Born-Oppenheimer Approximation: Fixing the Nuclei

What It Is and Where It Applies

The full molecular Hamiltonian couples every electron to every nucleus through Coulomb terms, and the nuclear kinetic-energy operator entangles the two sets of coordinates. Solving it exactly is intractable for anything beyond H₂⁺. The Born-Oppenheimer approximation cuts the knot by assuming the total wavefunction factorizes:

• Ψ(r, R) ≈ ψ_elec(r; R) · χ_nuc(R) — the electronic part depends on electron coordinates r and only parametrically on nuclear positions R (the semicolon signals this).
• The electrons "see" a static nuclear framework; the nuclei "see" only an averaged electronic energy.

This underpins essentially all of computational and physical chemistry: the very idea of molecular geometry, a bond length, a transition state, a reaction coordinate, or a spectroscopic normal mode presupposes a Born-Oppenheimer potential energy surface. Hartree-Fock, DFT, coupled cluster, and molecular dynamics all run on such a surface. The approximation is excellent for ground-state, closed-shell molecules far from electronic degeneracies — which covers most of organic and inorganic chemistry at room temperature.

The Derivation, Step by Step

Start with the molecular Schrödinger equation Ĥ Ψ = E Ψ, where Ĥ = T̂_nuc + T̂_elec + V̂_ee + V̂_en + V̂_nn (nuclear and electronic kinetic energy plus electron-electron, electron-nucleus, and nucleus-nucleus Coulomb terms).

• Step 1 — Clamp the nuclei. Drop T̂_nuc (set nuclear masses → ∞). The remainder is the electronic Hamiltonian Ĥ_elec, in which R appears as fixed numbers.
• Step 2 — Solve the electronic problem at each geometry: Ĥ_elec ψ_n(r; R) = E_n(R) ψ_n(r; R). The eigenvalue E_n(R), plus the constant V̂_nn(R), defines the potential energy surface U_n(R).
• Step 3 — Move the nuclei on that surface. Insert Ψ = ψ_n·χ into the full equation. Acting T̂_nuc on the product generates "extra" derivative terms — the non-adiabatic couplings — proportional to ⟨ψ_n|∇_R|ψ_m⟩.
• Step 4 — Neglect the couplings. Because they carry a factor ~1/M, they are tiny; dropping them yields the nuclear equation [T̂_nuc + U_n(R)] χ(R) = E χ(R).

Born's formal expansion parameter is κ = (m/M)^(1/4); electronic energies scale as κ⁰, vibrations as κ², rotations as κ⁴.

Key Quantities and a Worked Example

The whole scheme hinges on one number. For hydrogen, M/m = 1836, so κ = 1836^(−1/4) ≈ 0.152 — small enough that corrections at order κ⁴ ≈ 5×10⁻⁴ are typically sub-kJ/mol. The speed ratio is √1836 ≈ 42.8, so electrons complete tens of orbits before a nucleus shifts appreciably.

Worked example — H₂: Solve Ĥ_elec at a series of internuclear distances R and you trace a Morse-like PES. The minimum sits at the equilibrium bond length r_e ≈ 0.741 Å with well depth D_e ≈ 458 kJ/mol (4.75 eV). Diagonalizing the nuclear motion on that curve gives vibrational levels with ω_e ≈ 4401 cm⁻¹ and zero-point energy ≈ 26 kJ/mol, so the measured dissociation energy D₀ ≈ 432 kJ/mol (4.48 eV).

• The diagonal Born-Oppenheimer correction (DBOC) to the H₂ energy is only ~0.5 kJ/mol.
• H/D isotope effects (e.g., different zero-point energies) arise entirely from the nuclear step — the electronic PES is isotope-independent, a direct BO consequence.

How It Is Used in Practice

Every mainstream electronic-structure package is a Born-Oppenheimer engine. The workflow is always the same two-stage loop:

• Single point: fix R, solve Ĥ_elec (Hartree-Fock, DFT, MP2, CCSD(T)) → get U(R) and its gradient ∇U.
• Geometry / dynamics: move the nuclei — optimize to a minimum (∇U = 0, positive Hessian) or a transition state (one negative Hessian eigenvalue), or integrate Newton's equations in Born-Oppenheimer molecular dynamics (BOMD), re-solving the electronic problem at every step.

Spectroscopy leans on the same split: within BO, an electronic absorption band is dressed by vibrational structure via the Franck-Condon principle — nuclei are frozen during the ~femtosecond electronic transition, so intensities follow the overlap |⟨χ_final|χ_initial⟩|². Vibrational frequencies (IR/Raman) come from the PES Hessian; NMR shieldings and reaction barriers are PES-derived properties. Enzyme catalysis, combustion kinetics, and atmospheric photochemistry are all modeled as motion on Born-Oppenheimer surfaces (or, when that fails, on coupled surfaces).

Comparison to Related Concepts

The Born-Oppenheimer approximation is often conflated with its neighbors; the distinctions matter.

• Clamped-nucleus vs. adiabatic: The strict clamped-nucleus step (drop T̂_nuc entirely) gives U(R). The full adiabatic approximation keeps the diagonal nuclear-coupling term (the DBOC), so the nucleus follows a single surface but with a small mass-dependent correction. "BO" colloquially means the clamped/adiabatic family.
• Adiabatic vs. diabatic: Adiabatic states are the electronic eigenstates at each R (smooth energies, but couplings can spike). Diabatic states are rotated combinations chosen to make the derivative couplings vanish, trading them for smooth potential couplings — the natural basis near intersections.
• vs. Franck-Condon: Franck-Condon is a consequence of BO (fixed nuclei during electronic transitions), not a separate assumption.

Related mean-field ideas — the Hartree-Fock and Kohn-Sham approximations — operate inside the electronic step; they are orthogonal to, not substitutes for, the BO separation.

Exceptions, Breakdown, and Significance

The approximation fails precisely when the assumed energy gap between electronic surfaces collapses. Then the non-adiabatic coupling ⟨ψ_n|∇_R|ψ_m⟩ ∝ 1/(E_n − E_m) blows up, and a single surface no longer captures the physics.

• Conical intersections: two PESs touch in a genuine (not avoided) crossing, forming a double cone. The coupling literally diverges; nuclei funnel between electronic states on femtosecond timescales. This drives vision (retinal cis-trans isomerization), DNA photostability, and organic photochemistry.
• Avoided crossings & Jahn-Teller distortions in transition-metal and open-shell systems.
• Light-particle effects: proton-coupled electron transfer and hydrogen tunneling, where the "heavy" nucleus isn't heavy enough.

Historically, the BO paper (Annalen der Physik, 1927) made molecular quantum mechanics computable — it is arguably the enabling approximation of theoretical chemistry. High-accuracy work now goes beyond BO: explicitly correlated non-BO calculations on H₂ and its isotopologues reach spectroscopic accuracy (sub-cm⁻¹), and surface-hopping / Ehrenfest methods handle non-adiabatic dynamics where fixing the nuclei is no longer allowed.

Energy scales in a molecule as powers of the Born expansion parameter κ = (m/M)^(1/4), with typical magnitudes

Motion	Scaling with κ	Typical energy	Timescale
Electronic excitation	κ⁰ ≈ 1	1–10 eV (100–1000 kJ/mol)	~10⁻¹⁶ s (attoseconds–fs)
Nuclear vibration	κ² = (m/M)^(1/2)	0.05–0.5 eV (5–50 kJ/mol)	~10⁻¹⁴ s (tens of fs)
Molecular rotation	κ⁴ = m/M	10⁻⁴–10⁻³ eV	~10⁻¹¹ s (ps)
Non-adiabatic coupling (typical)	κ³ or higher	meV, often negligible	—
Coupling at conical intersection	diverges (→ ∞)	comparable to vibration	fs (ultrafast)