [18]annulene. Oth, Woo and Sondheimer (1973) reported formation of the [18]annulene dianion and proposed an NMR-based structural model from 60 MHz spectra at low temperature. Subsequent computational and crystallographic work on neutral [18]annulene debated bond-length alternation and dynamics, suggesting rapid interconversion among bond-localized structures despite an average high symmetry. Corannulene tetraanions are known to form lithium multi-decker “jammed” sandwiches with intercalated Li+, motivating exploration of related sandwich complexes from [18]annulene polyanions. Theoretical proposals also considered stacked annulene columns upon reduction, though often assuming the neutral conformation persisted.

Synthesis: [18]Annulene was prepared following literature routes with modifications (Supplementary Information Section 2). Glovebox and Schlenk techniques were used to exclude air and moisture for all reductions. Reductions and NMR: Reductions were performed in THF-d8. With K metal, [18]annulene afforded a green solution of the radical anion then the dianion 1-K2. With Li metal, sequential reductions provided the monoanion, dianion, and the tetraanion 1-4−. 1H NMR spectra were collected at 500 MHz, typically at −70 °C for the dianion and 25 °C for the tetraanion. Spectral assignments used 1H–13C HSQC, 1H–1H COSY, NOE/NOESY, and 2D exchange spectroscopy (HEXSY) to probe dynamics. 7Li NMR was recorded (194 MHz) at low temperatures (−60 to −80 °C) to resolve intercalated versus external Li+ environments. Crystallography: Single crystals of the lithium tetraanion salt were grown by layering hexanes over THF solutions. X-ray diffraction determined molecular geometry, Li+ positions (intercalated vs. external), inter-ring distances, and bond metrics. Computations: DFT optimizations employed BLYPxx functionals (37.5–54% exact exchange; especially BLYP45) with the def2-TZVP basis. High-accuracy single-point energies used DLPNO-CCSD(T*)-F12 with the cc-pVDZ-F12 basis. Conformers (A, B, C) of neutral, dianion, and tetraanion states were assessed. Calculated NMR chemical shifts and bond-length alternation (BLA) were compared to experiment.

• High-field 1H NMR of the dianion (THF-d8, −70 °C, 500 MHz) shows ten CH multiplets consistent with C2 symmetry: five inner protons strongly deshielded at δ ~ 27.97–29.47 ppm (2H, 2H, 1H) and thirteen outer protons shielded at δ ~ −0.50 to −1.54 ppm (various integrals), indicating five inner and 13 outer hydrogens.
• The lithium tetraanion 1-4− displays the same symmetry but reversed ring-current effects: inner five protons become strongly shielded at δ ≈ −8 to −9 ppm, confirming aromaticity with 22 π electrons; outer 13 protons shift oppositely relative to the dianion.
• Temperature dependence: The tetraanion 1H NMR is essentially temperature independent from −40 to 55 °C. The dianion broadens above −50 °C due to dynamic exchange. 2D exchange NMR shows rotation of two CH units (f and g) that switches the symmetry plane.
• Stability: Contrary to earlier reports, 1-2− solutions (in dry THF) are stable up to ~40 °C and for weeks at room temperature when sealed; exposure to O2 regenerates neutral [18]annulene.
• 7Li NMR: Tetraanion in THF-d8 (−80 °C) shows two lithium environments: δ7Li = −15.66 ppm (sharp; intercalated Li+ between aromatic decks) and −2.0 ppm (broad; external Li+). The dianion shows a single broad δ7Li ≈ 2.43 ppm, indicating weaker Li+ interaction.
• X-ray crystal structure (lithium tetraanion salt): A metallocene-like homo-sandwich forms with two [18]annulene tetraanions and five intercalated Li+ plus three external Li+(THF)3; centroid–centroid distance 3.8926(6) Å; relative rotation of C2 axes ~74°. C–C bond lengths average 1.411 Å with no significant BLA (<0.02 Å), consistent with aromaticity.
• Heteroleptic sandwich with corannulene: An equimolar mixture of [18]annulene and corannulene reduced with Li in THF-d8 yields a 1:1 heteroleptic sandwich (1–2–Li5) confirmed by X-ray. Inner [18]annulene protons shift further upfield to δ = −12.50 and −10.92 ppm due to shielding by the corannulene bowl. 7Li NMR shows three highly shielded Li environments at −17.92, −18.81 and −19.12 ppm (integrations 2:1:2).
• Theory: DLPNO-CCSD(T*)-F12 energies show that neutral [18]annulene prefers the high-symmetry A conformer (virtually D6h), while the dianion strongly prefers the lower-symmetry B conformer (five inner/13 outer H) by ~46 kJ mol−1 over A2−. The switch is driven by π-orbital stabilization for higher angular momentum modes and reduced Coulomb repulsion in B2−. BLYP45 reproduces experimental chemical shifts and predicts C2 symmetry; significant BLA in 1B2− (~0.06 Å) but none in 1B4− (<0.01 Å).

Modern high-field NMR and crystallography overturn the longstanding structural assignment of the [18]annulene dianion. The NMR pattern of five inner and 13 outer hydrogens, along with temperature-dependent exchange behavior, establishes a C2-symmetric anti-aromatic dianion distinct from earlier proposals. The discovery of a stable aromatic tetraanion that forms a lithium-intercalated sandwich demonstrates that additional reduction both reverses ring currents and stabilizes a metallocene-like architecture. 7Li NMR correlates Li+ positions with shielding by aromatic decks, tying electronic structure to ion coordination. Computations rationalize the conformational switch upon reduction: neutral [18]annulene favors a high-symmetry geometry, but electron addition stabilizes localized π orbitals and reduces Coulomb repulsion in the less symmetric B conformer, aligning with experimental NMR and the observed lack of BLA in the tetraanion. The work clarifies anti-aromatic versus aromatic behavior in a prototypical non-benzenoid ring and links redox state, conformation, and ion intercalation.

The widely accepted geometry of the [18]annulene dianion is incorrect; instead, reduction induces a substantial conformational change to a C2-symmetric anti-aromatic form with five inner hydrogens. Further reduction yields a stable aromatic tetraanion that assembles into lithium-intercalated homo- and heteroleptic sandwich complexes. These results, supported by high-field NMR, single-crystal X-ray structures, and high-level quantum calculations, highlight how electron count governs global conformation and aromaticity in macrocycles. The ability to reversibly access anti-aromatic and aromatic states and to intercalate multiple Li+ ions suggests potential uses in energy storage and molecular electronics. Future work could target: direct structural characterization of the dianion in the solid state; exploration of different counterions and solvents to tune sandwich assembly; extending the concept to other annulenes and macrocycles; probing electrochemical cycling and ion transport in bulk materials; and refining computational models for anti-aromatic systems.

The dianion structure is inferred from solution NMR and computation; no solid-state crystal structure of the dianion was reported. Many NMR characterizations require low temperatures due to dynamic exchange, potentially complicating spectral interpretation. The tetraanion is observed crystallographically only as lithium-bound sandwich dimers; behavior with other cations or in the absence of sandwich assembly remains to be established. Computational conclusions depend on functional choice and basis sets, although corroborated by high-level DLPNO-CCSD(T*)-F12 single points. Stability requires rigorously anhydrous, oxygen-free conditions.