Time evolution of CO vibrational populations during photodesorption by light pulses

Abstract

We describe the direct photodesorption of CO from the Ni 001 metal surface during the absorption of a short pulse of UV light by the adsorbate and account for the dynamics of the desorbing species coupled to electronic excitations of the substrate, which lead to energy dissipation. The interaction potentials and couplings for the ground and excited states are obtained from electronic structure calculations and from experimental information. The time evolution of CO vibrational populations is studied for propagating wavepackets for the adsorbate᎐substrate complex with a splitoperator algorithm, followed by a perturbative treatment of dissipation and response to the light pulse. Direct photodesorption is found to occur predominantly in the excited electronic state and is compatible with dissipation. Results are presented for CO photodesorption by Gaussian pulses with several durations to show the effect of light pulse shapes on the time evolution of populations. Shorter pulses in the femtosecond range are shown to give relatively larger populations of excited vibrational states during the desorption of CO.

Molecular photoexcitation in a medium: Density operator approach

“…(36) can be done using a split‐operator propagation (SOP), advancing time by small steps 32, 33. This is in fact the way we proceeded to obtain results for direct photodesorption of CO/Ni(001), where dissipation was modeled by a dissipative Hamiltonian term and unperturbed amplitudes were generated by a modified SOP to include dissipation 34. It should be noted that we can solve the direct excitation problem numerically without using perturbation theory, provided the p region can be described with a finite basis set of electronic states, and a set of grid points for the nuclear displacement variables.…”

Dissipative dynamics in many‐atom systems: A density matrix treatment

“…Examples of physical systems of interest are collision‐induced and photoinduced dynamics of molecules on solid surfaces, and of atoms and molecules in clusters. In our work, we have made use of density matrices to describe the photodesorption of diatomics from metal surfaces including dissipative effects, to calculate desorption yields, the time delays detected in femtosecond pulse excitation, and the effect of vibrational motions of adsorbates 10–15, and to predict the effect of chirped light pulses on yields 16. A related subject is the dynamics of electronically excited systems, where electronic motions are quantized and nuclear ones can be approximated as quasi‐classical.…”

“…The equations of motion for the density operator can be solved in a variety of ways, depending on whether they involve only a hamiltonian commutator, or both a hamiltonian commutator and a dissipative Liouvillian term. In our approach with a dissipative hamiltonian it is possible to decompose the density operator into a sum of density amplitude functions which can be propagated in time as wavepackets, using a split‐operator propagator suitable for a time‐dependent and non‐hermitian hamiltonian 11, 13. Several general procedures have been described in the literature to solve the L–vN equation for the density operator by means of expansions in a basis of time‐independent 40, 41 or time‐dependent states 42, 43, through numerical grid methods and polynomial approximations of propagators 44–47, employing stochastic numerical methods 48–50, and using path integral formulations 20, 51.…”

Density matrix theory and computational aspects of quantum dynamics in active medium

“…The density matrix equations can be solved in a variety of ways, depending on whether they involve only a Hamiltonian commutator or both a Hamiltonian commutator and a dissipative Liouvillian term. In our approach, the presence of only a dissipative Hamiltonian led us to the calculation of density amplitudes using a split‐operator propagator in time, suitable for a time‐dependent and non‐Hermitian Hamiltonian 17, 18. Several general procedures have been described in the literature to solve the L–vN equation by means of expansions in a basis of time‐independent 19, 20 or of time‐dependent states 21, 22, numerical grid methods, and polynomial approximations of propagators 23–26, stochastic numerical methods 27–29, and using path integral formulations 30, 31.…”