Preisach model for spin-glass remanences

P. Mitchler, R. Roshko, E. Dan Dahlberg, E. Wesseling

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


We present numerical calculations, based on a finite temperature Preisach model of hysteresis for a collection of thermally activated two-level subsystems, which yield peaks in the field dependence of the remanence, with identical systematics to those observed experimentally in spin glasses and some fine particle suspensions. The model allows thermally activated switching events over all energy barriers W<W≡kT ln(t/(Formula presented)), where T is the temperature, t is the observation time, and (Formula presented) is a microscopic time, in addition to those induced by an applied field (Formula presented) . The Preisach distribution is a product of a Gaussian distribution of interaction fields (Formula presented), with zero mean, and a Gaussian distribution of coercive fields (Formula presented) . The remanence peaks are generated by allowing the mean coercive field h-(Formula presented) to decrease in proportion to the magnetization m impressed on the system, as h-(Formula presented) =h-(Formula presented) -γm, where h-(Formula presented) and γ are positive constants. The isothermal remanent magnetization (IRM), obtained by assuming the same value for the cutoff parameter W for both the magnetizing and remanence processes, exhibits a peak which becomes sharper and shifts towards lower applied fields as W increases. The thermoremanent magnetization, obtained by assuming an 'infinite' value for W for the magnetizing process and a finite value for the remanence process, exhibits a peak which is larger and occurs at lower fields than the peak in the corresponding IRM.

Original languageEnglish (US)
Pages (from-to)5880-5885
Number of pages6
JournalPhysical Review B - Condensed Matter and Materials Physics
Issue number9
StatePublished - 1997


Dive into the research topics of 'Preisach model for spin-glass remanences'. Together they form a unique fingerprint.

Cite this