Standard Density Functional Theory (DFT) is remarkably successful at modeling covalent, metallic, and ionic bonds. However, standard local and semilocal exchange-correlation functionals—such as the Local Density Approximation (LDA) or the Generalized Gradient Approximation (GGA)—suffer from a fundamental blindspot: they fail to capture long-range, non-local electron correlations. These correlations give rise to London dispersion forces, the attractive component of van der Waals (vdW) interactions.
Because semilocal functionals only evaluate the electron density and its local gradient at a specific point in space, they cannot perceive the instantaneous, mutually induced dipoles that fluctuate between separated, non-overlapping electron clouds. As a result, standard DFT severely underestimates or completely omits these weak attractive forces. This omission leads to significant structural errors, such as overestimating the interlayer spacing of layered materials, miscalculating the density of liquids, or incorrectly predicting the phase stability of molecular crystals and polymers.
To fix this blindspot without incurring the prohibitive computational cost of high-level wave function methods, researchers use dispersion-corrected DFT.
Semiclassical Pairwise Corrections and Damping
The most computationally efficient strategy to capture these missing forces is the semiclassical pairwise approach, exemplified by Grimme's D3 correction. In this framework, the total energy of the system is calculated by adding an empirical or semi-empirical energy term ($E_{\text{disp}}$) directly to the standard DFT energy ($E_{\text{DFT}}$):
$$E_{\text{total}} = E_{\text{DFT}} + E_{\text{disp}}$$
The dispersion term is typically expanded as a power series of inverse interatomic distances, dominated by the leading-order term $-C_6/R^6$, where $C_6$ represents the dispersion coefficient between an atom pair and $R$ is their distance. As noted by Marcus Elstner et al. in Improving intermolecular interactions in DFTB3 using extended polarization from chemical-potential equalization, integrating these corrections is vital for capturing non-covalent forces in larger systems where base functionals fail.
However, a pure $-1/R^6$ term poses a physical problem: as two atoms get closer, the term approaches infinity. This would unphysically distort short-range covalent bonds that standard DFT already models accurately. To prevent this, a damping function must be introduced to smoothly turn off the dispersion correction at short ranges.
Justin M. Notestein et al. demonstrate in Comprehensive Phase Diagrams of MoS2 Edge Sites Using Dispersion-Corrected DFT Free Energy Calculations that combining dispersion corrections with Becke-Johnson damping significantly improves predictions of interlayer spacing and surface stability. Semiclassical corrections offer a highly scalable solution; as Christopher J. Stein shows in Semiclassical Dispersion Corrections Efficiently Improve Multiconfigurational Theory with Short-Range Density-Functional Dynamic Correlation, adding these terms allows models to capture attractive forces between distant atoms efficiently, bypassing the high cost of multireference perturbation theory.
Nonlocal Van der Waals Functionals
While pairwise corrections work well, they rely on atom-centered parameters. A more physically rigorous alternative is to modify the electronic functional itself by introducing a nonlocal correlation energy term that depends directly on the electron density rather than atomic coordinates.
Nonlocal van der Waals density functionals (such as vdW-DF) evaluate a double spatial integral over a nonlocal kernel:
$$E_{\text{c}}^{\text{nl}} = \frac{1}{2} \iint n(\mathbf{r}) \phi(\mathbf{r}, \mathbf{r}') n(\mathbf{r}') d\mathbf{r} d\mathbf{r}'$$
Here, the kernel $\phi$ accounts for the interaction between density fluctuations at positions $\mathbf{r}$ and $\mathbf{r}'$. This allows the functional to seamlessly capture dispersion forces directly from the electron distribution, making it highly adaptable to complex, shifting bonding environments.
In Origins of the structural phase transitions in MoTe2 and WTe2, Hyun‐Jung Kim et al. show that high-precision modeling of these nonlocal interactions is essential for accurately predicting the tiny energy barriers that govern structural phase transitions in layered materials. Similarly, Nicolas Mounet et al. leveraged these specialized vdW functionals in Two-dimensional materials from high-throughput computational exfoliation of experimentally known compounds to screen thousands of bulk materials for potential 2D exfoliation. They used the structural discrepancy between vdW-inclusive and standard relaxations as a clear signature to isolate compounds held together by weak dispersive forces.
Beyond Pairs: The Many-Body Response
Simple pairwise models assume that the dispersion interaction between two atoms is completely independent of the surrounding environment. However, in highly polarizable materials, the electrons respond collectively to fluctuations, creating a screening effect that alters the effective interactions.
To address this, Many-Body Dispersion (MBD) models treat the system as a collection of coupled quantum harmonic oscillators. This approach captures the collective electronic response of the entire system rather than treating interactions as purely additive pairs.
The importance of this mechanism is highlighted by Min Fang et al. in Revisiting the anchoring behavior in lithium-sulfur batteries: many-body effect on the suppression of shuttle effect. The authors found that standard additive models overestimate binding energies on highly polarizable materials like MXenes because they ignore collective screening effects. By accounting for the many-body response, MBD provides a more accurate representation of non-covalent binding in complex chemical environments.
Bridging the Accuracy Gap
Ultimately, dispersion corrections allow DFT to act as a bridge between computationally cheap but incomplete local models and high-level quantum chemistry methods like coupled-cluster theory. As Tamara Husch et al. demonstrate in Calculation of Ligand Dissociation Energies in Large Transition-Metal Complexes, modern dispersion corrections enable various DFT models to converge toward high-level coupled-cluster results for large molecular systems. Without these corrections, electronic structure simulations remain fundamentally blind to the subtle, long-range forces that dictate the architecture of condensed matter.
Key Foundational Work
Ab initiostudy of the vibrational properties of crystalline<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi>TeO</mml:mi><mml:mn>2</mml:mn></mml:msub></mml:mrow></mml:math>: The<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi>α</mml:mi></mml:mrow></mml:math>,<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi>β</mml:mi></mml:mrow></mml:math>, and<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi>γ</mml:mi></mml:mrow></mml:math>phases (2006) — contributed 1 concept elements to this direction.
Inelastic diffraction and spectroscopy of very weakly bound clusters (2002) — contributed 1 concept elements to this direction.