Monte Carlo simulation of exchange bias and training effects in ferromagnetic/antiferromagnetic bilayers with different Néel temperatures

A modified Monte Carlo Metropolis algorithm is employed to simulate the field-cooling and training dependencies of low-temperature exchange bias (HE) and coercivity (HC) in the ferromagnetic/antiferromagnetic bilayers with various antiferromagnetic Néel temperatures (TN), which are modulated by altering the antiferromagnetic exchange coupling constant (JAF). It is found that the smaller the JAF value, the more pronounced is the HE. However, strong cooling fields (HFC) may also induce a saturated low-temperature HE value. Interestingly, the low-temperature HC behaviors with JAF and HFC both exhibit a minimum value corresponding to the steepest change in HE. The evolutions of microscopic domain walls and domain sizes in the ferromagnetic layer are used to reflect the change in the antiferromagnetic configurations and, thus, to interpret the novel phenomena. On the other hand, HE in the bilayers with TN lower than the Curie temperature (TC) indicates a training effect due to the fact that the antiferromagnetic configurations near the interface which are created partially by their adjacent ferromagnetic layer via interfacial exchange coupling during field cooling are able to be rearranged at low temperature by repeating magnetizing. In other words, the completely frozen antiferromagnetic spins in the bilayers with TN > TC at low temperature lack the dynamics to cause the absence of training effect.