Inferensys

Glossary

Potential Energy Surface (PES)

A Potential Energy Surface (PES) is a mathematical function that maps the energy of a molecular system as a function of its atomic coordinates, defining the fundamental landscape upon which all chemical reactions and molecular motions occur.
Developer building agentic RAG system, retrieval pipeline diagram on laptop, technical workspace with notes.
FOUNDATIONAL CONCEPT

What is Potential Energy Surface (PES)?

The Potential Energy Surface (PES) is a mathematical function that defines the energy of a molecular system as a function of its atomic coordinates, forming the fundamental landscape upon which all chemical reactions and molecular motions occur.

A Potential Energy Surface (PES) is a hypersurface that maps the total electronic energy of a molecule or collection of atoms against their geometric configuration. For a system with N atoms, the PES is a function of 3N-6 internal coordinates, where each point represents a unique molecular geometry and its associated energy. Stable molecules correspond to local minima on this surface, while transition states—the fleeting configurations through which bonds break and form—are first-order saddle points.

The PES is the central object of theoretical chemistry, derived from solving the electronic Schrödinger equation under the Born-Oppenheimer approximation, which separates fast electronic motion from slow nuclear motion. The forces acting on atoms are the negative gradient of the PES, driving molecular dynamics. Modern machine learning, particularly neural network potentials, aims to learn this high-dimensional function directly from quantum mechanical reference data, bypassing expensive electronic structure calculations to enable simulations at ab initio accuracy.

THE ENERGY LANDSCAPE

Key Features of a Potential Energy Surface

A Potential Energy Surface (PES) is the fundamental mathematical landscape governing all chemical phenomena. Understanding its key features is essential for interpreting reaction mechanisms, molecular vibrations, and spectroscopy.

01

Stationary Points

Points on the PES where the first derivative (gradient) of the energy with respect to all atomic coordinates is zero, meaning the net force on every atom vanishes.

  • Global Minimum: The absolute lowest energy point on the entire surface, representing the most stable conformation of the molecule.
  • Local Minimum: A basin or well on the surface corresponding to a metastable conformer or isomer. A molecule can be trapped here.
  • Saddle Point: A point that is a minimum in all directions except one, where it is a maximum. The highest energy point along the minimum energy path connecting two minima.
  • Transition State: A first-order saddle point that represents the critical bottleneck structure through which reactants must pass to become products.
0
Force at Stationary Points
02

Reaction Coordinate

A continuous, one-dimensional parameter that tracks the progress of a chemical reaction from reactants to products along the minimum energy path (MEP). It is not a simple bond length but a collective coordinate involving the concerted motion of many atoms.

  • The energy profile plotted along this coordinate is the familiar reaction profile diagram.
  • The maximum along this coordinate corresponds to the transition state.
  • The difference in energy between the transition state and the reactants defines the classical barrier height.
1D
Dimensionality of Path
03

Hessian Matrix & Normal Modes

The Hessian is the matrix of second derivatives of the energy with respect to atomic coordinates. Its eigenvalues and eigenvectors reveal the local curvature of the PES.

  • At a Minimum: All eigenvalues of the Hessian are positive, indicating the energy increases in every direction.
  • At a Transition State: Exactly one eigenvalue is negative, corresponding to the reaction coordinate. The associated eigenvector points toward reactants and products.
  • Harmonic Vibrational Frequencies: Calculated from the mass-weighted Hessian eigenvalues. An imaginary frequency (reported as a negative number) is the hallmark of a transition state.
3N-6
Vibrational Modes (non-linear)
04

Minimum Energy Path (MEP)

The steepest-descent path connecting a transition state to its two adjacent minima. It represents the kinetically most favorable route for a reaction if the system has negligible momentum.

  • Often calculated using the Intrinsic Reaction Coordinate (IRC) method, which follows the gradient downhill in mass-weighted coordinates.
  • The MEP confirms that a transition state correctly connects the intended reactants and products.
  • The curvature of the MEP can reveal dynamical effects, where the reaction path tunnels through a barrier or bypasses a shallow intermediate.
0 K
Theoretical Temperature of IRC
05

Conical Intersections

A multidimensional seam where two electronic potential energy surfaces become exactly degenerate (equal in energy). These are not isolated points but funnels that allow extremely fast, non-radiative transitions between electronic states.

  • Critical for understanding photochemistry, such as vision, photosynthesis, and DNA photoprotection.
  • At a conical intersection, the Born-Oppenheimer approximation breaks down completely.
  • Characterized by the branching space defined by two vectors: the gradient difference and the non-adiabatic coupling vector.
Femtoseconds
Timescale of Decay
06

Avoided Crossings

A region where two potential energy surfaces of the same symmetry approach each other very closely but do not touch, repelling each other due to the non-crossing rule for diatomic molecules or states of the same symmetry in polyatomics.

  • The energy gap at the closest approach is a measure of the electronic coupling between the two states.
  • The character of the wavefunction swaps rapidly across the avoided crossing, leading to abrupt changes in molecular properties like the dipole moment.
  • A key signature in electron transfer and charge separation processes.
> 0
Minimum Energy Gap
PES FUNDAMENTALS

Frequently Asked Questions

Clear, technical answers to the most common questions about potential energy surfaces, their mathematical structure, and their role in computational chemistry and machine learning.

A potential energy surface (PES) is a mathematical function that describes the electronic energy of a molecular system as a function of its atomic coordinates, operating within the Born-Oppenheimer approximation which separates nuclear and electronic motion. For a system with N atoms, the PES is a hypersurface in 3N-6 dimensions (3N-5 for linear molecules), where each point represents a specific molecular geometry and its corresponding energy. The fundamental equation is E = f(R₁, R₂, ..., Rₙ), where Rᵢ are the nuclear coordinates. The PES serves as the conceptual landscape upon which all chemical processes unfold—stable molecules occupy local minima, transition states sit at first-order saddle points, and reaction paths trace minimum energy pathways between minima. In practice, the PES is computed by repeatedly solving the electronic Schrödinger equation at discrete geometries using methods like Density Functional Theory (DFT) or Coupled Cluster, then interpolating between points. The forces acting on atoms are the negative gradient of the PES, F = -∇E, driving molecular dynamics simulations. Understanding the topography of the PES—its valleys, passes, and basins—is essential for predicting reaction rates, spectroscopic signatures, and thermodynamic properties.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.