Potential Energy Surface: Valleys, Passes, and Reaction Paths

What the potential energy surface is and where it applies

The PES arises directly from the Born–Oppenheimer approximation (Max Born and J. Robert Oppenheimer, 1927). Because nuclei are roughly 1,800× heavier than electrons, the electrons adjust essentially instantaneously to any nuclear configuration. So you can freeze the nuclei, solve the electronic Schrödinger equation for that fixed geometry, and get a single energy. Repeat over all geometries and you trace out E as a smooth function of the nuclear coordinates — the surface on which the nuclei then move.

For a nonlinear molecule of N atoms there are 3N−6 internal degrees of freedom (bond lengths, angles, dihedrals), so the true PES lives in a high-dimensional space we can only plot in slices. Where it matters:

• Kinetics: barrier heights set activation energies and rate constants.
• Mechanism: the connectivity of valleys and passes is the mechanism.
• Spectroscopy: curvature at a minimum gives vibrational frequencies.
• Structure: the deepest valley is the equilibrium geometry.

Every equilibrium bond length and every reaction mechanism you draw is a claim about this landscape.

The mechanism: from valleys to a saddle-point pass

A stationary point is any geometry where the gradient vanishes: ∇E = 0, meaning the net force on every nucleus is zero. To classify it you build the Hessian — the matrix of second derivatives ∂²E/∂qᵢ∂qⱼ — and diagonalize it. Its eigenvalues (mass-weighted, these become the vibrational force constants) tell you the local shape:

• All eigenvalues positive → minimum. Every direction curves upward; it's a stable valley (reactant, product, or intermediate).
• Exactly one negative eigenvalue → first-order saddle point. The surface curves up in every direction but one, and curves down along that single direction — the reaction coordinate. This is the transition state, the mountain pass.

The negative eigenvalue corresponds to an imaginary vibrational frequency (since ν ∝ √k and the force constant k is negative). A genuine transition state has exactly one imaginary frequency, and its motion visualizes the bond-making/bond-breaking. Following the steepest-descent path downhill from the saddle in mass-weighted coordinates traces the intrinsic reaction coordinate (IRC, Fukui, 1970), rigorously connecting the TS to its two flanking minima.

Key quantities and a worked example

The chemically important numbers are all energy differences read off the surface. The classical barrier height is E‡ = E(saddle) − E(reactant minimum); adding zero-point vibrational energy gives the 0 K activation energy.

Worked example — F + H₂ → HF + H, a benchmark PES:

• Classical barrier: ≈ 6.6 kJ/mol (1.6 kcal/mol) — a low, early pass.
• Reaction exoergicity: ΔE ≈ −140 kJ/mol — a steep drop into the product valley, which is why F + H₂ is a classic chemical laser reaction.
• The saddle is reactant-like (early transition state), consistent with Hammond's postulate for a strongly exoergic step.

Contrast the symmetric H + H₂ exchange: barrier ≈ 40 kJ/mol with a linear (collinear) transition state, H···H···H, and ΔE = 0 by symmetry, so the saddle sits exactly midway. Curvature at a minimum gives frequencies: for H₂ the well depth (De ≈ 458 kJ/mol) and curvature reproduce its 4401 cm⁻¹ stretch. These few numbers — barrier, reaction energy, and curvatures — feed transition-state theory to predict the rate.

How the surface is measured and used in practice

You almost never have the full analytic PES; instead you compute the points you need with quantum chemistry. Typical workflow:

• Geometry optimization walks downhill (following −∇E) to locate minima — modern methods use analytic gradients and BFGS-type updates of the Hessian.
• Transition-state search (e.g. the Berny/eigenvector-following algorithm, or the nudged elastic band / string method) climbs to the first-order saddle.
• Frequency analysis confirms the stationary point: 0 imaginary frequencies for a minimum, exactly 1 for a TS.
• IRC calculation verifies the saddle connects the intended reactant and product valleys.

Energies come from density functional theory (B3LYP, ωB97X-D, ~2–5 kcal/mol accuracy) up to gold-standard CCSD(T) (~1 kcal/mol, "chemical accuracy"). Barrier heights then plug into Eyring transition-state theory, k = (k_B·T/h)·exp(−ΔG‡/RT), to give rate constants. Experimentally, molecular-beam scattering and femtosecond pump–probe (Zewail's femtochemistry, Nobel 1999) probe motion across the surface in real time, testing the computed passes.

How the PES relates to nearby concepts

The PES is the parent object; several familiar diagrams are just projections of it.

• Reaction coordinate diagram: the 1-D energy-vs-progress curve every textbook draws is a cut through the multidimensional PES along the minimum-energy path.
• Activation energy: the Arrhenius Ea is closely related to the barrier height E‡ read from the surface (differing by thermal and ZPE corrections).
• Gibbs free energy surface: add entropy and temperature and the potential energy surface becomes a free-energy surface; TS theory is built on the free-energy saddle, which can shift with T.
• Potential energy curve: the diatomic Morse or Lennard-Jones curve is simply the 1-D special case (3N−6 = 1 for N=2 diatomics is 3·2−5 = 1).

Crucially, the PES is independent of temperature and mechanism assumptions — it's pure electronic structure. Kinetics, thermodynamics, and spectroscopy are all derived from it, which is why it is the true starting point.

Exceptions, breakdowns, and famous cases

The single-surface picture is powerful but has real limits:

• Conical intersections: when two electronic states become degenerate, surfaces touch in a cone and the Born–Oppenheimer approximation fails. These "funnels" govern ultrafast photochemistry — vision (retinal isomerization in rhodopsin, <100 fs) and DNA photoprotection both route through conical intersections, not classical saddles.
• Barrierless reactions: many radical–radical recombinations and ion–molecule reactions have no saddle at all; the path runs monotonically downhill and rates are governed by long-range capture, not a pass.
• Bifurcating (valley-ridge) surfaces: a single TS can lead to two products when the path splits after the saddle — dynamics, not the static PES, then decide the ratio, breaking simple TS theory.
• Roaming reactions: in H₂CO photodissociation, trajectories "roam" far from the minimum-energy path to give H₂ + CO, discovered spectroscopically in 2004.

The historic F + H₂ and H + H₂ surfaces (Eyring, Polanyi, and later high-accuracy ab initio work) remain the proving grounds where computed passes are checked against molecular-beam experiments atom by atom.

Stationary points on a potential energy surface, classified by the number of negative eigenvalues (imaginary frequencies) of the Hessian (second-derivative matrix).

Feature	Gradient ∇E	Negative Hessian eigenvalues	Imaginary frequencies	Chemical meaning
Minimum (valley)	0	0	0	Stable reactant, product, or intermediate
First-order saddle (pass)	0	1	1	Transition state connecting two minima
Second-order saddle	0	2	2	Not a valid TS; artifact or higher symmetry point
Maximum (peak)	0	3N−6	3N−6	All directions downhill; never a reaction path
Non-stationary point	≠ 0	n/a	n/a	Generic point along a slope