How Geometry Optimization Works¶
A Geometry Optimization calculation aims to determine the most stable arrangement of atoms in a molecule by locating a minimum on the Potential Energy Surface (PES). Unlike a Single Point Energy calculation, which evaluates the electronic energy for a fixed molecular structure, a Geometry Optimization allows the atomic positions to change until the molecular energy reaches its lowest possible value.
The optimization process is iterative. Starting from an initial molecular geometry, Gaussian repeatedly calculates the electronic energy, determines the forces acting on the atoms, and adjusts the atomic coordinates to move the structure toward a lower-energy configuration. This cycle continues until the geometry satisfies the predefined convergence criteria.
The Potential Energy Surface¶
Every possible arrangement of atoms in a molecule corresponds to a unique electronic energy. If all these energies are plotted as a function of the nuclear coordinates, the result is known as the Potential Energy Surface (PES).
For a simple diatomic molecule, the potential energy surface can be visualized as a curve.
Energy
^
|
| * *
| * *
| * *
| * *
| * *
| *
+------------------------------> Bond Length
Minimum Energy
The lowest point on the curve represents the equilibrium bond length, where the molecule is most stable.
For larger molecules containing many atoms, the potential energy surface becomes multidimensional, making it impossible to visualize directly. Nevertheless, the objective remains the same: locate a minimum on this surface.
Starting from an Initial Geometry¶
The optimization begins with the molecular geometry provided in the input file.
This initial structure may
- be drawn using GaussView,
- be obtained from an experimental crystal structure,
- come from a molecular database,
- or be generated from a previous calculation.
The initial geometry is only a starting guess and is not necessarily the most stable structure.
Evaluating the Electronic Energy¶
For the current molecular geometry, Gaussian performs a complete Self-Consistent Field (SCF) calculation to determine the electronic energy.
This energy represents the position of the molecule on the potential energy surface.
If the geometry is far from equilibrium, the energy will generally be higher than the minimum energy.
Calculating the Forces on the Atoms¶
Once the electronic energy has been determined, Gaussian calculates the forces acting on every atom.
These forces indicate how each atom should move to reduce the total energy.
For example,
- atoms that are too close together experience strong repulsive forces,
- atoms that are too far apart may experience attractive forces,
- distorted bond angles generate restoring forces that move the molecule toward a more stable geometry.
The calculated forces provide the direction in which the geometry should be modified.
Updating the Molecular Geometry¶
Using the calculated forces, Gaussian generates a new set of atomic coordinates.
The displacement is usually small to ensure that the optimization remains stable.
The updated geometry becomes the starting point for the next optimization cycle.
Repeating the Optimization Cycle¶
Geometry optimization is not completed in a single step.
Instead, Gaussian repeatedly performs the following sequence:
Initial Geometry
│
▼
SCF Calculation
│
▼
Electronic Energy
│
▼
Calculate Forces
│
▼
Update Geometry
│
▼
New Geometry
│
▼
Repeat
With each iteration, the molecular geometry moves closer to the minimum on the potential energy surface.
Monitoring Convergence¶
After every optimization step, Gaussian checks whether the optimization has converged.
Several quantities are monitored, including
- the maximum force,
- the root-mean-square (RMS) force,
- the maximum atomic displacement,
- the RMS displacement.
When all of these quantities fall below their respective convergence thresholds, the optimization is considered complete.
Gaussian then prints the message
indicating that a stationary point has been successfully located.
The Optimized Structure¶
The optimized geometry represents a stationary point on the potential energy surface.
For most calculations, this stationary point corresponds to a local minimum, meaning that small movements of the atoms in any direction increase the molecular energy.
However, a Geometry Optimization alone cannot determine whether the stationary point is a minimum or a transition state.
For this reason, an optimized geometry is usually followed by a Frequency Calculation.
- No imaginary frequencies indicate a local minimum.
- One imaginary frequency indicates a first-order transition state.
- More than one imaginary frequency usually indicates that the optimization has not reached the desired stationary point.
The interpretation of vibrational frequencies is discussed in the Frequency Analysis section of this documentation.
Geometry Optimization Workflow¶
The complete optimization process can be summarized as
Read Initial Geometry
│
▼
Perform SCF Calculation
│
▼
Calculate Electronic Energy
│
▼
Calculate Atomic Forces
│
▼
Move the Atoms
│
▼
Generate New Geometry
│
▼
Convergence Achieved?
│ │
│ No │ Yes
▼ ▼
Repeat Optimized Geometry
Each optimization cycle gradually lowers the molecular energy until the convergence criteria are satisfied.
Summary¶
Geometry Optimization is an iterative process that searches for a stable molecular structure by repeatedly evaluating the electronic energy, calculating the forces acting on the atoms, and updating the molecular geometry. Starting from an initial guess, Gaussian moves the atoms step by step toward a stationary point on the potential energy surface. Once the convergence criteria for the forces and atomic displacements are satisfied, the optimized geometry is obtained. In practice, this optimized structure is typically verified by a subsequent Frequency Calculation to confirm that it corresponds to a true energy minimum.
Next Section¶
In the next chapter, we develop the mathematical framework behind Geometry Optimization. We introduce energy gradients, the Hessian matrix, Taylor series expansion of the potential energy surface, and the optimization algorithms used by Gaussian, including the Berny optimization method.