4  Force Fields

4.1 Force Fields

A force field (or interatomic potential) is an analytical expression for the potential energy surface \(U(\mathbf{r}_1, \ldots, \mathbf{r}_N)\) that appears in Newton’s equations of motion (Eq. 3.1, Chapter 2).

4.2 Functional Forms

A typical classical force field decomposes \(U\) into bonded and non-bonded contributions:

\[ U = U_{\text{bond}} + U_{\text{angle}} + U_{\text{dihedral}} + U_{\text{vdW}} + U_{\text{elec}} \tag{4.1}\]

4.2.1 Pair Potentials

The simplest models use pair potentials — \(U\) depends only on interatomic distances \(r_{ij} = |\mathbf{r}_i - \mathbf{r}_j|\):

\[ U = \sum_{i < j} \phi(r_{ij}) \tag{4.2}\]

Lennard-Jones Potential

The Lennard-Jones (LJ) potential models van der Waals interactions:

\[ \phi_{\text{LJ}}(r) = 4\varepsilon \left[\left(\frac{\sigma}{r}\right)^{12} - \left(\frac{\sigma}{r}\right)^{6}\right] \tag{4.3}\]

where \(\varepsilon\) is the well depth and \(\sigma\) is the collision diameter (the distance at which \(\phi = 0\)). The minimum occurs at \(r_{\min} = 2^{1/6}\sigma\).

Embedded Atom Method (EAM)

For metals, the Embedded Atom Method (EAM) is more accurate than pair potentials. The energy is:

\[ U = \sum_i F_i(\bar{\rho}_i) + \frac{1}{2}\sum_{i \neq j} \phi(r_{ij}) \tag{4.4}\]

where \(F_i\) is the embedding energy, \(\bar{\rho}_i = \sum_{j\neq i} \rho(r_{ij})\) is the local electron density, and \(\phi(r_{ij})\) is a short-range pair repulsion.

NoteComing up

Chapter 7 will revisit the limitations of classical force fields like Eq. 4.3 and Eq. 4.4, and show how machine learning potentials overcome them by learning \(U\) directly from quantum mechanical data.

4.3 Cutoffs and Periodic Boundary Conditions

This section is under construction.

Exercises

  1. Plot Eq. 4.3 for argon (\(\varepsilon = 0.0104\) eV, \(\sigma = 3.40\) Å). At what distance is the force zero? At what distance is the potential minimum?

  2. Compute the force \(F = -d\phi_{\text{LJ}}/dr\) and show that it is repulsive for \(r < r_{\min}\) and attractive for \(r > r_{\min}\).