Conical Intersection

A conical intersection (CI) is a point—or, more precisely, an (N_int − 2)-dimensional seam—in the nuclear coordinate space of a polyatomic molecule at which two adiabatic electronic potential energy surfaces (PES) of the same spin multiplicity become exactly degenerate, touching each other like the apexes of an upright and an inverted cone when plotted against the two symmetry-breaking nuclear coordinates that lift the degeneracy. First analyzed mathematically by John von Neumann and Eugene Wigner in 1929 and connected to molecular photochemistry by Edward Teller in 1937, H. Christopher Longuet-Higgins in 1958–1975, and Fernando Bernardi, Michael Robb, and Wolfgang Domcke in the 1990s, conical intersections serve as the quantum-mechanical "photochemical funnels" of chemistry and biochemistry.

In a diatomic molecule (which has only N_int = 1 internal internuclear distance coordinate R), the von Neumann–Wigner non-crossing rule forbids two electronic states of identical symmetry from intersecting, producing an avoided crossing instead. In any non-linear polyatomic molecule with N ≥ 3 atoms (N_int = 3N − 6 ≥ 3 internal coordinates), however, two independent algebraic conditions for degeneracy can be satisfied simultaneously across a continuous (N_int − 2)-dimensional seam. At the apex of a conical intersection, the energy gap ΔE vanishes, the Born–Oppenheimer approximation breaks down completely, and the non-adiabatic vibronic coupling diverges, allowing photoexcited molecules to convert electronic excitation energy into nuclear vibrational energy on an ultrafast femtosecond timescale (10 to 200 fs)—millions of times faster than fluorescence or intersystem crossing.

Interactive VSEPR Lab: Inspect 3D molecular symmetry axes and bending coordinates across polyatomic molecules in the VSEPR Lab!

Two Conditions for Degeneracy and the (3N − 8)-Dimensional Seam

Consider two interacting electronic states in a diatomic or polyatomic molecule represented in a crude diabatic basis {|Φ₁⟩, |Φ₂⟩} by the 2 × 2 electronic Hamiltonian matrix H(Q), with diagonal diabatic energies H₁₁(Q) and H₂₂(Q) and off-diagonal coupling H₁₂(Q) = H₂₁(Q). Diagonalizing H(Q) gives the upper and lower adiabatic potential energy surfaces E_±(Q):

E_±(Q) = ½ [ H₁₁(Q) + H₂₂(Q) ] ± √{ ¼ [ H₁₁(Q) − H₂₂(Q) ]² + H₁₂(Q)² }

For the two adiabatic surfaces to become exactly degenerate (E₊ = E₋) at a nuclear configuration Q_CI, the sum of two non-negative squares inside the square root must vanish, imposing two independent conditions:

1. Condition 1 (Diagonal Energy Matching): H₁₁(Q_CI) − H₂₂(Q_CI) = 0
2. Condition 2 (Off-Diagonal Coupling Vanishing): H₁₂(Q_CI) = 0

Expanding the matrix elements linearly in small nuclear displacements δQ = Q − Q_CI around the intersection point defines two fundamental vectors in nuclear coordinate space that span the 2D branching plane (or g–h plane):

• The Gradient Difference Vector (g): g = ½ ∇_Q [ H₁₁(Q) − H₂₂(Q) ]_QCI (typically a bond-stretching or skeletal-breathing coordinate that tunes the energy gap between the two diabatic configurations).
• The Non-Adiabatic Derivative Coupling Vector (h): h = ∇_Q H₁₂(Q)_QCI = ⟨Ψ₁ | ∇_Q H_el | Ψ₂⟩ (typically an asymmetric twisting, bending, or bond-alternation coordinate that mixes the two electronic states).

When the adiabatic energies E_±(x, y) are plotted over the 2D branching plane spanned by (x along g, y along h), they form a double cone:

E_±(x, y) = E_CI + s_x x + s_y y ± √( g² x² + h² y² )

Whereas moving in any direction within the 2D (g, h) branching plane lifts the degeneracy linearly with distance from the apex, moving along any of the remaining N_int − 2 = 3N − 8 orthogonal coordinates (the intersection space or seam space) preserves the exact degeneracy to first order. Depending on the tilt parameters (s_x, s_y), a conical intersection is classified as peaked (when both upper and lower cones slope inward toward the apex, funneling wavepackets directly into the seam) or sloped (when both sheets tilt steeply to one side, enabling recrossing).

Geometric (Berry) Phase and the Longuet-Higgins Sign Reversal

A profound quantum-topological property of every conical intersection—discovered by H. Christopher Longuet-Higgins and Gerhard Herzberg in 1958–1963 and generalized to all adiabatic quantum systems by Michael Berry in 1984—is the geometric phase (Berry phase) of γ = π.

In polar coordinates (ρ, θ) within the 2D branching plane around the conical intersection (g x = ρ cos θ, h y = ρ sin θ), the real adiabatic electronic wavefunctions |Ψ₋(θ)⟩ and |Ψ₊(θ)⟩ depend on the polar angle θ through half-angles:

|Ψ₋(θ)⟩ = − sin(θ / 2) |Φ₁⟩ + cos(θ / 2) |Φ₂⟩

If the nuclei are transported adiabatically around a closed loop in the branching plane that encircles the conical intersection once (θ: 0 → 2π) without ever passing through the apex (ρ > 0), the half-angle θ/2 advances from 0 to π, causing both real adiabatic electronic wavefunctions to reverse their algebraic sign:

|Ψ_±(θ + 2π)⟩ = − |Ψ_±(θ)⟩ = e^iπ |Ψ_±(θ)⟩

Because the total molecular wavefunction Ψ_total(r, Q) = Ψ_el(r; Q) χ_nuc(Q) must be strictly single-valued when the nuclei return to their starting coordinates (θ → θ + 2π), the nuclear vibrational wavefunction χ_nuc(ρ, θ) must also reverse sign upon completing a 2π circuit around the conical intersection (χ_nuc(ρ, θ + 2π) = −χ_nuc(ρ, θ)). This topological boundary condition forces the pseudorotation quantum number j in the Mexican hat trough of a Jahn–Teller molecule (such as the sodium trimer Na₃ or Cu₃) to take half-odd-integer values (j = ±½, ±3/2, ±5/2…) rather than integers (j = 0, ±1, ±2…), shifting the vibronic energy spectrum and creating a topological node on the far side of the cone in reactive molecular scattering (such as the H + H₂ → H₂ + H exchange reaction at its D_3h conical intersection).

Primary Photochemistry of Vision: 11-cis to all-trans Retinal Isomerization in Rhodopsin

In vertebrate vision, the capture of a single visible photon (λ_max ≈ 500 nm) by the G-protein-coupled photoreceptor rhodopsin in retinal rod cells triggers the photoisomerization of its covalently bound chromophore, the 11-cis retinal protonated Schiff base (PSB11), into the all-trans photoproduct (bathorhodopsin) with a quantum yield of Φ = 0.67. Remarkably, this large-amplitude torsional isomerization around the C₁₁=C₁₂ double bond completes in less than 200 femtoseconds (2 × 10⁻¹³ s)—one of the fastest and most coherent chemical reactions in biology:

1. Upon vertical excitation from the S₀ ground state (covalent configuration) to the bright spectroscopic S₁ (1¹B_u-like charge-transfer) excited state, positive charge translocates along the conjugated polyene chain toward the β-ionone ring, inverting the single/double bond-length alternation (BLA).
2. Within the first ~20–30 fs, high-frequency C=C/C–C bond-stretching vibrations (the gradient difference vector g) propel the nuclear wavepacket out of the Franck–Condon region, followed immediately by barrierless torsional twisting around the C₁₁=C₁₂ bond (angle φ ≈ 0° → ~90°, contributing to the coupling vector h) coupled with hydrogen-out-of-plane (HOOP) wagging.
3. At a twisted geometry near φ ≈ 90°, the S₁ potential surface meets the S₀ ground-state surface at a peaked S₁/S₀ conical intersection. The nuclear wavepacket drops through this conical funnel onto the S₀ surface within ~80–110 fs, conserving vibrational phase coherence all the way into the all-trans ground-state photoproduct well and initiating the visual transduction cascade.

Intrinsic UV Photoprotection of DNA and RNA Nucleobases

Solar ultraviolet-B and ultraviolet-C radiation (200–300 nm) is strongly absorbed by the aromatic purine and pyrimidine rings of DNA and RNA nucleobases (adenine, guanine, cytosine, thymine, and uracil), populating bright ¹ππESCAST excited states carrying ~4.5 to 5.0 eV of electronic energy—more than enough to cleave covalent bonds or cross-link adjacent bases into mutagenic cyclobutane pyrimidine dimers (CPDs). Yet monomeric canonical nucleobases in aqueous solution exhibit exceptionally low fluorescence quantum yields (Φ_F ~ 10⁻⁴) and decay back to the electronic ground state (S_₀) without chemical damage in 0.2 to 1.0 picoseconds (200–1,000 fs).

Femtosecond transient absorption and ab initio non-adiabatic dynamics simulations have established that this natural photostability of the genetic code is governed by barrierless access to ring-puckering S₁(ππESCAST)/S_₀ conical intersections:

• In adenine (9H-adenine), photoexcitation to the bright ¹L_a (ππ*) state drives ultrafast out-of-plane puckering at the C₂–H (ethylenic twist) or C₆–NH₂ positions of the six-membered pyrimidine ring, reaching an S₁/S₀ conical intersection in < 100 fs and dissipating the 4.6 eV photon energy harmlessly into surrounding water molecules as vibrational heat before triplet states or radical pathways can form.
• In pyrimidine bases (thymine, uracil, cytosine), out-of-plane twisting and pyramidalization of the C₅=C₆ ethylenic bond (analogous to the prototype twisted-pyramidalized conical intersection of ethylene) funnels > 99% of the excited population back to S₀ on a sub-picosecond timescale. By contrast, non-canonical synthetic nucleobase analogues (such as 2-aminopurine, an isomer of adenine where the amino group shifts from C₆ to C₂) face a significant energy barrier (~0.3–0.4 eV) on the path to the conical intersection, lengthening their excited-state lifetime by 10,000-fold (to ~10 nanoseconds) and rendering them strongly fluorescent and photochemically reactive.

Why does the Born–Oppenheimer approximation fail completely at a conical intersection?

In the standard Born–Oppenheimer approximation, the non-adiabatic derivative coupling vector d₁₂(Q) = ⟨Ψ₁ | ∇_Q | Ψ₂⟩ between two adiabatic electronic states is neglected because, by the Hellmann–Feynman theorem, it is inversely proportional to the adiabatic energy gap between the surfaces:

d₁₂(Q) = ⟨Ψ₁ | ∇_Q H_el | Ψ₂⟩ / [ E₂(Q) − E₁(Q) ] = h / ΔE(Q)

When two electronic states are well separated in energy (ΔE ≈ several eV), d₁₂ is tiny and nuclei move smoothly on a single potential energy surface. At a conical intersection, however, the denominator ΔE(Q_CI) → 0 while the numerator h remains finite, causing the non-adiabatic coupling d₁₂ to diverge to infinity (a mathematical singularity). As a nuclear wavepacket approaches the apex of the cone, the probability of jumping between the upper and lower surfaces approaches unity inside a single vibrational period (~10–50 fs).

Why can polyatomic molecules have conical intersections while diatomic molecules cannot?

According to the von Neumann–Wigner theorem (1929), forcing two real symmetric electronic states to be degenerate requires satisfying two independent algebraic equations simultaneously: the diagonal energies must be equal (H₁₁ = H₂₂) and the off-diagonal coupling must vanish (H₁₂ = 0). A diatomic molecule has only one internal degree of freedom (the bond length R). Unless the two states belong to different spatial or spin symmetries (where H₁₂ = 0 automatically by symmetry), varying a single parameter R cannot simultaneously satisfy two independent equations, resulting in an avoided crossing. A non-linear polyatomic molecule with N ≥ 3 atoms possesses 3N − 6 ≥ 3 internal coordinates, providing at least two independent degrees of freedom (the g and h vectors) to satisfy both conditions at once, leaving a continuous (3N − 8)-dimensional seam of true conical intersections.

How is the Jahn–Teller effect related to conical intersections?

Every first-order Jahn–Teller system—such as an E ⊗ e trigonal molecule (Na₃, Cu₃) or octahedral complex ([Cu(H₂O)₆]²⁺)—possesses a symmetry-required conical intersection right at its high-symmetry reference geometry (Q = 0). In fact, the central peak at the origin of the classic Jahn–Teller Mexican hat (sombrero) potential is the exact prototype of a peaked conical intersection, where point-group symmetry guarantees both H₁₁(0) = H₂₂(0) and H₁₂(0) = 0, and the two degenerate normal vibrational modes (Q_θ, Q_ε) form the orthogonal g and h vectors of the 2D branching plane. Modern photochemistry simply generalizes this Jahn–Teller cone from symmetry-enforced origins to accidental conical intersections occurring at distorted, low-symmetry geometries across excited states.

1. Teller, E. (1937). The crossing of potential surfaces. The Journal of Physical Chemistry, 41(1), 109–116. https://doi.org/10.1021/j150379a010
2. Yarkony, D. R. (1996). Diabolical conical intersections. Reviews of Modern Physics, 68(4), 985–1013. https://doi.org/10.1103/RevModPhys.68.985
3. Domcke, W., Yarkony, D. R., & Köppel, H. (Eds.). (2004). Conical Intersections: Electronic Structure, Dynamics & Spectroscopy. World Scientific. https://doi.org/10.1142/5406
4. Polli, D., Altoè, P., Weingart, O., Spillane, K. M., Manzoni, C., Brida, D., Tomasello, G., Orlandi, G., Kukura, P., Mathies, R. A., Garavelli, M., & Cerullo, G. (2010). Conical intersection dynamics of the primary photoisomerization event in vision. Nature, 467(7314), 440–443. https://doi.org/10.1038/nature09346