Perrella, Fulvio (2021) Development of an adaptive electronic approach for extended Lagrangian ab initio molecular dynamics. [Tesi di dottorato]

Preview

Text perrella_fulvio_33.pdf Download (103MB) | Preview

Collection description

Ab initio Molecular Dynamics (AIMD) is nowadays a popular tool for the caracterization of equilibrium, ground state, properties of molecular systems. Beside the Born-Oppenheimer (BO) traditional approach to AIMD, extended Lagrangian methods such as Car-Parrinello Molecular Dynamics offer a good compromise between accuracy and computational cost, within the Density Functional Theory (DFT) framework. In order to explore photophysical and photochemical properties, the DFT Time-Dependent formalism is instead often chosen to perform excited states BOMD simulations. Extended Lagrangian ground state trajectories are nevertheless required to extensively sample the population being excited. The Atom-centered Density Matrix Propagation (ADMP) method is an extended Lagrangian (Car-Parrinello-like) ab initio Molecular Dynamics approach which relies upon the concurrent propagation of both the nuclear degrees of freedom and the density matrix in an orthonormal basis. The adiabaticity between the two sub-systems is crucial in order to keep only slight deviations from a true Born-Oppenheimer dynamics. In the first part of this work, novel fictitious electronic mass-weighting approaches have been developed, aimed at improving the separation between the nuclear and the electronic power spectra and so the adiabaticity. A first class of approaches is based upon the definition of a molecular core subspace, defined from the core atomic MOs. Then, the projection of the orthonormal atomic or molecular orbitals onto such subspace allows for a decomposition into a valence and several core components, which is exploited to evaluate the corresponding mass. ADMP test trajectories of several model compounds reveal that these weighting schemes allow one to use time steps greater than 0.1 fs, while still ensuring an acceptable accuracy. A second approach is instead derived from the definition of the electronic harmonic normal modes and the electronic Hessian. A mass-matrix is calculated imposing a common frequency to such modes. This method proves to be highly accurate at the lowest time steps, showing little or no error in the calculated vibrational bands with respect to a BO dynamics. Beside smaller model systems, ADMP trajectories of the Ru(II) complex [Ru(dcbpy)2(NCS)2]4- ('N3', an efficient dye for light-harvesting applications) have been collected, again revealing an improved accuracy if a rational weighting scheme is implemented. In the second part of this work, photophysical properties of N3 dye in solution have been explored, starting from a ground state characterization through ADMP simulations. In water solution, all the N3 solvation sites are highly solvated, as expected. Moreover, combined solvation and anharmonicity effects account for a ∼ 100 cm-1 red-shift in several N3 vibrational bands. The dynamics of two relevant MLCT excited states (called 1MLCTA and 1MLCTB) after the photo-excitation has been then investigated through real-time TD-DFT. Simulations on N3 optimized gas-phase structure show that the hole oscillates between the NCS ligands and Ru (at ∼ 15000 cm-1), while the electron resides on the dcbpy ligands, although in 1MLCTB a complete electron migration is observed after 15 fs. Moreover, the overall electron mobility appears reduced with respect to the hole in 1MLCTA. Water polarization counterbalances the distortion effects in the N3 solution structure excited state dynamics. In particular, the hole is more displaced towards the Ru center, while in 1MLCTA the electron migration is still inhibited. On average, the charge carriers mobilities appear reduced in both states and the separation increased in 1MLCTA. An ultra-fast dcbpy/dcbpy inter-ligand electron transfer is instead observed in 1MLCTB also in water solution. As shown by this work, ab initio (adiabatic and non-adiabatic) Molecular and Electronic Dynamics prove to be powerful tools to gather insight into systems and processes of technological and biological interest.

Downloads

Actions (login required)

View Item