Skip to main content
Open Access Publications from the University of California

Development and Application of a Spin-symmetry Breaking Algorithm for Ab Initio Non-adiabatic Molecular Dynamics Simulations

  • Author(s): Vincent, Jordan
  • Advisor(s): Furche, Filipp
  • et al.

Fast photochemical processes are central to a number of problems challenging chemists today. Such problems include atmospheric air pollution driven by solar radiation, development of photocatalysts for clean energy sources and photodynamic therapies for treatment of diseases such as cancer, and light-driven biological phenomena such as vision and vitamin D photosynthesis. Continued improvement of computational methods which can describe and predict such photochemical processes is thus essential. Accordingly, this thesis focuses on two main topics: 1) The development of a spin-symmetry breaking algorithm which for the first time enables homolytic bond cleavage in 'on the fly' density functional non-adiabatic molecular dynamics simulations, and 2) a benchmarking application of the spin-symmetry breaking algorithm to acetaldehyde photochemistry.

Following a brief overview of the challenges inherent within this work, chapter 2 introduces fast photochemical processes and the theoretical foundations of ab inito non-adiabatic molecular dynamics methods. In Chapter 3, a new algorithm for breaking spin-symmetry within linear response time-dependent density functional theory is derived and implemented in the quantum-chemical software suite TURBOMOLE. H2 is used to illustrate the key features of the algorithm, which are: 1) Stability analysis is used to identify critical points on the PES, 2) tight convergence thresholds are used only near triplet instabilities, and 3) adaptive time stepping in the vicinity of triplet instabilities is employed to conserve total energy.

An application of the spin-symmetry breaking algorithm to acetaldehyde photochemistry is then discussed in Chapter 4. Ensembles of approximately 400 trajectories in the ground state and 400 trajectories initialized in the excited state are simulated. Compared to experiment, the spin-symmetry breaking algorithm yields qualitative accuracy for computed branching ratios and total kinetic energy probability distributions. Additionally, two dissociation mechanisms are identified and discussed: The non-transition state theory, ``roaming'', mechanism and new triple-fragmentation mechanisms. Finally, the spin-symmetry breaking algorithm reveals that an improved description of acetaldehyde photodissociation mechanisms may require inclusion of excited state dynamics.

Main Content
Current View