Entenent el comportament erràtic dels funcionals de bescanvi-correlació: descomposició atòmica i diatòmica de propietats òptiques no-lineals

Abstract

Nowadays, computational chemistry has a lot of applications in different chemistry fields due to the accurate theoretical values it offers that make it possible to understand and predict a wide range of phenomena. However, there are still some aspects where computational chemistry is unable to obtain correct results unless it uses the best available tools (such as coupled cluster calculations, which are very computationally demanding and require long calculation times). One of them is the calculation of non-linear optical properties (the response of a system under a radiation of high intensity), calculations that result in values way higher than expected when calculated using DFT (Density Functional Theory) methods (which are the best available methods with reasonable calculation times and resources). This overestimation is very noticeable in calculations of the second hyperpolarizability (γ), but the error in these calculations is high enough for this to also be seen in polarizability (α).

For this reason, this work is started with the objective of finding the cause of the difference between calculated values and experimental values, studying specifically the first oligomers of polyacetylene, a polymer with a conjugated π system where this phenomenon is easily noticeable. Because this investigation would be impossible analysing the total value of the polarizability, which would only show the increasing trend of this value but without giving enough information to find an explanation for it, we will be using methods that allow for a decomposition of the total polarizability of a molecule in the contributions of different fragments (which could be defined as the atoms, or groups of atoms, or the monomers in the polymer…). After this decomposition it will be possible to analyse the tendency of each of these fragments, trying to find the origin of the overestimation of the final result, and finding vital information for future works to analyse which mathematical terms in the computational calculations are responsible for the deviation observed in the final result, and finally making it possible to design optimized functionals that are capable of calculating these properties correctly