(217c) Computational Methods for Exploring Reaction Paths in Catalytic Processes
The first topic is development of efficient methods for locating paths that connect local minima on potential energy surfaces, using only knowledge of the reactant and product coordinates. A recent development of this type, the Freezing String Method (FSM) will be discussed in detail, and contrasted with the earlier Growing String Method (GSM), in terms of both formal properties, and numbers of gradients typically required. Since these methods will typically be employed with computationally demanding quantum mechanical evaluation of the energy and gradients, it is vital to minimize the number of these evaluations. Either an FSM or GSM calculation yields a high quality guess at a transition structure, which can be refined by local optimization methods. This yields a picture of the energy surface that would be traversed at zero temperature.
One of the simplest possible approaches that accounts for temperature effects is to perform quasiclassical trajectories (QCT) to characterize barrier crossing and recrossing dynamics. In QCT, initial velocities of the atoms are assigned in a way that is consistent with the quantum mechanical vibrational populations in harmonic normal modes. For catalytic systems, with metastable intermediates separating reactants from products, short trajectories launched from transition states can yield product distributions that cannot be learned from the reaction path alone. This will be illustrated with hydrocarbon cracking reactions in zeolites. The complementary information that can be learned from reaction paths versus trajectories will be discussed.