On the many-body Van der Waals binding energy of a dense fluid

We consider a dense system of neutral atoms. When the atoms are represented by isotropic oscillators (Drude-Lorentz model) interacting with nonretarded dipole-dipole forces, the binding energy of the system is given exactly by a well-known expression which is written as a sum of two-bond, three-bond, etc., Van der Waals interactions. For a Bravais lattice this expression for the binding energy can be computed numerically to arbitrary accuracy. This has been done for the f.c.c. lattices of the noble-gas solids by Lucas. For a fluid an exact evaluation would require the knowledge of higher-order molecular distribution functions. Various approximations are discussed for this case, the simplest of which is the so-called long-wavelength approximation due to Doniach. When this approximation is checked by comparison with the exact result for a lattice, it turns out that the two-bond contribution leads to a value which is more than twice too large. Some more refined approximations are considered which treat the two-bond contribution exactly. It is pointed out that the model is consistent only if the distance of closest approach between the atoms is not too small.