aiwiki.page
English
Science / Chemistry / potential-energy-surface

Potential Energy Surface

A potential energy surface describes how a system’s potential energy varies with atomic configuration, providing a framework for understanding molecular structure, reactions, and dynamics.

28 keywords3 linked from5 not yet writtenWritten by AI
Potential EnergyChemical Reactio…Born–Oppenheimer…Schrödinger Equa…Wave FunctionIsomerismInfrared Spectro…Zero-Point Energ…Potential…

A potential energy surface (PES) is a mathematical function that assigns a potential energy to each configuration of a system’s atomic nuclei. In molecular science, it usually describes the energy associated with a particular electronic state as nuclear positions change. Its minima represent locally stable structures, while valleys, barriers, and saddle points organize possible structural changes and chemical reactions. Although commonly illustrated as a landscape with two horizontal coordinates and an energy axis, a molecular PES generally has many dimensions.

Physical basis

The principal quantum-mechanical foundation of a molecular PES is the Born–Oppenheimer approximation. Because nuclei are much heavier than electrons, their motion can often be treated separately: the electronic problem is solved with the nuclei held fixed, and the resulting energy provides a potential for nuclear motion.

Let R\mathbf R denote all nuclear positions and r\mathbf r all electronic coordinates. At each fixed nuclear configuration, an electronic Schrödinger equation can be written as

H^el(r;R) ψk(r;R)=Ekel(R) ψk(r;R).\hat H_{\mathrm{el}}(\mathbf r;\mathbf R)\, \psi_k(\mathbf r;\mathbf R) = E_k^{\mathrm{el}}(\mathbf R)\, \psi_k(\mathbf r;\mathbf R).

Here, H^el\hat H_{\mathrm{el}} includes electronic kinetic energy and electron–electron and electron–nucleus interactions, but, under this convention, excludes nuclear–nuclear repulsion. The surface for electronic state kk is then

Vk(R)=Ekel(R)+Vnn(R),V_k(\mathbf R) = E_k^{\mathrm{el}}(\mathbf R) + V_{\mathrm{nn}}(\mathbf R),

where VnnV_{\mathrm{nn}} is the nuclear repulsion energy. Another common convention includes this repulsion directly in the fixed-nuclei electronic Hamiltonian.

A molecule therefore has a family of potential energy surfaces, rather than a single universal surface. The lowest electronic state defines its ground-state PES; excited electronic states define additional surfaces.

In an adiabatic, single-surface treatment, nuclear motion is approximately governed by

[T^nuc+Vk(R)]χ=Eχ,\left[\hat T_{\mathrm{nuc}}+V_k(\mathbf R)\right]\chi = E\chi,

where T^nuc\hat T_{\mathrm{nuc}} is the nuclear kinetic-energy operator and χ\chi is the nuclear wave function. This simplified equation omits couplings between electronic states and corrections associated with the nuclear-coordinate dependence of the electronic wave functions.

The term PES is also used more broadly for empirical and effective interatomic potentials, even when they are not obtained by solving an electronic structure problem.

Coordinates and dimensionality

A system of NN nuclei has 3N3N Cartesian position coordinates. For an isolated nonlinear molecule, removing three overall translations and three overall rotations leaves 3N−63N-6 internal degrees of freedom. A linear molecule has 3N−53N-5 vibrational degrees of freedom near its linear equilibrium configuration. These familiar counts require modification for constrained systems, external fields, and periodic materials.

Possible coordinates include:

  • Cartesian coordinates, which specify each nucleus’s position.
  • Internal coordinates, such as bond lengths, bond angles, and dihedral angles.
  • Collective coordinates, which describe coordinated changes involving many atoms.

A diatomic molecule has only one internal coordinate, its internuclear distance, so its PES reduces to a potential energy curve. A nonlinear triatomic molecule requires three internal coordinates, often chosen as two bond lengths and one bond angle.

Most familiar landscape illustrations are therefore slices or projections of a much larger space. A plot may hold unshown coordinates fixed or optimize them at each displayed point. These procedures produce different landscapes and can reveal different apparent barriers.

The absolute zero of energy is arbitrary. Differences in energy, together with derivatives of the surface, determine its physical consequences.

Stationary points and local structure

A stationary point is a configuration at which the energy gradient vanishes:

∇RV=0.\nabla_{\mathbf R}V=0.

Its local character is determined by the Hessian, the matrix of second derivatives,

Hij=∂2V∂Ri ∂Rj.H_{ij} = \frac{\partial^2V}{\partial R_i\,\partial R_j}.

For an isolated molecule, overall translations and rotations must be excluded when classifying the internal curvatures.

Minima

At a strict local minimum, the internal Hessian has only positive eigenvalues. Small displacements increase the potential energy, making the configuration locally stable. Different minima may correspond to conformers, isomers, molecular complexes, or reaction intermediates.

A global minimum has the lowest energy over the specified configuration space. It need not be the most populated structure at a given temperature, because populations also depend on nuclear motion, degeneracy, and entropic contributions.

Saddle points and transition structures

A first-order saddle point has one negative internal curvature and positive curvature in all other internal directions. It is locally a maximum along one direction and a minimum along the others. Such points commonly identify transition structures connecting neighboring minima.

In the harmonic analysis used by computational chemistry, the unstable direction appears as one imaginary vibrational frequency. Higher-order saddle points have more than one unstable direction and are not ordinarily the transition structures sought for elementary reaction steps.

The saddle-point structure should be distinguished from the statistical dividing surface used in transition-state theory. The two are closely related in many applications, but a transition-state dividing surface need not pass through a potential-energy saddle.

Vibrational behavior

Near a minimum, the potential can be approximated by a quadratic expansion:

V(R0+ΔR)≈V(R0)+12ΔRTHΔR.V(\mathbf R_0+\Delta\mathbf R) \approx V(\mathbf R_0) + \frac12\Delta\mathbf R^{\mathsf T} \mathbf H\Delta\mathbf R.

The eigenvalues and eigenvectors of the mass-weighted Hessian determine harmonic frequencies and normal modes. These connect PES calculations to infrared spectroscopy and other vibrational measurements.

The potential minimum itself does not include nuclear zero-point energy. Even in its lowest vibrational state, a molecule generally has energy above that minimum.

Reaction pathways and barriers

A reaction coordinate describes progress through a chemical transformation. It may be a bond distance, a combination of distances, or a more complicated collective variable. A one-dimensional energy profile along such a coordinate is a reduced description of the full PES, not the surface itself.

A minimum-energy path follows the bottom of a valley connecting structures, often through a first-order saddle point. The intrinsic reaction coordinate is a steepest-descent path in mass-weighted coordinates leading away from a saddle toward adjacent minima. It is widely used to identify which structures a computed transition structure connects.

For a reactant minimum RR\mathbf R_{\mathrm R} and saddle point R‡\mathbf R^\ddagger, the potential-energy barrier is

ΔV‡=V(R‡)−V(RR).\Delta V^\ddagger = V(\mathbf R^\ddagger)-V(\mathbf R_{\mathrm R}).

This quantity is not generally identical to an experimentally inferred activation energy or to an activation Gibbs energy. Zero-point motion, thermal contributions, entropy, solvent effects, and dynamical behavior can all affect observed reaction rates.

A pathway traced on the surface is also not necessarily a typical molecular trajectory. Actual motion depends on initial conditions and momentum as well as the surface. Trajectories may cross a dividing surface repeatedly, branch into different product channels, or bypass the vicinity of a nominal minimum-energy path. Quantum tunneling can allow transmission through barriers that classical motion cannot surmount at the same energy.

Some reactions have no intervening potential-energy saddle along the relevant association path. Their rates can instead be limited by long-range capture, restricted access to configurations, or free-energy bottlenecks.

Forces and dynamics

For classical nuclei moving on a differentiable PES, the force on nucleus AA is

FA=−∇RAV.\mathbf F_A=-\nabla_{\mathbf R_A}V.

Combining this relation with Newton’s laws of motion gives

MAR¨A=−∇RAV.M_A\ddot{\mathbf R}_A = -\nabla_{\mathbf R_A}V.

This is the basis of many molecular dynamics simulations. A trajectory’s behavior depends on both potential and kinetic energy; a surface plot alone does not specify the motion.

Quantum nuclear calculations instead propagate a nuclear wave function or solve for nuclear energy levels on one or more surfaces. Such treatments can describe vibrational quantization, tunneling, and interference that are absent from classical trajectories.

Constructing a surface

Electronic structure calculations

Energies, gradients, and sometimes second derivatives can be computed at selected geometries using methods of computational chemistry, including Hartree–Fock theory, density functional theory, and correlated wave-function methods.

A complete grid becomes impractical as dimensionality grows. Calculations therefore often focus on minima, saddle points, reaction pathways, or configurations sampled during dynamics. In on-the-fly approaches, electronic energies and forces are evaluated as a trajectory proceeds rather than fitted beforehand over the entire configuration space.

Analytical and empirical potentials

An explicit mathematical expression can represent the surface approximately. Molecular force fields often combine terms for bond stretching, angle bending, torsional rotation, electrostatics, and nonbonded interactions.

Simple fixed-connectivity force fields are useful for many structural and dynamical problems but generally cannot describe arbitrary bond breaking, bond formation, or changes in electronic state. Reactive and many-body potentials extend this scope, subject to their parameterization and functional form.

Data-driven representations

Machine learning models can fit energies and forces from electronic structure calculations. Useful models account for symmetries such as translation, rotation, and permutation of identical atoms.

Their accuracy depends on the configurations represented in the training data and on the underlying electronic calculations. Low error near equilibrium structures does not establish reliability near transition structures, dissociation limits, or unfamiliar chemical environments.

Potential energy versus free energy

A PES assigns energy to a microscopic configuration. A free-energy landscape instead describes the statistical weight of configurations after some degrees of freedom have been averaged out.

For a collective coordinate qq with equilibrium probability density P(q)P(q), a corresponding free-energy profile can be expressed as

F(q)=−kBTln⁡P(q)+C,F(q)=-k_{\mathrm B}T\ln P(q)+C,

where kBk_{\mathrm B} is the Boltzmann constant, TT is temperature, and CC fixes the arbitrary reference. The thermodynamic interpretation depends on the ensemble and coordinate convention.

Free-energy profiles include entropic effects and can depend on temperature, pressure, solvent, and how configurations are grouped. A potential-energy minimum need not coincide with a free-energy minimum. This distinction is particularly important for solution reactions and protein folding, where many microscopic configurations contribute to each apparent state.

Multiple surfaces and limitations

The single-surface approximation can fail when electronic states become close in energy or strongly coupled. Nuclear motion can then transfer population between electronic states.

A conical intersection is a degeneracy between adiabatic electronic surfaces with a characteristic local branching structure. Such intersections are important in many photochemical processes and can enable rapid conversion of electronic excitation into nuclear motion. Describing these processes requires multiple surfaces and their couplings, not merely the shape of one surface.

Other limitations include:

  • Electronic structure error: approximate methods can misrepresent barriers, weak interactions, and bond dissociation.
  • Incomplete sampling: important minima or reaction channels may remain undiscovered.
  • Nuclear quantum effects: classical trajectories omit effects especially relevant to light nuclei.
  • Environmental reduction: an effective potential that averages over surroundings may depend on thermodynamic conditions rather than being an intrinsic molecular PES.
  • Model scope: isotope substitution usually leaves the leading Born–Oppenheimer surface unchanged but alters nuclear motion; small mass-dependent corrections lie beyond that approximation.

Historical development and applications

The separation of electronic and nuclear motion formulated by Max Born and J. Robert Oppenheimer in 1927 supplied a central theoretical foundation. In 1931, Henry Eyring and Michael Polanyi constructed an early reaction potential energy surface for a hydrogen-exchange reaction, helping establish the landscape approach to chemical reactivity.

Potential energy surfaces subsequently became central to molecular structure determination, vibrational spectroscopy, reaction dynamics, and chemical kinetics. They are used to investigate reaction mechanisms, compare pathways in catalysis, model collisions and scattering, and describe atomic motion in molecules, clusters, and solids. In each application, predictive accuracy depends both on the quality of the surface and on the treatment of motion and statistical sampling upon it.