We solve numerically the dissipative Landau-Lifshitz-Gilbert equations to consider hysteresis in fine magnetic particles. Finite size effects are studied for two models with uniaxial anisotropy-bulk random axis and surface anisotropy only. It is demonstrated that the latter model introduces considerable effects for small enough particles when the coupling to the anisotropy is equal or greater than the coupling to the isotropic Heisenberg exchange. We show that some features of magnetization reversal are associated with spins at the surface of fine particles.

We develop and carry out Monte Carlo simulations for an ensemble of superparamagnetic particles uniformly distributed in a nonmagnetic matrix. We find the magnetization below the blocking temperature T B when it shows hysteresis and above T B in the superparamagnetic region. We determine the blocking temperature for a set of anisotropy strengths from the magnetization and the susceptibility of the particles. A fixed number of Monte Carlo steps with a constrained acceptance rate is shown to be equivalent to an observation time in the simulations that is much shorter than experimental observation times. We show how the blocking temperature obtained in the simulations can be converted into the corresponding experimentally measurable blocking temperature by using this difference in the observation times. This provides a new method to compare Monte Carlo simulation results with experiments, such as recent ones on fcc Co particles.

We have studied the magnetic properties of spherical fcc clusters at zero temperature using a numerical solution of the Landau-Lifshitz equations of motion with damping. Of particular interest is the effect caused by anisotropy which is uniaxial, radial, and present only on the surface. The hysteresis curves are obtained for different values of the anisotropy strength and for different cluster sizes. Insight into the spin-field configurations close to reversal is given. A specific dependence of the coercivity on the size of the clusters is found, which is explained qualitatively by the relative weight of the surface sites in the total number of sites in the cluster.