Computer analysis of electron paramagnetic resonance data using the Monte Carlo method

Publisher Website
Google Scholar
Abstract
A Monte Carlo computing procedure different from the standard least-squares fitting method is presented for evaluating the set of parameters which best fit the effective spin Hamiltonian. The parameter g and the electric-field parameters blm are fitted using the experimental data obtained from the electron paramagnetic resonance (EPR) spectra. This method is very powerful as regards decision making so that the labelling of the quantum levels involved in a transition is optional in most cases. The different steps in the simulated 'annealing' program are described in detail and a sample calculation is given for Gd3+ in a CaF2 matrix when it is located in tetragonal or cubic sites. This centre was chosen as the results obtained can be compared with previous determinations reported in the literature by using more conventional fitting procedures. This method is applicable even when the off-diagonal elements of the effective spin Hamiltonian matrix are large. The use of Monte Carlo optimisation methods, even if time consuming, is a modern approach that is easily available as it can be run on a microcomputer; this offers clear advantages for fitting the EPR data.