12.09.05 · quantum / identical-particles

Diatomic molecule and the Born-Oppenheimer approximation

shipped3 tiersLean: none

Anchor (Master): Landau, L. D. & Lifshitz, E. M., *Quantum Mechanics: Non-Relativistic Theory*, 3rd ed. (Pergamon, 1977), Ch. XI §§77–82; Born, M. & Huang, K., *Dynamical Theory of Crystal Lattices* (Oxford, 1954), Appendices VII–VIII

Intuition Beginner

A molecule is a swarm of light electrons darting around a few heavy nuclei. A proton weighs about two thousand times as much as an electron, so the nuclei lumber while the electrons race. The Born-Oppenheimer idea is to use this huge weight gap. Freeze the nuclei in place, let the fast electrons settle into their lowest arrangement around those fixed positions, and read off the energy. Then move the nuclei a little and repeat. The electrons keep up so quickly that they always look settled, no matter how the slow nuclei drift.

This gives a clean two-step picture. First you find the electron energy for every possible spacing of the nuclei. That energy, plotted against the distance between the nuclei, is a curve with a dip. The bottom of the dip is the comfortable bond length, where the molecule likes to sit. Push the nuclei closer and they repel; pull them apart and the bond weakens. The dip acts like a valley that the nuclei roll around in.

Why bother? Because it splits one hopeless problem into two solvable ones. You never have to track electrons and nuclei at the same time. You get the bond length, the stiffness of the bond, and the spacing of the molecule's energy levels, all from that one energy curve.

Visual Beginner

Picture a graph. The horizontal axis is the distance between the two nuclei of a diatomic molecule. The vertical axis is the electronic energy you get after letting the electrons settle at each fixed distance. The curve starts high on the left, where the nuclei are jammed together and repel hard. It swoops down to a lowest point, the dip, at the equilibrium bond length. Then it climbs back up and flattens out on the right, where the atoms have drifted so far apart that they barely feel each other.

Near the bottom, the valley looks like a smooth bowl, and the nuclei wobble in it the way a marble rocks in a salad bowl. Those wobbles are the vibrations of the molecule. Higher up, the valley gets lopsided and the wobbles change spacing, crowding closer together as you near the flat plateau where the bond finally breaks.

Worked example Beginner

Take the hydrogen molecule, two protons and two electrons. We want a rough estimate of how fast it vibrates, using only the shape of the energy valley near its bottom.

Step 1. Read off the valley. Experiment tells us the bottom of the energy curve sits at a bond length of about $0.74$ ångström. Near the bottom the curve is shaped like a parabola, and the stiffness of that parabola (how sharply it curves) is measured by a spring constant of about $510$ newtons per metre for hydrogen.

Step 2. Use the spring picture. A mass on a spring vibrates at a rate set by the spring constant divided by the mass, then square-rooted. For two protons sharing the wobble, the mass that matters is half a proton mass, which is about $8.4 \times 1 0^{- 28}$ kilograms.

Step 3. Compute the rate. Dividing the spring constant by the mass gives $510/ (8.4 \times 1 0^{- 28}) \approx 6.1 \times 1 0^{29}$ per second squared. Taking the square root gives about $7.8 \times 1 0^{14}$ wobbles per second (in angular units).

Step 4. Turn it into an energy. Multiplying that rate by Planck's constant gives a vibrational energy step of about $0.52$ electron-volts. The measured value is close to $0.54$ electron-volts.

What this tells us: the single energy curve, through nothing more than the stiffness of its valley floor, already predicts how the molecule vibrates and lands within a few percent of the real number.

Check your understanding Beginner

Exercise (easy, multiple choice).

What feature of a molecule lets the Born-Oppenheimer approximation treat the nuclei as nearly fixed while the electrons settle?

A. The electrons carry less charge than the nuclei.
B. The nuclei are far heavier than the electrons, so they move much more slowly.
C. The electrons are held in place by the chemical bond.
D. The molecule is at very low temperature.

Hint

Compare the mass of a proton to the mass of an electron, and think about which one moves faster.

Answer

B. The nuclei are far heavier than the electrons, so they move much more slowly.

A proton outweighs an electron by roughly two thousand to one. The light electrons race around and reach their settled arrangement almost instantly compared to the slow drift of the heavy nuclei, which is what lets us freeze the nuclei while solving for the electrons. Charge, bonding, and temperature are not the reason.

Formal definition Intermediate+

Consider a diatomic molecule with nuclei of masses $M_{A}, M_{B}$ at positions $R_{A}, R_{B}$ and $n$ electrons at positions $r = (r_{1}, \dots, r_{n})$ . In atomic units the molecular Hamiltonian is $$ \hat H = \hat T_{\mathrm{nuc}} + \hat H_{\mathrm{el}}(\mathbf R), \qquad \hat T_{\mathrm{nuc}} = -\frac{1}{2M_A}\nabla_{\mathbf R_A}^2 - \frac{1}{2M_B}\nabla_{\mathbf R_B}^2, $$ where the electronic Hamiltonian at clamped nuclear configuration $R = (R_{A}, R_{B})$ is $$ \hat H_{\mathrm{el}}(\mathbf R) = -\tfrac12\sum_{i=1}^{n}\nabla_{\mathbf r_i}^2

\sum_{i,X} \frac{Z_X}{|\mathbf r_i - \mathbf R_X|}

\sum_{i<j}\frac{1}{|\mathbf r_i - \mathbf r_j|}
\frac{Z_A Z_B}{|\mathbf R_A - \mathbf R_B|}. $The nuclear kinetic operator carries the small prefactors $1/M_X$; the electron mass is set to one. The ratio that controls the whole construction is the mass disparity, and the natural expansion parameter is the fourth root$ \kappa = \left(\frac{m_e}{M_N}\right)^{1/4}, $$ introduced by Born and Oppenheimer because the four powers of $κ$ organise the successive corrections: electronic energy at order $κ^{0}$ , vibrational quanta at order $κ^{2}$ , and rotational quanta at order $κ^{4}$ .

For each fixed $R$ , the operator $\hat{H}_{el} (R)$ acts on the electronic Hilbert space $H_{el} = ⋀^{n} (L^{2} (R^{3}) \otimes C^{2})$ and has a discrete spectrum of electronic terms $$ \hat H_{\mathrm{el}}(\mathbf R),\phi_k(\mathbf r;\mathbf R) = E_k(\mathbf R),\phi_k(\mathbf r;\mathbf R), \qquad k = 0,1,2,\dots, $$ where the eigenfunctions $ϕ_{k} (r; R)$ depend on $R$ as a parameter. Because the molecule is invariant under rotation about the internuclear axis, $E_{k}$ depends only on the scalar internuclear distance $R = ∣ R_{A} - R_{B} ∣$ , and each curve $E_{k} (R)$ is a potential-energy curve. The Born-Oppenheimer ansatz writes the full molecular wavefunction as a single product $$ \Psi(\mathbf r, \mathbf R) = \chi(\mathbf R),\phi_k(\mathbf r;\mathbf R), $$ with $χ$ a nuclear wavefunction to be determined. The associated effective nuclear potential is the chosen electronic term $E_{k} (R)$ , and the equilibrium internuclear distance $R_{e}$ is the minimiser of $E_{k} (R)$ , satisfying $E_{k}^{'} (R_{e}) = 0$ and $E_{k}^{''} (R_{e}) > 0$ .

Counterexamples to common slips

The clamped-nucleus eigenfunctions $ϕ_{k} (r; R)$ are parametrically dependent on $R$ , not dynamical functions of it. Treating $\nabla_{R}$ as ignoring this dependence drops the derivative couplings and discards the entire correction structure.
$E_{k} (R)$ already contains the nuclear repulsion $Z_{A} Z_{B} / R$ . Forgetting that constant term removes the repulsive wall and the curve loses its minimum.
The ansatz $Ψ = χ ϕ_{k}$ is a single product, not a sum over electronic states. It is exact only in the limit of vanishing nuclear kinetic energy; the neglected off-diagonal couplings to other $ϕ_{l}$ are precisely the non-adiabatic terms.

Key theorem with proof Intermediate+

Theorem (Born-Oppenheimer separation). Insert the product ansatz $Ψ = χ (R) ϕ_{k} (r; R)$ into $\hat{H} Ψ = E Ψ$ with $ϕ_{k}$ normalised, $⟨ ϕ_{k} ∣ ϕ_{k} ⟩_{r} = 1$ for every $R$ . Projecting onto $ϕ_{k}$ yields the nuclear equation $$ \Big[,\hat T_{\mathrm{nuc}} + E_k(R) + U_{kk}(\mathbf R),\Big]\chi(\mathbf R) = E,\chi(\mathbf R), $$ where the diagonal Born-Oppenheimer correction is $$ U_{kk}(\mathbf R) = \sum_{X}\frac{1}{2M_X}\Big[ -\langle\phi_k|\nabla_{\mathbf R_X}^2|\phi_k\rangle_{\mathbf r} \Big], $$ and the off-diagonal couplings to other electronic states $ϕ_{l}$ ( $l \neq = k$ ) are neglected. Dropping $U_{k k}$ as well gives the standard Born-Oppenheimer nuclear equation with effective potential $E_{k} (R)$ .

Proof. Apply $\hat{H} = \hat{T}_{nuc} + \hat{H}_{el}$ to $Ψ = χ ϕ_{k}$ . The electronic part acts on $ϕ_{k}$ alone: $$ \hat H_{\mathrm{el}}(\chi\phi_k) = \chi,\hat H_{\mathrm{el}}\phi_k = \chi,E_k(R)\phi_k. $$ The nuclear kinetic term differentiates a product. For each nucleus $X$ , using $\nabla^{2} (χ ϕ_{k}) = (\nabla^{2} χ) ϕ_{k} + 2 (\nabla χ) \cdot (\nabla ϕ_{k}) + χ (\nabla^{2} ϕ_{k})$ with $\nabla = \nabla_{R_{X}}$ , $$ \hat T_{\mathrm{nuc}}(\chi\phi_k) = \sum_X\frac{-1}{2M_X}\Big[ (\nabla^2\chi)\phi_k + 2(\nabla\chi)\cdot(\nabla\phi_k) + \chi(\nabla^2\phi_k) \Big]. $$ Project the eigenvalue equation onto $ϕ_{k}$ by taking $⟨ ϕ_{k} ∣ \cdot ⟩_{r}$ , integrating over electronic coordinates with $⟨ ϕ_{k} ∣ ϕ_{k} ⟩ = 1$ . The first term gives $- \sum_{X} (2 M_{X})^{- 1} \nabla^{2} χ = \hat{T}_{nuc} χ$ . The middle term contributes the first-derivative (non-adiabatic) coupling $$ \mathbf F_{kk}^{(X)} = \langle\phi_k|\nabla_{\mathbf R_X}\phi_k\rangle_{\mathbf r}, $$ which vanishes for real $ϕ_{k}$ : differentiating $⟨ ϕ_{k} ∣ ϕ_{k} ⟩ = 1$ gives $F_{k k} + F_{k k}^{*} = 0$ , so $F_{k k}$ is purely imaginary, hence zero when the phase is chosen real. The third term, projected, is $- \sum_{X} (2 M_{X})^{- 1} χ ⟨ ϕ_{k} ∣ \nabla^{2} ϕ_{k} ⟩$ , which is $χ U_{k k}$ . Collecting the surviving pieces and the electronic eigenvalue $E_{k} (R)$ , $$ \big[\hat T_{\mathrm{nuc}} + E_k(R) + U_{kk}(\mathbf R)\big]\chi = E\chi. $$ The terms discarded in the projection are the off-diagonal couplings $⟨ ϕ_{l} ∣\nabla ϕ_{k} ⟩$ and $⟨ ϕ_{l} ∣ \nabla^{2} ϕ_{k} ⟩$ for $l \neq = k$ , each carrying a factor $1/ M_{X}$ and therefore suppressed when the nuclei are heavy. Neglecting $U_{k k}$ as well — itself of order $m_{e} / M_{N}$ relative to $E_{k}$ — leaves the nuclear motion governed by the single electronic surface $E_{k} (R)$ . $□$

Bridge. This separation builds toward the entire theory of molecular spectra and appears again in 12.09.03, where the same clamped-nucleus electronic problem is what a Hartree-Fock calculation solves at each fixed geometry to trace out $E_{k} (R)$ . The foundational reason the product ansatz works is the mass disparity packaged in $κ = (m_{e} / M_{N})^{1/4}$ : the off-diagonal couplings that the proof discards are smaller than the retained terms by powers of $κ$ , so the single-surface picture is the leading order of a controlled expansion rather than a guess. This is exactly the adiabatic principle in disguise — the fast electronic subsystem follows the slow nuclear coordinates so faithfully that it stays in one eigenstate — and it generalises the time-dependent adiabatic theorem from a one-parameter time variable to the multidimensional nuclear configuration space. The central insight is that the electronic energy becomes a potential for the nuclei: putting these together, the spectrum of one operator (the clamped-nucleus electronic Hamiltonian) supplies the potential energy in a second operator (the nuclear Schrödinger equation), and the equilibrium bond length, the vibrational quanta, and the rotational quanta all fall out of the shape of that potential near its minimum.

Exercises Intermediate+

Exercise 7 (hard, short-answer).

The exact Morse spectrum is $E_{v} = ℏ ω (v + \frac{1}{2}) - ℏ ω x_{e} (v + \frac{1}{2})^{2}$ with anharmonicity constant $x_{e} = ℏ ω / (4 D_{e})$ . Derive the dissociation energy $D_{0}$ (energy from $v = 0$ to the top of the well) and explain why $D_{0} < D_{e}$ .

Hint

$D_{e}$ is measured from the bottom of the well; $D_{0}$ is measured from the $v = 0$ level, which already sits above the bottom by the zero-point energy.

Answer

The $v = 0$ level lies at $E_{0} = \frac{1}{2} ℏ ω - \frac{1}{4} ℏ ω x_{e}$ above the well bottom (the zero-point energy, reduced slightly by anharmonicity). The dissociation limit sits at depth $D_{e}$ . Therefore $D_{0} = D_{e} - E_{0} = D_{e} - \frac{1}{2} ℏ ω + \frac{1}{4} ℏ ω x_{e}$ . The spectroscopic dissociation energy $D_{0}$ is smaller than the well depth $D_{e}$ precisely because the molecule can never sit at the bottom of the well: the zero-point motion lifts the ground state, so less energy is needed to pull the bond apart. Rubric: full credit for $D_{0} = D_{e} - E_{0}$ with the zero-point explanation.

Exercise 8 (hard, short-answer).

Explain what a conical intersection is and why the Born-Oppenheimer approximation fails near one.

Hint

Consider what happens to the energy gap $E_{l} (R) - E_{k} (R)$ between two electronic surfaces and how the derivative coupling depends on that gap.

Answer

A conical intersection is a nuclear configuration where two electronic potential surfaces $E_{k} (R)$ and $E_{l} (R)$ become degenerate, touching in a double-cone shape (codimension two in nuclear space for a real Hamiltonian, the von Neumann-Wigner non-crossing rule). The off-diagonal derivative coupling has the form $⟨ ϕ_{k} ∣\nabla \hat{H}_{el} ∣ ϕ_{l} ⟩ / (E_{l} - E_{k})$ , which diverges as the gap $E_{l} - E_{k} \to 0$ . The single-surface ansatz then breaks down because the nuclear motion can no longer stay on one surface: the wavepacket splits, and population funnels between electronic states. This non-adiabatic transfer is the dominant pathway for fast radiationless decay in photochemistry. Rubric: full credit for the degeneracy/cone description plus the gap-driven divergence of the coupling.

Advanced results Master

The adiabatic expansion in powers of $κ$ . Born and Oppenheimer organised the molecular eigenvalue problem as a perturbation series in $κ = (m_{e} / M_{N})^{1/4}$ . Writing the nuclear displacement from equilibrium as $ξ = κ^{- 1} (R - R_{e})$ rescales the nuclear kinetic operator so that the leading nuclear term enters at order $κ^{2}$ . The expansion then reads $E = E_{el} (R_{e}) + κ^{2} E^{(2)} + κ^{4} E^{(4)} + \dots$ , with $E^{(2)}$ the harmonic vibrational energy $ℏ ω (v + \frac{1}{2})$ and $E^{(4)}$ collecting the rotational energy $ℏ^{2} J (J + 1) /2 I$ together with the leading anharmonic and rotation-vibration corrections. Odd powers of $κ$ vanish by the reflection symmetry of the harmonic well, which is why the expansion proceeds in even powers and the rotational scale is suppressed relative to the vibrational scale by a further factor $κ^{2} \sim m_{e} / M_{N}$ .

Rovibrational structure near the minimum. Expanding $E_{k} (R)$ about $R_{e}$ and adding the rotational barrier $ℏ^{2} J (J + 1) /2 μ R^{2}$ gives the effective radial potential. To leading order the levels separate into a sum $$ E_{vJ} = E_k(R_e) + \hbar\omega\Big(v+\tfrac12\Big) + \frac{\hbar^2}{2I}J(J+1), $$ with $I = μ R_{e}^{2}$ . The Morse potential $V (R) = D_{e} [1 - e^{- a (R - R_{e})}]^{2}$ refines the vibrational ladder to the exact closed form $E_{v} = ℏ ω (v + \frac{1}{2}) - ℏ ω x_{e} (v + \frac{1}{2})^{2}$ with $x_{e} = ℏ ω /4 D_{e}$ , reproducing the crowding of levels toward dissociation and supporting a finite number of bound states $v_{m a x} = ⌊ 1/ (2 x_{e}) - 1/2 ⌋$ . Coupling to rotation through the $R$ -dependence of $I$ produces the rotation-vibration interaction constant $α_{e}$ in the standard spectroscopic expansion $E_{v J} / h c = ω_{e} (v + \frac{1}{2}) - ω_{e} x_{e} (v + \frac{1}{2})^{2} + [B_{e} - α_{e} (v + \frac{1}{2})] J (J + 1)$ .

Non-adiabatic couplings and the derivative-coupling matrix. Retaining the off-diagonal terms restores the exact problem in the form of coupled nuclear equations $$ \Big[\hat T_{\mathrm{nuc}} + E_k(\mathbf R) - E\Big]\chi_k = \sum_{l\ne k}\Big[ \frac{1}{M}\mathbf F_{kl}\cdot\nabla_{\mathbf R} + \frac{1}{2M}G_{kl} \Big]\chi_l, $$ with the first-derivative coupling $F_{k l} = ⟨ ϕ_{k} ∣ \nabla_{R} ϕ_{l} ⟩$ and the second-derivative coupling $G_{k l} = ⟨ ϕ_{k} ∣ \nabla_{R}^{2} ϕ_{l} ⟩$ . By the Hellmann-Feynman relation $F_{k l} = ⟨ ϕ_{k} ∣\nabla \hat{H}_{el} ∣ ϕ_{l} ⟩ / (E_{l} - E_{k})$ for $l \neq = k$ , so the coupling grows as the electronic gap shrinks and diverges at a degeneracy. This is the analytic signature of the breakdown of the single-surface picture.

Conical intersections and the geometric phase. At a point where two real electronic surfaces meet, the non-crossing rule of von Neumann and Wigner requires two independent nuclear parameters to be tuned, so the seam of intersection has codimension two and the surfaces form a double cone. Transporting a real electronic eigenstate around a loop that encircles the intersection returns it to minus itself: the wavefunction acquires a sign change, the molecular Berry phase. A consistent single-valued total wavefunction then forces a compensating boundary condition on the nuclear factor $χ$ , which shifts the allowed nuclear angular momenta by a half-integer and reorganises the low-lying rovibronic spectrum. The geometric phase is the topological fingerprint of the intersection on the nuclear motion.

Crude-adiabatic versus adiabatic bases. Two natural choices of electronic basis appear. The adiabatic basis diagonalises $\hat{H}_{el} (R)$ at every $R$ , giving smooth potential surfaces but singular derivative couplings at intersections. The diabatic basis instead trades curve smoothness for a basis whose derivative couplings vanish (or are minimised), replacing the singular kinetic coupling by a smooth potential coupling — at the cost that exact diabatic states do not generally exist in more than one dimension (the Mead-Truhlar obstruction). Spectroscopy and reaction dynamics choose between them according to whether the relevant gap is large (adiabatic) or small (diabatic).

Synthesis. The Born-Oppenheimer separation is the foundational reason chemistry has a geometry at all: the central insight is that diagonalising the fast electronic Hamiltonian at every clamped nuclear configuration manufactures a potential-energy surface, and this is exactly the object on which bond lengths, force constants, and reaction barriers are defined. The $κ = (m_{e} / M_{N})^{1/4}$ expansion identifies the electronic, vibrational, and rotational energy scales with successive even powers of $κ$ , and that ordering generalises the time-dependent adiabatic theorem to the multidimensional nuclear manifold, so the molecular spectrum is the adiabatic theorem made spectroscopic. Putting these together, the diagonal correction $U_{k k}$ , the Morse anharmonicity, and the rigid-rotor ladder are all refinements of a single surface, while the derivative couplings $F_{k l}$ measure exactly where that surface picture must be abandoned — the same gap-driven divergence that produces conical intersections also produces the Berry phase, so the topology of the electronic degeneracy is dual to the boundary condition on the nuclear wavefunction. The bridge from this unit to electronic-structure practice is that the surface $E_{k} (R)$ is what a self-consistent-field calculation computes geometry by geometry, which appears again wherever a potential-energy surface is scanned, optimised, or propagated.

Full proof set Master

Proposition (vanishing of the diagonal first-derivative coupling). For a non-degenerate electronic eigenstate $ϕ_{k} (r; R)$ with a smooth real phase, the diagonal first-derivative coupling vanishes: $F_{k k} (R) = ⟨ ϕ_{k} ∣ \nabla_{R} ϕ_{k} ⟩_{r} = 0$ .

Proof. Normalisation $⟨ ϕ_{k} ∣ ϕ_{k} ⟩ = 1$ holds for every $R$ . Differentiate with respect to a component of $R$ : $$ 0 = \nabla_{\mathbf R}\langle\phi_k|\phi_k\rangle = \langle\nabla_{\mathbf R}\phi_k|\phi_k\rangle + \langle\phi_k|\nabla_{\mathbf R}\phi_k\rangle = \mathbf F_{kk}^* + \mathbf F_{kk}. $$ Thus $Re F_{k k} = 0$ and $F_{k k}$ is purely imaginary. For a non-degenerate state the eigenfunction is determined up to a phase $e^{i θ (R)}$ ; choosing $ϕ_{k}$ real (possible locally because $\hat{H}_{el}$ is real-symmetric and the level is isolated) makes $F_{k k}$ real. A quantity that is both real and purely imaginary is zero. $□$

Proposition (Hellmann-Feynman form of the off-diagonal coupling). For $l \neq = k$ with $E_{l} \neq = E_{k}$ , $$ \mathbf F_{kl}(\mathbf R) = \langle\phi_k|\nabla_{\mathbf R}\phi_l\rangle_{\mathbf r} = \frac{\langle\phi_k|,(\nabla_{\mathbf R}\hat H_{\mathrm{el}}),|\phi_l\rangle_{\mathbf r}}{E_l(\mathbf R) - E_k(\mathbf R)}. $$

Proof. Start from the parametric eigenvalue equation $\hat{H}_{el} ϕ_{l} = E_{l} ϕ_{l}$ and apply $\nabla_{R}$ : $$ (\nabla\hat H_{\mathrm{el}})\phi_l + \hat H_{\mathrm{el}}(\nabla\phi_l) = (\nabla E_l)\phi_l + E_l(\nabla\phi_l). $$ Take the inner product with $ϕ_{k}$ on the left. Since $\hat{H}_{el}$ is self-adjoint, $⟨ ϕ_{k} ∣ \hat{H}_{el} ∣\nabla ϕ_{l} ⟩ = E_{k} ⟨ ϕ_{k} ∣\nabla ϕ_{l} ⟩ = E_{k} F_{k l}$ . The term $⟨ ϕ_{k} ∣ (\nabla E_{l}) ϕ_{l} ⟩ = (\nabla E_{l}) ⟨ ϕ_{k} ∣ ϕ_{l} ⟩ = 0$ for $l \neq = k$ by orthogonality. Rearranging, $$ \langle\phi_k|(\nabla\hat H_{\mathrm{el}})|\phi_l\rangle + E_k,\mathbf F_{kl} = E_l,\mathbf F_{kl}, $$ hence $F_{k l} = ⟨ ϕ_{k} ∣ (\nabla \hat{H}_{el}) ∣ ϕ_{l} ⟩ / (E_{l} - E_{k})$ , valid whenever the denominator is non-zero. $□$

Proposition (exact Morse vibrational spectrum). The one-dimensional Schrödinger equation with potential $V (R) = D_{e} [1 - e^{- a (R - R_{e})}]^{2}$ and reduced mass $μ$ has bound-state energies $$ E_v = \hbar\omega\Big(v+\tfrac12\Big) - \frac{\hbar^2 a^2}{2\mu}\Big(v+\tfrac12\Big)^2, \qquad \omega = a\sqrt{\frac{2D_e}{\mu}}, $$ for integer $v = 0, 1, \dots, v_{m a x}$ , with $v_{m a x}$ the largest integer below $λ - \frac{1}{2}$ where $λ = 2 μ D_{e} / (a ℏ)$ .

Proof. Substitute $y = e^{- a (R - R_{e})}$ , mapping $R \in (- \infty, \infty)$ to $y \in (0, \infty)$ and converting the radial equation into a confluent-hypergeometric (Whittaker) equation. With the constant $λ = 2 μ D_{e} / (a ℏ)$ and the scaled energy $ϵ = - 2 μ E / (a ℏ)$ , the bound-state solutions normalisable as $y \to 0, \infty$ exist only when the series terminates, which requires $λ - ϵ - \frac{1}{2} = v$ for a non-negative integer $v$ . Solving for $E$ gives $- 2 μ E / (a ℏ) = λ - (v + \frac{1}{2})$ , so $$ -E = \frac{a^2\hbar^2}{2\mu}\Big[\lambda - (v+\tfrac12)\Big]^2. $$ Measuring energies from the dissociation limit and using $D_{e} = a^{2} ℏ^{2} λ^{2} /2 μ$ , expand: $- E = D_{e} - a ℏ 2 D_{e} / μ (v + \frac{1}{2}) + \frac{a ^{2} ℏ ^{2}}{2 μ} (v + \frac{1}{2})^{2}$ . Identifying $ω = a 2 D_{e} / μ$ and shifting the zero to the well bottom yields the stated $E_{v}$ . The series terminates only while $λ - (v + \frac{1}{2}) > 0$ , giving a finite number of bound states with $v_{m a x} = ⌊ λ - \frac{1}{2} ⌋$ . The anharmonicity constant is $x_{e} = ℏ ω / (4 D_{e}) = a ℏ/ (2 2 μ D_{e})$ , recovering $ℏ ω x_{e} = ℏ^{2} a^{2} /2 μ$ . $□$

Proposition (codimension of a real conical intersection). For a real symmetric electronic Hamiltonian, the locus where two eigenvalues coincide has codimension two in nuclear configuration space.

Proof. Restrict $\hat{H}_{el} (R)$ to the two-dimensional space spanned by the two near-degenerate states. In a real orthonormal basis the restricted operator is a real symmetric $2 \times 2$ matrix $\begin{psmallmatrix} a & c \\ c & b\end{psmallmatrix}$ with eigenvalues $\frac{a + b}{2} \pm (\frac{a - b}{2})^{2} + c^{2}$ . Degeneracy requires the radical to vanish, which forces both $a - b = 0$ and $c = 0$ — two independent real conditions on the entries, each a function of $R$ . Two conditions cut a codimension-two locus from nuclear space, so for a molecule with $\geq 2$ internal nuclear coordinates the intersection is generically a manifold of codimension two (von Neumann-Wigner). Near such a point the upper and lower eigenvalues, as functions of the two tuning coordinates, trace a double cone, which is the conical intersection. $□$

Connections Master

Hartree-Fock self-consistent field method 12.09.03. The clamped-nucleus electronic Schrödinger equation $\hat{H}_{el} (R) ϕ_{k} = E_{k} (R) ϕ_{k}$ is precisely the problem a Hartree-Fock or post-Hartree-Fock calculation solves at each fixed molecular geometry; scanning the nuclear coordinates traces out the potential-energy surface $E_{k} (R)$ that this unit feeds into the nuclear equation, so the self-consistent field supplies the input to the Born-Oppenheimer machinery and the two units are the electronic and nuclear halves of one calculation.
Angular momentum operators and SU(2) 12.05.01. The rigid-rotor rotational levels $E_{J} = ℏ^{2} J (J + 1) /2 I$ are the spectrum of the squared angular-momentum operator on the rotational degrees of freedom of the molecule, and the same $J (J + 1)$ eigenvalue structure, selection rules, and $2 J + 1$ degeneracy that organise atomic angular momentum reappear as the rotational fine structure superimposed on each vibrational level of a diatomic.
Hilbert-space formalism 12.02.01. The full molecular state lives in the tensor product of the electronic and nuclear Hilbert spaces, and the Born-Oppenheimer ansatz is the statement that the ground state is well approximated by a single product vector in that tensor product; the derivative couplings $F_{k l}$ measure the entanglement between the electronic and nuclear factors that the product ansatz discards, connecting the approximation directly to the inner-product and tensor-product structure of the formalism.
Fermionic Fock space and Pauli exclusion 12.13.02. The electronic eigenstates $ϕ_{k} (r; R)$ are antisymmetric many-electron wavefunctions in the fermionic sector, so the ordering and symmetry of the molecular electronic terms (bonding versus antibonding, singlet versus triplet) are governed by the exclusion principle and the antisymmetry of the electronic Fock space, which is what determines whether a given $E_{k} (R)$ has a bound minimum at all.

Historical & philosophical context Master

Max Born and J. Robert Oppenheimer published the separation of electronic and nuclear motion in 1927 in Annalen der Physik ^{[Born-Oppenheimer 1927]}, within a year of Schrödinger's wave equation. Their decisive technical move was to identify the fourth root $κ = (m_{e} / M_{N})^{1/4}$ as the natural expansion parameter and to show that the molecular energy organises into successive even powers of $κ$ : electronic energy at $κ^{0}$ , vibrational quanta at $κ^{2}$ , rotational quanta at $κ^{4}$ . The choice of the fourth root, rather than the more obvious mass ratio itself, was what made the perturbation series converge in a controlled way, and it explained quantitatively why molecular spectra display the nested hierarchy of widely separated electronic bands, closer vibrational lines, and finely spaced rotational structure that spectroscopists had catalogued empirically.

Philipp Morse supplied the exactly solvable anharmonic model two years later, in 1929 in Physical Review ^{[Morse 1929]}, giving the closed-form vibrational spectrum $E_{v} = ℏ ω (v + \frac{1}{2}) - ℏ ω x_{e} (v + \frac{1}{2})^{2}$ that reproduced the observed crowding of levels toward dissociation and the finite count of bound states. The systematic adiabatic expansion, the precise status of the product ansatz, and the diagonal correction were placed on firmer footing by Born and Huang in the 1954 appendices to Dynamical Theory of Crystal Lattices ^{[Born-Huang 1954]}, which distinguished the crude-adiabatic from the adiabatic treatment and isolated the derivative couplings as the exact residual. Landau and Lifshitz present the diatomic molecule in Chapter XI of Quantum Mechanics, deriving the electronic terms, the vibrational and rotational structure, and the conditions for the validity of the separation ^{[Landau-Lifshitz 1977]}. The later recognition that the derivative couplings diverge at electronic degeneracies, and that encircling such a degeneracy imprints a sign change on the electronic wavefunction, connected the molecular problem to the geometric-phase structure that Berry would formalise in 1984, by which point conical intersections had become the accepted mechanism for ultrafast radiationless transitions in photochemistry.

Bibliography Master

@article{BornOppenheimer1927,
  author  = {Born, Max and Oppenheimer, J. Robert},
  title   = {Zur Quantentheorie der Molekeln},
  journal = {Annalen der Physik},
  volume  = {389},
  number  = {20},
  year    = {1927},
  pages   = {457--484},
  note    = {Vierte Folge, Band 84}
}

@article{Morse1929,
  author  = {Morse, Philip M.},
  title   = {Diatomic Molecules According to the Wave Mechanics. II. Vibrational Levels},
  journal = {Physical Review},
  volume  = {34},
  number  = {1},
  year    = {1929},
  pages   = {57--64}
}

@book{BornHuang1954,
  author    = {Born, Max and Huang, Kun},
  title     = {Dynamical Theory of Crystal Lattices},
  publisher = {Oxford University Press},
  year      = {1954},
  note      = {Appendices VII--VIII}
}

@article{Born1951Adiabatic,
  author  = {Born, Max},
  title   = {Kopplung der Elektronen- und Kernbewegung in Molekeln und Kristallen},
  journal = {Nachrichten der Akademie der Wissenschaften G{\"o}ttingen, Math.-Phys. Klasse IIa},
  year    = {1951},
  pages   = {1--3}
}

@book{LandauLifshitzQM1977,
  author    = {Landau, Lev D. and Lifshitz, Evgeny M.},
  title     = {Quantum Mechanics: Non-Relativistic Theory},
  edition   = {3},
  publisher = {Pergamon Press},
  year      = {1977},
  note      = {Course of Theoretical Physics, Vol. 3, Ch. XI}
}

@article{vonNeumannWigner1929,
  author  = {von Neumann, John and Wigner, Eugene},
  title   = {{\"U}ber das Verhalten von Eigenwerten bei adiabatischen Prozessen},
  journal = {Physikalische Zeitschrift},
  volume  = {30},
  year    = {1929},
  pages   = {467--470}
}

@article{Berry1984,
  author  = {Berry, Michael V.},
  title   = {Quantal Phase Factors Accompanying Adiabatic Changes},
  journal = {Proceedings of the Royal Society of London A},
  volume  = {392},
  number  = {1802},
  year    = {1984},
  pages   = {45--57}
}

@article{MeadTruhlar1982,
  author  = {Mead, C. Alden and Truhlar, Donald G.},
  title   = {Conditions for the Definition of a Strictly Diabatic Electronic Basis for Molecular Systems},
  journal = {Journal of Chemical Physics},
  volume  = {77},
  number  = {12},
  year    = {1982},
  pages   = {6090--6098}
}

@article{HagedornJoye2001,
  author  = {Hagedorn, George A. and Joye, Alain},
  title   = {A Time-Dependent Born-Oppenheimer Approximation with Exponentially Small Error Estimates},
  journal = {Communications in Mathematical Physics},
  volume  = {223},
  number  = {3},
  year    = {2001},
  pages   = {583--626}
}

@book{BransdenJoachain2003,
  author    = {Bransden, Brian H. and Joachain, Charles J.},
  title     = {Physics of Atoms and Molecules},
  edition   = {2},
  publisher = {Pearson},
  year      = {2003}
}

Prerequisites

12.02.01
12.05.01
12.13.02
12.09.03

Tier anchors

beginner: Griffiths, *Introduction to Quantum Mechanics*, 2nd ed. (Pearson, 2005), §7.3 (the hydrogen molecule ion); Atkins, P., *Physical Chemistry*, Ch. 12 (molecular spectra)
intermediate: Bransden, B. H. & Joachain, C. J., *Physics of Atoms and Molecules*, 2nd ed. (Pearson, 2003), Ch. 9; Atkins, P. & Friedman, R., *Molecular Quantum Mechanics*, 5th ed. (Oxford, 2011), Ch. 10
master: Landau, L. D. & Lifshitz, E. M., *Quantum Mechanics: Non-Relativistic Theory*, 3rd ed. (Pergamon, 1977), Ch. XI §§77–82; Born, M. & Huang, K., *Dynamical Theory of Crystal Lattices* (Oxford, 1954), Appendices VII–VIII

References

Born, M. & Oppenheimer, R. — Zur Quantentheorie der Molekeln · Annalen der Physik 84, 457–484 (1927) (the adiabatic separation and the $\kappa = (m_e/M_N)^{1/4}$ expansion)
Landau, L. D. & Lifshitz, E. M. — Quantum Mechanics: Non-Relativistic Theory · 3rd ed. (Pergamon, 1977), Ch. XI §§77–82 (the diatomic molecule; electronic terms, vibrational and rotational structure)
Morse, P. M. — Diatomic Molecules According to the Wave Mechanics. II. Vibrational Levels · Physical Review 34, 57–64 (1929) (the Morse potential and its exact vibrational spectrum)
Born, M. & Huang, K. — Dynamical Theory of Crystal Lattices · Oxford University Press (1954), Appendices VII–VIII (the systematic adiabatic expansion and the diagonal correction)
Born, M. — Kopplung der Elektronen- und Kernbewegung in Molekeln und Kristallen · Nachr. Akad. Wiss. Göttingen, Math.-Phys. Kl. IIa, 1–3 (1951) (the adiabatic / crude-adiabatic distinction)

Estimated time

beginner: 18m
intermediate: 45m
master: 80m