Rovibronic Coupling and Nonadiabatic Effects

Molecular spectroscopy cashes in one result from quantum mechanics: solve the electronic Schrödinger equation at frozen nuclei to get a potential energy surface, then quantize nuclear vibration and rotation on it. That two-step recipe is the Born-Oppenheimer approximation, and it works spectacularly for the ground states of stable molecules because electronic energy gaps dwarf vibrational and rotational quanta. But the recipe throws something away: the nuclear kinetic-energy operator does not only act on the nuclear wavefunction, it also differentiates the electronic states, which depend parametrically on where the nuclei sit. Those discarded terms are the nonadiabatic couplings, and they knit electronic states back together through nuclear motion. The coupling scales as the inverse of the electronic energy gap, so it is invisible deep in a molecular well and explosive wherever two surfaces approach. Same-symmetry surfaces forbidden to cross bunch into an avoided crossing; surfaces of different symmetry, or a polyatomic in more than one nuclear dimension, can touch exactly at a conical intersection, where electrons can no longer follow the nuclei and the system leaps between surfaces with no photon involved. The fingerprints crowd real spectra: bound levels above a dissociation limit broaden into diffuse bands (predissociation), regular term series buckle where a dark state crosses (perturbations), forbidden bands borrow intensity, and orbital angular momentum coupled to rotation splits levels into Λ-doublets — and this article traces every one of them back to a single operator.

Key takeaways

The Born-Oppenheimer separation expands the molecular wavefunction in adiabatic electronic states; the neglected nuclear-momentum terms that differentiate those states are the nonadiabatic couplings.
The first-derivative coupling equals a smooth Hellmann-Feynman element over the electronic gap, so it diverges wherever two adiabatic surfaces approach or touch.
A conical intersection is a genuine degeneracy that, for a nonlinear molecule, forms a double cone in a two-dimensional branching space — an ultrafast radiationless funnel carrying a geometric phase of π.
Nonadiabatic coupling is measurable: predissociation and line broadening, perturbed term series, intensity borrowing, and Λ-doubling all read out its strength, and the Landau-Zener formula turns a crossing into a hop probability.
Key idea: nonadiabatic coupling is the nuclear kinetic-energy operator acting on electronic wavefunctions; it goes as ⟨φᵢ|∇ᵣφⱼ⟩ over the electronic gap, so it governs the physics exactly where adiabatic surfaces meet.

The core insight

Nonadiabatic coupling is the action of the nuclear kinetic-energy operator on electronic wavefunctions; it is proportional to $\langle \phi_i | \nabla_R \phi_j\rangle$ and inversely proportional to the electronic energy gap, so it dominates the physics precisely where adiabatic surfaces approach or intersect.

Where this lives today

Nonadiabatic dynamics is the heart of ultrafast science. Femtosecond pump-probe spectroscopy — opened by Ahmed Zewail with the sodium-iodide avoided crossing, watching a wavepacket oscillate and leak across the gap in real time — now routinely clocks wavepackets funneling through conical intersections in molecules from retinal to DNA bases, where the intersection lets a nucleobase dump ultraviolet energy as heat in under a picosecond instead of undergoing damaging photochemistry. Surface-hopping and multi-configuration time-dependent Hartree methods propagate the nuclear wavepackets across coupled surfaces using exactly the diabatic couplings and Landau-Zener logic developed here, while attosecond experiments probe the breakdown of Born-Oppenheimer directly by resolving electron and nuclear motion on their own timescales, and cold-molecule experiments invert the Landau-Zener formula with chirped-field sweeps to prepare specific states. What Herzberg extracted from diffuse bands and buckled term plots, today's instruments watch happen frame by frame.

The physical picture

Picture the nuclei as slow, heavy dancers and the electrons as a light, fast cloud that reshapes instantly to whatever geometry it is handed — that instantaneous reshaping is the adiabatic picture, and it is why a molecule has clean potential energy curves at all. Nonadiabatic coupling is what happens when the nuclei move so fast, or the electron cloud has two nearly equal ways to arrange itself, that the cloud cannot keep up and flips from one arrangement to another. The energy gap between surfaces acts as a stiffness. A large gap locks the cloud rigidly to one configuration, so no nuclear jostling can shake it loose and Born-Oppenheimer holds; shrink the gap and the cloud softens, until at zero gap an infinitesimal nuclear step flips the system between two degenerate configurations. That is the physical meaning of the energy denominator: every nonadiabatic quantity blows up as the gap closes, as the first figure shows — same-symmetry adiabatic curves repelling into an avoided crossing over the smooth diabatic curves beneath.

Figure

Two smooth diabatic curves cross at Rc but the same-symmetry adiabatic surfaces repel into an avoided crossing whose minimum separation is exactly 2V₁₂ = 0.20 eV ≈ 1600 cm⁻¹

A conical intersection is the more violent cousin of the avoided crossing. The upper surface acts as a funnel: a wavepacket slides toward the tip of a cone and drops onto the lower surface within a single vibrational period, before it can fluoresce. Because the derivative coupling diverges at the tip, the passage is near-unit-efficient, which is why conical intersections dominate radiationless decay in photochemistry and photobiology. Rotation adds one more layer: as the molecule tumbles, the Coriolis interaction mixes electronic and vibrational angular momenta, so the electronic, vibrational, and rotational ladders become one entangled rovibronic problem.

Figure

At a conical intersection two adiabatic surfaces touch at a point and split linearly along the branching coordinates Qx and Qy, forming a double cone; a wavepacket funnels from upper to lower surface in under 100 fs and picks up a geometric phase of π on any loop enclosing the tip

The core equations

The molecular Hamiltonian splits into a nuclear kinetic term and an electronic Hamiltonian that depends on nuclear positions only as fixed parameters. Diagonalizing the electronic part at each geometry defines the adiabatic states and their surfaces.

\hat H_{\text{mol}} = \hat T_N + \hat H_{el}(r;R),\qquad \hat H_{el}\,\phi_j(r;R) = E_j(R)\,\phi_j(r;R) \qquad (1)

where: $\hat T_N=-\sum_A \hbar^2/(2M_A)\,\nabla_{R_A}^2$ — nuclear kinetic operator; $\hat H_{el}(r;R)$ — electronic Hamiltonian at frozen nuclei; $\phi_j(r;R)$ — adiabatic state ( $r$ electronic, $R$ nuclear); $E_j(R)$ — potential energy surface. The parametric semicolon is the whole story: the electronic states carry hidden $R$ -dependence that $\hat T_N$ differentiates. Because the adiabatic states are a complete electronic basis at every geometry, the exact molecular wavefunction expands in them with nuclear coefficients. Keeping one term is Born-Oppenheimer; keeping all of them is exact and is where the coupling is born.

\Psi(r,R) = \sum_j \chi_j(R)\,\phi_j(r;R) \qquad (2)

where: $\chi_j(R)$ — nuclear (vibration-plus-rotation) wavefunction on surface $j$ ; the sum runs over the complete set of adiabatic states. The $\chi_j$ are the unknowns; the $\phi_j$ come from the electronic-structure calculation. Substituting the expansion and projecting onto one electronic state gives nuclear equations that are no longer independent — each surface's motion is driven by the others through a coupling operator.

\left[\hat T_N + E_i(R) - E\right]\chi_i = \sum_{j}\hat\Lambda_{ij}\,\chi_j \qquad (3)

where: $\hat\Lambda_{ij}$ — nonadiabatic coupling operator linking surfaces $i$ and $j$ . When every $\hat\Lambda_{ij}$ is negligible the equations decouple into independent single-surface problems. The dominant piece is the first-derivative coupling. Differentiating the electronic eigenvalue equation and using the Hellmann-Feynman theorem recasts it as a smooth numerator over the gap.

\mathbf F_{ij}(R) = \langle \phi_i|\nabla_R\phi_j\rangle = \frac{\langle \phi_i|\nabla_R\hat H_{el}|\phi_j\rangle}{E_j(R)-E_i(R)}\quad(i\ne j) \qquad (4)

where: $\mathbf F_{ij}$ — derivative coupling vector; the numerator is a smooth Hellmann-Feynman element; $E_j-E_i$ — adiabatic gap. The divergence at a crossing is entirely the fault of the vanishing denominator, the signature that the adiabatic basis has failed. Near a crossing only two states matter, so the electronic problem collapses to a real symmetric two-by-two matrix in a diabatic basis. Its eigenvalues are the two adiabatic surfaces, and their form encodes the noncrossing rule.

E_\pm(R) = \bar V \pm \sqrt{\left(\tfrac{V_{11}-V_{22}}{2}\right)^2 + V_{12}^2} \qquad (5)

where: $E_\pm$ — upper and lower adiabatic energies; $\bar V=(V_{11}+V_{22})/2$ ; $V_{11},V_{22}$ — smooth diabatic energies that may cross; $V_{12}$ — diabatic coupling. The square root vanishes only if both $V_{11}=V_{22}$ and $V_{12}=0$ , and where $V_{11}=V_{22}$ the surfaces are split by exactly $2|V_{12}|$ . When a wavepacket sweeps through the crossing at finite speed, whether it stays on one adiabatic surface or hops is fixed by the Landau-Zener probability — the exact solution of the linearized sweep.

P_{\text{diab}} = \exp\!\left(-\frac{2\pi V_{12}^2}{\hbar\, v\,\left|\Delta F\right|}\right),\qquad \Delta F=\left.\frac{\mathrm d(V_{11}-V_{22})}{\mathrm dR}\right|_{R_c} \qquad (6)

where: $P_{\text{diab}}$ — probability of staying diabatic (changing electronic state); $v$ — nuclear speed; $\Delta F$ — diabatic slope difference. Fast passage or weak coupling drives $P_{\text{diab}}\to1$ ; slow passage or strong coupling drives it to zero, keeping the wavepacket on one adiabatic surface. Away from crossings, the rovibronic energy stacks electronic, vibrational, and rotational terms, and residual electronic-rotational coupling lifts the degeneracy of orbitally degenerate states.

E(v,J) \approx T_e + \omega_e\!\left(v+\tfrac12\right) + B_v J(J+1),\qquad \Delta E_{\Lambda} \propto B_v\, J(J+1) \qquad (7)

where: $T_e$ — electronic term energy; $\omega_e$ — harmonic constant; $B_v$ — rotational constant; $\Delta E_\Lambda$ — Λ-doubling of a $\Lambda\ne0$ state, growing with rotation. The additive form is the Born-Oppenheimer idealization and $\Delta E_\Lambda$ is its first crack. Finally, a bound level pushed above the dissociation limit of a coupled repulsive state leaks into the continuum, and Fermi's golden rule turns the coupling into a measurable width.

\Gamma = \frac{\hbar}{\tau} \approx 2\pi\,\bigl|\langle \chi_{\text{bound}}|\hat\Lambda|\chi_{\varepsilon}\rangle\bigr|^2\,\rho(\varepsilon) \qquad (8)

where: $\Gamma$ — homogeneous FWHM linewidth; $\tau$ — lifetime; $\chi_\varepsilon$ — continuum nuclear function of the repulsive state; $\rho(\varepsilon)$ — density of continuum states. A sharp line broadens into a diffuse one whose width reads out the coupling.

Derivation ANonadiabatic coupling and the origin of the energy denominator

The goal is to start from the exact Born-Huang expansion, project the molecular Schrödinger equation onto a single adiabatic state, and show that the first-derivative coupling carries the electronic gap in its denominator. This one calculation explains everything downstream — why the coupling is invisible in a deep well, why it diverges at a crossing, and why the adiabatic basis is the wrong one to compute with near an intersection.

Step 1. Substitute $\Psi=\sum_j\chi_j\phi_j$ into $(\hat T_N+\hat H_{el})\Psi=E\Psi$ and use $\hat H_{el}\phi_j=E_j\phi_j$ . Why: this replaces the electronic operator by its eigenvalue, leaving only $\hat T_N$ to act on the geometry-dependent states.

Step 2. The product rule for $\hat T_N$ on $\chi_j\phi_j$ generates a purely nuclear term plus a cross term $-\sum_A(\hbar^2/M_A)(\nabla_{R_A}\chi_j)\!\cdot\!(\nabla_{R_A}\phi_j)$ . Why: the electronic states depend on $R$ , so the nuclear Laplacian cannot pass through cleanly — the cross term seeds the coupling.

Step 3. Project onto $\langle\phi_i|$ and integrate using $\langle\phi_i|\phi_j\rangle=\delta_{ij}$ ; the off-diagonal collects into $\hat\Lambda_{ij}$ whose leading part is $\mathbf F_{ij}=\langle\phi_i|\nabla_R\phi_j\rangle$ . Why: orthonormality isolates the coupling as an overlap between one state and the gradient of another.

Step 4. Differentiate $\hat H_{el}\phi_j=E_j\phi_j$ with respect to $R$ and left-multiply by $\langle\phi_i|$ with $i\ne j$ ; the $\nabla_R E_j$ term dies since $\langle\phi_i|\phi_j\rangle=0$ , and $\langle\phi_i|\hat H_{el}=E_i\langle\phi_i|$ gives $\langle\phi_i|\nabla_R\phi_j\rangle=\langle\phi_i|\nabla_R\hat H_{el}|\phi_j\rangle/(E_j-E_i)$ . Why: this Hellmann-Feynman step trades the derivative of a wavefunction for a matrix element of the derivative of the Hamiltonian, and the gap lands in the denominator — coupling grows without bound as $E_j\to E_i$ .

Go deeper · Derivation B · 1 minTwo-state avoided crossing and the noncrossing rule

Derivation BTwo-state avoided crossing and the noncrossing rule

The goal is to diagonalize the real symmetric two-level Hamiltonian that governs any near-crossing and extract, from its eigenvalue structure alone, the von Neumann-Wigner noncrossing rule: two adiabatic curves of the same symmetry cannot cross in a diatomic. Along the way we find the minimum gap equals twice the diabatic coupling, the number a spectroscopist measures.

Step 1. Write the near-crossing Hamiltonian in a two-state diabatic basis as $H=\begin{pmatrix}V_{11}&V_{12}\\V_{12}&V_{22}\end{pmatrix}$ with smooth real entries. Why: diabatic states vary slowly with $R$ , pushing the fast physics into the ordinary potential coupling $V_{12}$ rather than a divergent derivative coupling.

Step 2. Diagonalize to $E_\pm=\bar V\pm\sqrt{((V_{11}-V_{22})/2)^2+V_{12}^2}$ with $\bar V=(V_{11}+V_{22})/2$ . Why: the square root is the object that decides whether the surfaces can ever meet.

Step 3. Demand $E_+=E_-$ : the radicand is a sum of two squares, so it vanishes only if $V_{11}=V_{22}$ and $V_{12}=0$ both hold. Why: two independent conditions must be met at once, not one — that counting is the entire content of the rule.

Step 4. A diatomic has one nuclear coordinate; one parameter cannot generically solve two equations, so same-symmetry curves avoid crossing. If symmetry forbids the coupling ( $V_{12}\equiv0$ for different-symmetry states) one condition remains and a real crossing is allowed. Where $V_{11}=V_{22}$ the gap is $(E_+-E_-)_{\min}=2|V_{12}|$ . Why: polyatomics with a two-dimensional branching space can host conical intersections while diatomics only show avoided crossings, and measuring the closest approach hands you $V_{12}$ directly.

Go deeper · Derivation C · 1 minLandau-Zener probability from the linearized crossing

Derivation CLandau-Zener probability from the linearized crossing

The goal is to convert the static two-state picture into a dynamical hop probability by letting the nuclei sweep through the crossing at constant velocity, reducing the problem to a linear two-level sweep whose exact solution is the Landau-Zener exponential. This is the workhorse estimate for curve-crossing rates, and it makes precise that speed favours a diabatic jump while sluggishness keeps the system adiabatic.

Step 1. Linearize the diabatic energy difference about the crossing $R_c$ as $V_{11}-V_{22}\approx\Delta F\,(R-R_c)$ and feed in uniform motion $R(t)=R_c+vt$ . Why: near the crossing only the local slope matters, and a linear sweep is the one case with a known closed form.

Step 2. The amplitudes obey $i\hbar\,\dot c=\begin{pmatrix}\tfrac12\Delta F v t&V_{12}\\V_{12}&-\tfrac12\Delta F v t\end{pmatrix}c$ , a linear ramp of the detuning with constant coupling. Why: this is exactly the model Landau, Zener, and Stückelberg solved in 1932, so we can quote the exact result.

Step 3. The exact asymptotic solution is $P_{\text{diab}}=\exp(-2\pi V_{12}^2/\hbar v|\Delta F|)$ , and the complement $P_{\text{adiab}}=1-P_{\text{diab}}$ follows one surface. Why: the exponent is dimensionless and compares the coupling energy to the sweep rate — large $V_{12}$ or slow $v$ makes the hop unlikely and keeps the wavepacket on the lower adiabatic curve, the classical limit of a reaction on a single surface.

Limiting cases

The strength of every nonadiabatic effect is set by the contest between the coupling and the adiabatic gap.

Case	Result	Meaning
Adiabatic (large gap)	$\bigl\|\langle\phi_i\|\hat T_N\|\phi_j\rangle\bigr\| \ll \|E_i-E_j\|$	Electrons follow the nuclei; Born-Oppenheimer holds and surfaces stay uncoupled. Ground state far from any crossing.
Strong coupling / near-degeneracy	$\mathbf F_{ij}=\dfrac{\langle\phi_i\|\nabla_R\hat H_{el}\|\phi_j\rangle}{E_j-E_i}\to\infty$	Derivative coupling diverges as the gap closes; the adiabatic basis becomes useless — switch to diabatic.
Sudden (diabatic) passage	$P_{\text{diab}}=\exp(-2\pi V_{12}^2/\hbar v\|\Delta F\|)\to1$	Fast passage keeps diabatic character, so the electronic state changes; large $v$ or small $V_{12}$ favours the hop.
Conical intersection (nonlinear)	$\dim(\text{seam})=3N-6-2$	Two branching coordinates lift the degeneracy linearly; the rest form the seam. Needs more than one nuclear dimension.

Common trapAdiabatic curves of the same symmetry can cross if you just push them together

They cannot, and assuming otherwise is the classic error. In a diatomic there is a single nuclear coordinate, and a genuine degeneracy demands two conditions at once — $V_{11}=V_{22}$ and $V_{12}=0$ . One knob cannot satisfy two equations generically, so same-symmetry curves are forbidden to cross and instead form an avoided crossing split by $2|V_{12}|$ ; only states of different symmetry, for which $V_{12}\equiv0$ , may truly cross. Genuine touchings of same-symmetry states — conical intersections — require at least two nuclear coordinates, so they live in polyatomics. A subtler trap: the divergence of the derivative coupling at an intersection is an artefact of the adiabatic representation, not a physical singularity — the diabatic energies stay perfectly smooth through the crossing, which is why diabatic bases are used to compute the dynamics.

Worked examples

1 · Landau-Zener hop at a curve crossing (confidence builder)

A diatomic has an avoided crossing with diabatic coupling $V_{12}=0.10$ eV and slope difference $\Delta F=2.0$ eV/Å. A wavepacket passes at nuclear speed $v=5.0\times10^{3}$ m/s. Find the probability it stays diabatic (changes electronic state) and the crossover speed at which the exponent equals one.

Step 1. Put everything in SI: $V_{12}=0.10\times1.602\times10^{-19}=1.602\times10^{-20}$ J, $\Delta F=2.0\ \text{eV/Å}=3.204\times10^{-9}$ J/m, $\hbar=1.0546\times10^{-34}$ J·s, $v=5.0\times10^{3}$ m/s.

Step 2. Assemble the exponent $2\pi V_{12}^2/(\hbar v|\Delta F|)$ . Numerator $2\pi(1.602\times10^{-20})^2=1.613\times10^{-39}$ J²; denominator $(1.0546\times10^{-34})(5.0\times10^{3})(3.204\times10^{-9})=1.689\times10^{-39}$ J². The ratio is dimensionless, as it must be: $\text{exponent}=0.955$ .

Step 3. Exponentiate: $P_{\text{diab}}=e^{-0.955}=\mathbf{0.38}$ , so $P_{\text{adiab}}=\mathbf{0.62}$ — more likely to follow the lower surface than to hop, but the outcomes are comparable.

Step 4. The crossover speed sets the exponent to one: $v^*=2\pi V_{12}^2/(\hbar|\Delta F|)=\mathbf{4.8\times10^{3}\ m/s}$ . Our $v$ sits just above $v^*$ , so the exponent is just below one and $P_{\text{diab}}$ just below $1/e$ — consistent with $0.38$ . P_diab = 0.38, P_adiab = 0.62, crossover 4.8×10³ m/s.

2 · Predissociation lifetime from a diffuse linewidth

A rotational line in the O₂ Schumann-Runge bands is broadened by predissociation to a homogeneous width $\Gamma=2.0$ cm⁻¹ (FWHM). Find the lifetime, and the lifetime-width scaling.

Step 1. Convert the width to energy: $\Gamma=hc\,\tilde\nu=(6.626\times10^{-34})(2.998\times10^{10}\ \text{cm/s})(2.0\ \text{cm}^{-1})=3.97\times10^{-23}$ J.

Step 2. Apply $\Gamma=\hbar/\tau$ : $\tau=\hbar/\Gamma=1.0546\times10^{-34}/3.97\times10^{-23}=\mathbf{2.7\times10^{-12}\ s}$ , about $2.7$ ps — a few picoseconds before the coupling dumps the level into the repulsive continuum.

Step 3. The shortcut follows directly: $\tau[\text{ps}]=\hbar/(hc\,\tilde\nu)=5.31/\Gamma[\text{cm}^{-1}]$ . A $1$ cm⁻¹ line means $5.3$ ps; a $4$ cm⁻¹ line only $1.3$ ps. τ = ħ/Γ ≈ 2.7 ps, and τ[ps] = 5.31/Γ[cm⁻¹] — wider lines mean shorter-lived levels. Note the same coupling ties both examples together: the crossing of Example 1 has a minimum adiabatic gap $(E_+-E_-)_{\min}=2|V_{12}|=0.20$ eV $=1.6\times10^3$ cm⁻¹ — a splitting easily resolved optically — and that measured $V_{12}$ is exactly the number that reappears squared in the Landau-Zener exponent, fixing both the static splitting and the dynamical hop.

Physics of the portrait

The phenomena — predissociation, perturbations, Λ-doubling — were seen in spectra long before the theory matured, and no one did more to organize them than the founder of molecular spectroscopy.

Origins · 1 minGerhard Herzberg and how this was discovered

Gerhard Herzberg

1904–1999 · molecular spectroscopy: predissociation, perturbations, and Λ-doubling

Gerhard Herzberg, born in Hamburg in 1904 and forced to emigrate in 1935, built at Saskatchewan and later the National Research Council of Canada the most complete experimental picture of molecular structure of his century. His three-volume Molecular Spectra and Molecular Structure became the standard reference because it tied the abstract machinery of coupled electronic states to measurable spectra. It is in Herzberg that predissociation is systematized — a sharp band going diffuse as a bound level couples to a repulsive continuum, the onset of diffuseness locating the crossing — and that local perturbations and Λ-doubling are each read as signatures of the coupling that produces them. He won the 1971 Nobel Prize in Chemistry for this work. With Longuet-Higgins in 1963 he also identified the sign change — the geometric phase — that a wavefunction acquires on a loop around a conical intersection, two decades before Berry named the effect. The fingerprints this article decodes are, to a remarkable degree, the ones Herzberg taught spectroscopists to recognize.

History

1927

— Born and Oppenheimer justify separating electronic and nuclear motion using the small mass ratio

1929

— von Neumann and Wigner prove the noncrossing rule for terms of the same symmetry

1932

— Landau, Zener, and Stückelberg independently solve the two-level curve-crossing problem

1937

— Teller shows that same-symmetry surfaces can meet at a conical intersection in polyatomics

1950

— Herzberg systematizes predissociation, perturbations, and Λ-doubling in Molecular Spectra and Molecular Structure

1963

— Herzberg and Longuet-Higgins identify the geometric phase acquired around a conical intersection

Going further

Texts. Herzberg, Molecular Spectra and Molecular Structure I: Spectra of Diatomic Molecules (predissociation, perturbations, Λ-doubling) · Demtröder, Laser Spectroscopy, Vol. 1 (nonadiabatic coupling, predissociation, perturbations) · Foot, Atomic Physics (Born-Oppenheimer separation) · Saleh and Teich, Fundamentals of Photonics (molecular levels) · Loudon, The Quantum Theory of Light (radiative widths) · Baer, Beyond Born-Oppenheimer (diabatic representations and the geometric phase).

Provenance. The coupled-nuclear equations and derivative-coupling denominator follow the standard Born-Huang treatment in Herzberg and Demtröder; the noncrossing rule is von Neumann-Wigner (1929); the two-level hop is Landau-Zener-Stückelberg (1932); the conical intersection and its geometric phase trace to Teller (1937) and Herzberg-Longuet-Higgins (1963). A verifier should recompute: (1) the Hellmann-Feynman step giving $\mathbf F_{ij}=\langle\phi_i|\nabla_R\hat H_{el}|\phi_j\rangle/(E_j-E_i)$ ; (2) $(E_+-E_-)_{\min}=2|V_{12}|$ ; (3) the Landau-Zener exponent and Example 1 ( $P_{\text{diab}}=0.38$ at $v=5.0\times10^3$ m/s, crossover $4.8\times10^3$ m/s); (4) $\tau=\hbar/\Gamma=5.31/\Gamma[\text{cm}^{-1}]$ ps; (5) $0.20$ eV $=1.6\times10^3$ cm⁻¹. Inputs $V_{12}=0.10$ eV and $\Delta F=2.0$ eV/Å are illustrative; the O₂ width is a representative order of magnitude.

Builds on. The Born-Oppenheimer Approximation · The Chemical Bond and Molecular Potential Energy Curves · Molecular Term Symbols · Electronic Transitions in Molecules · The Rigid Rotor Model. Leads to. The Franck-Condon Principle · Vibration-Rotation Spectra and P, Q, R Branches · Anharmonicity and the Morse Potential.

Ask the tutor: ask me to walk a wavepacket through a specific avoided crossing and compute its Landau-Zener hop probability step by step, or to turn a measured diffuse linewidth into a predissociation lifetime.

What is the difference between adiabatic and diabatic states?

Adiabatic states are eigenstates of the electronic Hamiltonian at each fixed geometry, so they carry smooth surfaces but a derivative coupling that diverges at crossings. Diabatic states are a rotated basis chosen so their electronic character changes as little as possible with geometry, which pushes the fast physics out of the divergent kinetic coupling and into a smooth potential coupling $V_{12}$ . The trade is deliberate: adiabatic surfaces are what you diagonalize and plot, but diabatic states are what you propagate dynamics with, because their coupling stays finite straight through a crossing. A rigorously diabatic basis does not exist in general, but excellent quasi-diabatic ones are built routinely.

Why is the Born-Oppenheimer approximation so accurate yet not exact?

Its accuracy comes from the mass ratio: electrons are so much lighter than nuclei that they adjust almost instantaneously, and the neglected coupling scales as the mass ratio times an energy-gap ratio, so deep in a well the single-surface picture is superb. It fails not because the mass ratio changes but because the gap can collapse: near an avoided crossing or conical intersection the denominator $E_j-E_i$ shrinks toward zero and the once-negligible coupling dominates. So Born-Oppenheimer is a statement about being far from other electronic states, and it breaks exactly where you are not.

Is a conical intersection just an avoided crossing in more dimensions?

They are relatives but not the same. An avoided crossing occurs along one coordinate where same-symmetry surfaces approach to a minimum gap $2|V_{12}|$ but never touch. A conical intersection is a true degeneracy: the surfaces meet at a point and split linearly in the two-dimensional branching space, tracing a double cone. Slice it along a generic one-dimensional path and you see an avoided crossing, which is why a diatomic — one coordinate — only ever shows avoided crossings. The extra dimension is decisive: it makes the degeneracy generic rather than accidental and endows any loop around the tip with a geometric phase.

What is the geometric phase around a conical intersection?

When a molecular wavefunction is transported adiabatically around a closed loop enclosing a conical intersection, it returns with its sign reversed — a phase of $\pi$ . Herzberg and Longuet-Higgins found this in 1963, two decades before Berry showed it was a general feature of adiabatic transport. The phase is not optional bookkeeping: it changes the allowed vibrational levels and the interference between paths passing on opposite sides of the intersection, and because the nuclear wavefunction must stay single-valued the electronic sign change forces a compensating change in the nuclear part — entangling the two precisely where Born-Oppenheimer fails.

How is predissociation different from ordinary photodissociation?

In direct photodissociation the molecule is excited straight onto a repulsive surface and flies apart within a vibrational period, giving a broad, structureless absorption. In predissociation it is first excited to a genuinely bound level, which would give a sharp line — but that level is nonadiabatically coupled to a repulsive continuum at the same energy, so it leaks apart over a longer time and its rotational lines stay resolvable but diffuse, broadened by $\Gamma=\hbar/\tau$ . The width measures the coupling to the continuum, and the quantum number at which lines first blur out locates the crossing.

What exactly are Λ-doubling and rovibronic coupling?

In a state with nonzero electronic orbital angular momentum along the axis ( $\Lambda\ne0$ ) the two senses of orbital circulation are degenerate in a nonrotating molecule. Rotation breaks that symmetry: the orbital motion couples to molecular rotation and the degeneracy splits into a Λ-doublet spaced roughly as $B_v J(J+1)$ . Rovibronic coupling is the general name for this entanglement of rotation, vibration, and electronic motion — the Coriolis and rotational-electronic terms that the clean additive energy $E(v,J)=T_e+\omega_e(v+\tfrac12)+B_vJ(J+1)$ ignores. Measuring the doublet spacing versus $J$ yields the electronic-rotational coupling constants directly.