SEMICLASSICAL MOLECULAR DYNAMICS OF ADIABATIC AND NONADIABATIC PROCESSES

Molecular dynamics over the years has been established as a powerful yet sim- ple technique capable of offering new physical insights into molecular processes. Some of the most desirable features are that it retains favorable scalability with the increasing number of degrees of freedom and can also be used to reproduce processes happening on relatively long time scales. On the other hand, since molec- ular dynamics is based on classical mechanics, it neglects the quantum nature of the nuclear motion and will eventually lead to some noticeable deviations from the real behavior of molecular systems. Instead, when considering a quantum picture of the nuclear dynamics all of the previously cited desirable features of molecular dynamics tend to vanish. Making the whole picture usually more complicated and computationally daunting. Semiclassical methods are in part able to conciliate the best features of both worlds, allowing for an intuitive description of the underlying nuclear quantum dynamics based on classical molecular dynamics and at the same time keeping the scalability in function of the system dimensionality favorable. Throughout this thesis, semiclassical methods are employed to compute vibrational and vibronic spectra. More specifically, the first part of the manuscript will be dedi- cated to a quick review of the theory of the semiclassical initial value representation and its application to spectroscopy. The objective of the first Chapter is not only to show the origins of the semiclassical approximation but also to describe some strategies and further approximations that can pave the way for the simulations of molecular systems featuring up to 10000 degrees of freedom. Later, in the same part of the manuscript, it is also explored the idea of including nonadiabatic effects to semiclassical vibrational spectroscopy by employing the Meyer-Miller-Stock-Thoss mapping. In Chapter 3 a simple study case regarding the formaldehyde molecule is used to dig deeper into a tedious question regarding what can be regarded as a classical anharmonic effect and what can be classified as a "quantum effect" in vibrational spectroscopy. This Chapter also serves as a point to assess the validity of classical methods when compared to quantum mechanical ones and validate some of the best practices to employ regarding the former. In Chapter 4 both quasiclassical and semiclassical methods are used to reproduce the experimental spectrum of the solvated G-quadruplex and explain the effects of solvation on the observed vibrational features. To describe the dynamics of such a large system, an empirical force field must be employed, a topic that is not central to the Chapter and will be used as a starting point for the next one, regards the reliability of this way to describe the potential energy surface. To avoid the loss of accuracy that comes with employing a classical force field, but at the same time avoid the computational overheads that can be faced when employing a full ab initio description of the system, in Chapter 5 semiclassical molecular dynamics is employed alongside hybrid QM/MM potentials to obtain a new physical picture of solvation for the case of solvated thymidine. Such system turned out to be a very intriguing study case where the modeling of solute-solvent interactions is essential for the correct reproducibility of the experimentally observed spectroscopic behav- ior. Finally, in Chapter 6 applications of the novel time-averaged semiclassical approach to nonadiabatic vibronic spectroscopy are displayed for analytical model systems of growing difficulty. During this part of the manuscript, some of the features and limitations of the method will highlighted upon comparison with numerically exact results. Conclusions of the entire work and future perspectives are drawn in Chapter 7.