Speaker
Description
There is a range of quantum mechanical methods that can be used to calculate binding energies of molecular clusters or between molecules and surfaces. The individual approaches then differ in computational cost they require and accuracy they can provide. We will discuss the random-phase approximation (RPA), which is a method that, in terms of accuracy and computational demands, sits between advanced coupled clusters (CC) techniques and less demanding density functional theory (DFT) approximations. For RPA we obtained binding energies of several molecular solids and of methane in water clathrate cage. To analyse the accuracy of RPA we do not only consider the binding energy of the whole cluster or solid but also break it down using many-body expansion (MBE). In this way we obtain a large set of molecular dimers, trimers, and tetramers for which we obtain reference CC binding energies as well as energies for RPA and other schemes. The results show that RPA dimer binding energies significantly depend on the distance between molecules in dimer. Errors are large for small separations but very low for large separations. For the non-additive energies of trimers we identified two problems: first, the RPA energies have much larger dependence on the basis-set size compared to CC values and, second, they partly inherit the often large errors that DFT methods exhibit. We discuss several ways that we identified that can be used to reduce these issues.
