What Is A London Dispersion Force

Temporary dipoles that hold everything together

London dispersion forces are the weakest of the intermolecular interactions, but they are also the most universal. Every molecule with electrons experiences them, whether polar or nonpolar. They arise from momentary fluctuations in electron density that create temporary dipoles, which then induce dipoles in neighboring molecules. The attraction between those temporary dipoles is what you get when you strip away hydrogen bonding and dipole-dipole interactions and are left with nothing but the raw quantum mechanical noise of electrons moving around.

I spent years working with polymer processing and saw this force matter more than anything else when people tried to predict why certain nonpolar systems behaved completely differently than their simulations suggested. The DFT functionals we used routinely underestimated dispersion by a factor of two or three in stacked aromatic systems. That sounded impossible until you watched actual samples gel at concentrations the model said should have remained fluid. The technical name is the van der Waals attraction arising from correlated electron fluctuations. In modern computational chemistry it is often separated out explicitly because standard Hartree-Fock theory misses it entirely, and even many density functionals treat it poorly. The energy scales with the polarizability of the molecules involved and falls off as the sixth power of distance, which is why it is negligible at long range but dominates at contact. The practical upshot is that larger molecules with more diffuse electron clouds experience significantly stronger dispersion interactions. This is why n-pentane is a gas at room temperature while n-eicosane is a waxy solid. No hydrogen bonds are involved. No permanent dipoles exist. It is purely the cumulative effect of billions of tiny instantaneous dipoles reinforcing each other across a large surface area.

I once had a client dealing with a separation column where two nonpolar hydrocarbons with nearly identical boiling points would not resolve. We tried temperature optimization, flow rate adjustments, everything standard. The breakthrough came when I calculated the polarizability difference between the two compounds and realized we needed a stationary phase with specifically tuned dispersion characteristics rather than just adjusting operational parameters. We switched from a standard C18 column to one with a phenyl-hexyl bonded phase and got clean separation in a single run. The phenyl group provided additional pi-conjugation that responded differently to each analyte's dispersion profile. It was a twenty minute change that saved us three weeks of method development.

How to estimate and work with dispersion in practice

If you are doing this computationally, use a dispersion-corrected functional like B3LYP-D3 or wB97X-V. Standard B3LYP without empirical dispersion correction will give you geometries and binding energies that are qualitatively wrong for noncovalent systems. The D3 correction adds a simple -C6/R^6 term with damping at short range, and it costs virtually nothing in terms of computational overhead. If you are working experimentally and need to predict solubility or phase behavior, look at polarizability and molecular surface area rather than just molecular weight. Two isomers with the same mass can have wildly different dispersion contributions if one is compact and the other is extended. Linear alkanes interact more strongly through dispersion than their branched counterparts, which is why n-hexane boils at 69 degrees Celsius while 2,2-dimethylbutane boils at 50 degrees Celsius despite having the same formula. There is also the matter of solvent effects. Dispersion forces are heavily screened in solution because the solvent molecules compete for interaction sites. What looks like a strong dispersion-driven association in the gas phase can be essentially nonexistent in a polar solvent. I learned this the hard way when my lab assumed a guest-host complex was held together primarily by dispersion interactions based on crystallographic data, only to find it dissociated immediately upon dissolving in anything more polar than hexane.

Where dispersion fails as an explanatory tool

The biggest mistake beginners make is assuming dispersion explains everything that looks like an intermolecular attraction. It does not account for directional preferences in hydrogen bonding, it does not explain the specificity of electrostatic interactions, and it is completely irrelevant when ionic bonds or covalent bonding are the dominant forces. You will see people overattribute phenomena to London dispersion when dipole-dipole or induction effects are actually responsible. Another limitation is that dispersion interactions are additive in a way that makes them dominate only at large contact areas. For small isolated molecules in dilute solution, the effect can be so minor that it is buried within experimental error. If you are trying to measure binding energies below about 2 kilojoules per mole, you are likely measuring noise rather than dispersion. The good news is that the field has moved past the point where dispersion was considered a minor correction. Modern force fields like OPLS-AA and CHARMM now include explicit dispersion terms calibrated against ab initio data, and machine learning potentials are beginning to capture higher-order many-body dispersion effects that traditional pairwise approximations miss. If you are building models, check whether your force field includes many-body dispersion. The difference between a two-body and a three-body treatment can shift predicted densities of liquid hydrocarbons by several percent, which matters if you are simulating material properties rather than just qualitative behavior.