Anda di halaman 1dari 7

ARTICLE IN PRESS

Journal of Magnetism and Magnetic Materials 320 (2008) 24362442 www.elsevier.com/locate/jmmm

Connection between microstructure and magnetic properties of soft magnetic materials


G. Bertotti
INRIM-Istituto Nazionale di Ricerca Metrologica, Strada delle Cacce 91, 10135 Torino, Italy Available online 7 April 2008

Abstract The magnetic behavior of soft magnetic materials is discussed with some emphasis on the connection between macroscopic properties and underlying micromagnetic energy aspects. It is shown that important conceptual gaps still exist in the interpretation of macroscopic magnetic properties in terms of the micromagnetic formulation. Different aspects of hysteresis modeling, power loss prediction and magnetic non-destructive evaluation are discussed in this perspective. r 2008 Elsevier B.V. All rights reserved.
PACS: 75.60.d Keywords: Micromagnetics; Magnetic hysteresis; Hysteresis model; Power loss; Non-destructive evaluation

1. Introduction The behavior of soft magnetic materials is dominated by magnetostatic effects. Through careful adjustment of relevant microstructural factors (e.g., grain size, crystallographic texture, induced anisotropies) this results in a wide variety of domain arrangements [1], whereby one can tailor the magnetic response of the material to specic needs. There are cases where domain structures take remarkable simplicity, like the bar-like domains observed in grain-oriented silicon steels or in magnetostrictive amorphous ribbons under stress. On the other hand, it is by no means obvious to get detailed information about the seemingly complex domain structures that are responsible for the hysteretic response of ne-grained materials like non-oriented silicon steels, where grain size is of the order of 50100 mm and magnetic domain size is accordingly smaller. Prediction of magnetic domain structures is a fundamental part of the comprehension of soft magnetic materials. However, the derivation of magnetic domain
Tel.: +39 11 3919 835; fax: +39 11 3919 834.

E-mail address: g.bertotti@inrim.it URL: http://www.inrim.it 0304-8853/$ - see front matter r 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jmmm.2008.04.001

structures from structural information about the particular system considered is a problem of formidable difculty. The theoretical formulation of this problem in terms of energy minimization is known in the literature under the name of micromagnetics [14]. Micromagnetics is the leading theoretical approach used to describe the behavior of magnetic materials at scales where atomic granularity is not resolved and the material is described as a continuous medium whose magnetic state is dened by the magnetization vector as a function of position inside the body. The physical foundation of micromagnetics was laid by Landau and Lifshitz [5]. It was then given a clear variational formulation by Brown [6,7], who showed the form taken by the micromagnetic free energy functional and discussed how minimization of this functional may lead in principle to predictions for the large variety of magnetic congurations observed in ferromagnetic bodies. Few analytical solutions are known for micromagnetics, due to the complex nonlinear nature of the equations involved. Conversely, numerical micromagnetics has acquired a leading role after computing power has become a broadly available resource. Nowadays, micromagnetic simulations are of key importance in micrometer or sub-micrometer applications to magnetic recording and

ARTICLE IN PRESS
G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442 2437

magneto-electronics [8]. In this area, attention has in fact shifted from the original micromagnetic goal of reconstructing the distribution of the energy minima responsible for hysteresis to the time integration of the LandauLifshitz Gilbert equation governing magnetization dynamics [9]. In contrast to the widespread use of micromagnetic codes in magnetic recording and magnetic nanotechnology developments, challenging difculties are still encountered in the application of numerical micromagnetics to bulk materials dominated by structural disorder. Congurations minimizing the micromagnetic free energy are the result of the competition of various energy contributions: ferromagnetic exchange, crystal and induced anisotropies, magnetoelastic interactions, magnetostatic interactions. Of these, magnetostatic effects are long-ranged and prevent any possibility of reducing the energy minimization problem to a set of local problems in individual volumes of the material. Micromagnetic states are usually characterized by ne spatial oscillations (presence of magnetic domains and domain walls) down to sub-micrometer scales. At the same time, the consequences of magnetostatic interactions are revealed only at the macroscopic scale set by the body geometrical dimensions. This makes full-detail simulations of bulk systems computationally prohibitive. An exception is represented by micromagnetic investigations of bulk permanent magnets [10]. However, in that case the exceeding strength of local crystal-anisotropy forces makes large-scale magnetostatic effects of secondary importance and justies the assumption that simulations of small aggregates of few crystalline grains may be representative of the behavior of bulk materials. These simplifying assumptions completely fail for soft magnetic materials, where the strength of crystal anisotropy is smaller than that of magnetostatic interactions. The impossibility of developing rigorous micromagnetic tools for the description of bulk magnetic materials is reected by the numerous phenomenological hysteresis models that have been proposed for the treatment of bulk properties. A similar situation is encountered in power loss modeling, where losses are estimated by making a certain number of phenomenological assumptions about magnetization processes. A feature of great potential clearly expressed by the micromagnetic formulation is the intimate connection existing between microstructural aspects (presence of secondary phases, dislocations, grain size, texture, anisotropies) and magnetic domain congurations or magnetization processes. This means that the macroscopic magnetic behavior of a material must reect its microstructural state and thus in principle can provide a great deal of information about microstructural features [11]. This is the main reason why the development of nondestructive tools for magnetic inspection of materials and products has attracted and is still attracting considerable attention. A case of importance is represented by steel processing, where complex sequences of casting/rolling/annealing steps

are used to tune the steel microstructure in order to achieve nal products with homogeneous properties within the desired acceptance intervals. Here, magnetic testing can lead to more effective process control, with the ability to identify within short times the causes for unexpected deviations from desired product quality. In addition, a point which should not be overlooked is that magnetic inspection can be operated without contact on moving products, a decisive feature for on-line testing inside steelmaking plants. On the other hand, magnetic testing is of broad interest in the inspection of operating components for the early detection of damage or approach to failure. An example is the case of radiation damage in nuclear reactor vessels [12]. The main issue in magnetic testing methods is that the correlation between testing results and the microstructural/ mechanical properties of interest is often built up by crude empirical methods leading to look-up tables for individual products which say little about the origin of undesired deviations from expected product quality or operation. This state of affairs reects the fact that there still exists a conceptual gap in the comprehension of the relationship between bulk magnetic properties and the underlying microscale coupling of magnetization to structural disorder. Starting from the classical formulation of micromagnetics, the connection between microstructural and magnetic aspects could in principe be better revealed by resorting to intermediate-scale formulations of the energetics of a ferromagnet, valid for polycrystals on the scale of the grain size. These formulations may be based on asymptotic models recently proposed in the literature [13], which permit one to work out the form taken by the free energy of a ferromagnetic polycrystal when the spatial resolution used to describe the problem does not resolve the details of the magnetic domain structure present in each grain. Developments based on this general framework could lead to the implementation of efcient numerical codes for the simulation of the connection between microstructural disorder and magnetic properties under realistic conditions, close to the ones encountered in steel industry and in non-destructive magnetic evaluation. 2. Micromagnetics and multiscale approaches In micromagnetics, the state of a ferromagnetic body is described by the vector eld Mr representing the local magnetization at every point r inside the body. For ferromagnets well below their Curie temperature, exchange interactions prevail over all other forces at short spatial scales. This fact is taken into account by imposing that the magnitude of the local magnetization vector must be equal to the spontaneous magnetization Ms : jMrj Ms . Conversely, the orientation of Mr is in general nonuniform, i.e., it varies from point to point. At equilibria, the spatial distribution of these orientations must correspond to a minimum of the GibbsLandau free energy G L M:; Ha of the system, where Ha represents the externally applied

ARTICLE IN PRESS
2438 G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442

magnetic eld. In micromagnetics, the free energy GL for a ferromagnetic body occupying the region O is expressed as the following volume integral: Z " A G L M:; Ha rMx 2 rMy 2 rMz 2 2 O Ms # m0 f AN M M HM m0 M Ha dV . 2 (1) The rst term inside the integral represents the exchange energy (A is the so-called exchange stiffness constant, typically of the order of 1011 J=m), f AN M describes crystal anisotropy effects, while the two last terms represent magnetostatic energy and energy of interaction with the applied eld Ha . The magnetostatic contribution is governed by the associated magnetostatic eld HM , which is solution of magnetostatic Maxwell equations: r H M 0; r HM r M, (2) subject to the appropriate interface conditions at the body surface. The micromagnetic free energy may contain additional terms describing other mechanisms, like for example magnetoelastic interactions, which have not been included in the simplied expression above. The equilibrium magnetization states under given applied eld Ha correspond to the condition dGL 0, where dG L represents the free energy variation with respect to arbitrary variations of the vector eld Mr subject to the constraint jMrj Ms . By using standard variational calculus, one obtains that dG L 0 if at each point in O the following equation is satised (Browns equation): M Heff 0. Here Heff represents the effective eld, dened as Heff HEX HAN HM Ha , (4) HEX and HAN being the exchange eld and the anisotropy eld, respectively: HEX 2A r 2 M; m0 M2 s HAN 1 qf AN . m0 qM (5) (3)

Eq. (3) expresses the fact that the local torque exerted on the magnetization by the effective eld must be zero at equilibrium. However, this equation determines all possible magnetization equilibria regardless of their stability. According to the thermodynamic principle of free energy minimization, only G L minima will correspond to stable equilibria which can be in principle physically observable. The information on the nature of equilibria can be obtained by deriving the second variation of GL and determining if it is positive under arbitrary variations of the vector eld Mr, subject to the constraint jMrj Ms . Micromagnetic equations are complex nonlinear equations giving detailed information about the local orientation of magnetization at each point inside the body. However, for bulk materials, such complete information

about all micromagnetic details is seldom needed. The large-scale behavior of the body is expected to depend only on robust qualitative features emerging when one develops appropriate coarse-graining methods, and one connects the micromagnetic scale to the coarse scale by multiscale techniques [13]. At the same time, thermodynamic approaches may be effective in capturing the essential aspects of the macroscopic behavior, rst of all the appearance of hysteresis phenomena [14]. A rst remarkable example of coarse-graining theory of el phase theory for bulk single crystals (see this sort is Ne Ref. [3] for an overview). This theory is based on the assumption that the magnetic state of the body can be described in terms of the relative volume fractions (magnetic phases) magnetized along the various anisotropy directions in the crystal, all details about the size and orientation of individual magnetic domains being irrelevant. More precisely, the simplifying assumptions introduced in phase theory are the following: (i) the body is of ellipsoidal shape and the applied eld is spatially uniform; (ii) magnetic domains are ne enough that, on the scale of the specimen geometry, the materials appears as homogeneous, which, together with assumption (i), implies that the coarse-grained magnetization and demagnetizing eld are spatially uniform as well; (iii) the domain structure is such that the formation of local magnetic charges is completely avoided (principle of pole avoidance), so that the magnetostatic energy is only related to magnetic charges at the surface of the body; (iv) there are no impediments, due for example to structural disorder, to domain wall motion. Under these conditions, the behavior of the body is governed only by the anisotropy energy and the energy of interaction with the applied and magnetostatic elds. The degree of renement of the domain structure becomes irrelevant: the state of the body can be described just in terms of the relative fractions of the body volume magnetized along different anisotropy easy axes. el phase approach proves useful in many cases where Ne the competition between crystal anisotropy and magnetostatic effects dominates the problem. In a recent interesting work [15] some aspects of phase theory have been applied to the description of grain-oriented silicon steels magnetized along directions different from the rolling direction. el phase theory have been adapted in The principles of Ne order to take into account the specic role played by Goss texture and the fact that hysteresis effects cannot be neglected. On this basis, one is able to work out rather accurate predictions for magnetization curves, hysteresis loops and power losses for laminations cut at arbitrary angles with respect to the rolling direction, once the magnetic properties along the rolling direction and the transverse direction are measured. Another remarkable example of large-scale theory for soft materials is that due to Van den Berg [16,17], originally developed for the case where no external eld is applied, and subsequently extended to the case of nonzero eld as well [18]. In Van den Berg approach, one justies on physical grounds the fact that if one deals with large and

ARTICLE IN PRESS
G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442 2439

ideally soft thin lms, the magnetic congurations minimizing the micromagnetic free energy under zero applied eld can be determined by purely geometrical methods by making the following simplifying assumptions (i): the magnetization M is uniform across the lm thickness and is in-plane (the component of M normal to the lm plane is zero); (ii) M fullls the saturation constraint jMj Ms ; (iii) the magnetization states are such that no magnetic poles are created, namely M is divergence-free and tangential at the body surface; (iv) the other energy contributions coming from anisotropy and exchange can be completely ignored. The problem of determining the magnetization vector eld in the body is then solved by the method of characteristics and translated into graphical rules for the construction of the admissible magnetization congurations for a lm of given geometry. These rules do not lead to a unique solution, but to families of solutions which are totally equivalent. As a matter of fact, one expects that this equivalence should be destroyed once the energy contributions neglected in the construction (e.g., domain wall energy, anisotropy energy inside domains) were to be taken into account. The geometrical nature of Van den Berg construction reects the fact that domain congurations in soft thin lms are the result of topological constraints coming from the requirement of accommodating a divergence-free vector eld in a given bidimensional region. In particular, it shows how the divergence-free character of magnetization congurations naturally leads to the appearance of domain walls and domain structures even if exchange and anisotropy are completely neglected. The coarse-graining models discussed above led to important progress in the conceptual understanding of the equilibrium patterns in large bodies. However, the connection of these models with the general theory of micromagnetics is still an open eld of investigation. A thorough study of the connection between micromagnetic el phase theory was carried out by De theory and Ne Simone [19]. He showed that in the limit where the volume of the body becomes very large, the limiting behavior of the magnetic congurations minimizing the micromagnetic free energy (1) is represented by the congurations minimizing the modied energy functional: Z h i m G L M:; Ha fM 0 M HM m0 M Ha dV , 2 O (6) under the relaxed constraint MrjpMs . The function fM represents the so-called convex envelope of the function f AN M in Eq. (1). This transformation (convex envelope) is the vector equivalent of the well-known Maxwell construction used in the theory of phase coexistence and is the appropriate construction for the description of the coexistence of different magnetic phases in the body. Indeed, the appearance of fM and of the relaxed constraint MrjpMs reect the fact that in the large-body limit one accepts congurations consisting of

ne mixtures of domains along different directions, whose details are not resolved but expressed in terms of a coarsegrained net magnetization averaged over elementary volumes containing many domains.. One can prove that in the case of an ellipsoidal body subject to a uniform applied eld the theory described by Eq. (6) is equivalent to el phase theory. Ne Considerable mathematical work has also been devoted to the study of the connection between the behavior of magnetically soft thin-lm and micromagnetics [13,20]. In this case too, one can dene a suitable limiting procedure when one of the dimensions of the body (the lm thickness) tends to zero. By examining the scaling behavior of the various micromagnetic energy terms in this limit, it is possible to derive a coarse-grained energy functional for the thin-lm, which should be the basis for the rigorous justication of the validity of Van den Berg-type theories. An open issue in this respect is the fact that in the coarsegraining approach there appears the relaxed micromagnetic constraint jMrjpMs and not the full constraint jMrj Ms used in Van den Berg construction. The asymptotic models discussed above have been developed so far for single crystals under idealized conditions. It is foreseeable that they could be used as the basis for generalized formulations, valid for polycrystals in the presence of signicant structural disorder. Two lines of development seem to be of interest. On the one hand, asymptotic approaches could be applied in spacedependent form, inside each individual crystal grain. On the other hand, the fact that the coarse-grained magnetization actually describes a distribution of magnetic domains implies the presence in the coarse-grained energy functional of additional energy terms, neglected in idealized models. In particular, one expects a term describing the energy of the magnetic domain wall surface associated with a given conguration, and a random energy term associated with pinning of domain walls with structural disorder. These terms will be functions of the coarse-grained magnetization vector, with parameters dependent on the micromagnetic details of the underlying domain distribution, details not directly resolved by the asymptotic description. The application of multiscale methods to micromagnetics is part of the general progress under way in the eld of multiscale approaches. Algebraic multigrid methods have been applied in their adaptive form to large sparse matrix equations arising in the nite-element description of stray elds in micromagnetic problems with nonuniform or unstructured grids [21]. Systematic upscaling has been applied to studies of Maxwell equations within heterogeneous magnetic materials [22,23]. Methods based on homogenisation theory [24] have been proposed in relation to the minimization of the micromagnetic energy functional [25]. Of particular interest are the multiscale-niteelement method [26] and the heterogeneous-multiscale method [27], which propose computational-cost-efcient ways to include the effect of microscale structures in the coarse-grained nite-element description.

ARTICLE IN PRESS
2440 G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442

3. Magnetic hysteresis The most evident macroscopic result of micromagnetic mechanisms is the appearance of hysteresis phenomena of complex nature. Various models have been proposed for the description of these phenomena: Preisach-type models [3,28], JilesAtherton-type models [29,30], and various attempts at vector hysteresis modeling [31]. These models have distinct interesting features but a common shortcoming: they are based on ad hoc assumptions, phenomenological in nature, for which no clear justication is given in terms of micromagnetic mechanisms. Vector hysteresis models attempt various strategies aimed at capturing the basic consequences that a certain vector eld history will have on the material response. JilesAthertontype models identify the main factor of interest in the competition between the natural tendency of the material to evolve toward minimum energy states (the anhysteretic curve) and the hindering effect of structural disorder which opposes this tendency. Finally, in Preisach-type models, structural disorder results in a statistical distribution of elementary magnetization reversal events, under the assumption that interaction effects, and rst of all magnetostatic interactions, can be expressed in frozen, state-independent form. The Preisach approach has been often used in the past as a tool to classify and compare materials based on differences in their Preisach distribution [3234]. Representing a material by its Preisach distribution can be useful in order to give evidence of certain characteristic features of the hysteretic response. However, this approach suffers from an inherent shortcoming. It was proven that there exist two properties, known as the wiping-out or returnpoint memory property and the congruency property, which express the necessary and sufcient conditions for the representation of a generic scalar hysteretic system by the Preisach model [28,35]. Soft magnetic materials satisfy these properties only approximately, sometimes with signicant discrepancies. Therefore, in most cases the Preisach representation is the result of manipulations of the experimental data forcing them into the Preisach framework. To overcome this difculty, a more exible representation has been proposed, known under the name of FORC diagrams [36], where FORC stands for rst-order reversal curves. In fact, the set of rst-order magnetization reversal curves is what permits one, through appropriate double differentiation, to reconstruct the Preisach distribution. If the same differentiation is carried out without assuming that the system response is describable by the Preisach model, one obtains diagrams in which a function analogous to the Preisach distribution is generated without forcing the symmetries typical of Preisach systems. FORC representations are being applied to various systems and materials [37]. Recently, attempts have been made to apply this method to soft magnetic materials, in order to show possible similarities between experimental outcomes

and predictions of specic models (e.g., JilesAtherton model) [38]. If one accepts the idea that the phenomenological hysteresis models previously mentioned can be an effective way of lumping all the details of underlying micromagnetic mechanisms into a limited number of parameters, then one can investigate the dependence of these parameters on specic microstructural features. A case of particular importance in this respect is the dependence of macroscopic hysteresis on plastic deformation [39]. In recent papers [40,41], this problem has been addressed in the frame of JilesAtherton-type models. The two model parameters k and a that control the strength of pinning effects and the slope of the anhysteretic curve are assumed to have identical dependence on grain size s and dislocation density zd , this p dependence being of the type: a; k / G 1 G 2 =s zd , where G 1 and G 2 are constant coefcients. Once the dislocation density is connected to the ow stress and plastic deformation by appropriate models of plastic ow, one is capable of predicting how the hysteresis loop will evolve as a function of plastic strain. 4. Magnetic power losses A case, of central importance to soft magnetic materials, where micromagnetics is bypassed by ad hoc simplifying assumptions is the description of power losses. From the micromagnetic point of view, the loss problem would amount to the quite formidable task of solving the equations for the dynamic evolution of the magnetic domain structure in a metal (i.e., micromagnetic equations + Maxwell equations), by including into the effective eld (Eq. (4)) the magnetic eld generated by eddy currents at each point in the material. It was suggested in the frame of the so-called statistical loss theory [42] that the loss behavior is dominated by certain qualitative features of the magnetization process, rst of all the fact that magnetization reversal takes place in a strongly correlated fashion inside correlation regions of dimensions controlled by the microstructure (e.g., grain size). Once the phenomenological assumption of the existence of a certain number of statistically independent correlation regions is accepted, the basic aspects of the loss behavior can be understood and reproduced in a satisfactory way. First of all, the statistical approach shows that the so-called loss separation method is not a mere empirical procedure for loss analysis, but has indeed a precise physical origin. In fact, one can prove by appropriate statistical methods of the theory of Markovian stochastic processes [3] that, under rather broad conditions, the total loss P is the sum of three terms for which the theory gives denite expressions in terms of the correlation properties of the magnetization process: P Physt Pclass Pexc . (7)

These terms are the so-called hysteresis Physt , classical Pclass , and excess Pexc components. In particular, one nds that the expression for Pclass predicted by the

ARTICLE IN PRESS
G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442 2441

statistical approach coincides with the loss that one would have if no hysteresis and no magnetic domains were present. The decomposition of the loss into these contributions reects the coexistence of three dominant space-time scales in the magnetization process: fast and localized elementary Barkhausen jumps giving rise to Physt ; largescale eddy current patterns controlled by the system geometry Pclass ; and intermediate-scale losses consequent to the existence of the correlation regions mentioned above. Pexc . A particularly simple and interesting case is that of materials with ne domain structures and important structural disorder (e.g., non-oriented silicon steels). In that case, the loss per unit volume under sinusoidal ux in a lamination of thickness d takes the form: p2 sd 2 I p f 2 C I p f 3=2 , (8) 6 where f is the magnetization frequency, I p the peak magnetization, s the electrical conductivity, and C a material-dependent parameter related to the correlation region properties. The parameter C turns out to be intimately connected to the hysteresis loss itself, as a consequence of the fact that the distribution of local coercive elds controlling quasi-static hysteresis also controls the number of active correlation regions under given excitation conditions. In addition, comparison with loss measurements shows that C often contains some residual dependence on the peak magnetization I p . A generalized version of Eq. (8), proposed in Ref. [43], proves extremely useful in all those cases where the material response is non-sinusoidal. It is a formula for the instantaneous loss at time t:    3=2 sd 2 dI 2 0 dI Pt Physt t C , (9) dt 12 dt P Physt expressed in terms of the instantaneous magnetization rate of change dI =dt and a parameter, C 0 , strictly connected to the parameter C of Eq. (8). The effect of ux distortions on losses can be estimated quite generally by taking the average of Eq. (9) over a magnetization cycle. The loss problem is part of the more general problem of predicting the shape of the dynamic hysteresis loop traversed by a soft magnetic specimen under given excitation conditions. The so-called dynamic Preisach model [44] was developed for this purpose. Related to this problem are the attempts made by several authors to include adequate descriptions of the local hysteretic response of the material into Maxwell solvers for the study and design of transformers, electrical machines and various types of magnetic devices [45]. The dynamic Preisach mode can be useful in this respect too. More recent developments along a different line [46] are particularly suited to the description of electrical steels. The statistical loss theory was originally developed and applied with success to silicon steels, nickeliron materials, amorphous ribbons. Recently, the same approach has been extended to MnZn ferrites [47,48]. Although in this case

eddy currents are not the main cause of dissipation, the description in terms of active correlation regions can still be applied, while assuming that dissipation is due to spin relaxation through LandauLifshitz dynamics. This leads to an expression for the loss of the type of Eq. (8), where the second term, the so-called classical loss, is negligible under most circumstances, while the excess loss term consists of two contributions, one due to domain wall motion and the other to magnetization rotation inside grains. The domain wall term is of the form: Pexc / ahsiPhyst n I p f 2 1=n1 , (10)

where a is the dimensionless damping constant appearing in LandauLifshitz equation (of the order of some 102 ), hsi is the average grain size, and n is a material-dependent exponent, which is found typically to be in the range: 0:5pnp0:8. In this case too, a direct connection between excess loss and hysteresis loss is found. A detailed analysis of the residual role played by eddy currents in soft ferrites can be carried out by appropriate homogenization techniques [49,50] applied to the heterogeneous microstructure of the ferrite core, consisting of grains and inter-grain layers. This approach permits one, in particular, to determine the conditions associated with qualitatively different dynamical regimes, one where closed eddy current ow patterns are present inside each individual grain, and the other where eddy currents ow across inter-grain layers and generate patterns extending over the entire ferrite core. Finally, it is worth mentioning that the statistical loss theory has been recently applied also to systems remarkably far from the ones for which the theory was originally conceived, that is ultra-thin lms [51]. For these systems, there has been some debate in the recent past about the fact that measured dynamic hysteresis loops were or were not giving experimental evidence for certain cross-over effects between expected regimes of dynamical response. In Ref. [51], it is shown that no ad hoc assumption about the existence of specic dynamical regimes is needed, because all the experimental data about losses in ultra-thin lms are consistent with the predictions of the statistical loss theory, the differences observed for different lms being simply due to differences in the number of active correlation regions. Acknowledgments The author is indebted to O. Bottauscio and C. Serpico for helpful discussions about the use of multiscale approaches in micromagnetics. References
[1] A. Hubert, R. Scha efer, Magnetic Domains, Springer, Berlin, 1998. [2] A. Aharoni, Introduction to the Theory of Ferromagnetism, Oxford University Press, Oxford, 1996. [3] G. Bertotti, Hysteresis in Magnetism, Academic Press, Boston, 1998.

ARTICLE IN PRESS
2442 G. Bertotti / Journal of Magnetism and Magnetic Materials 320 (2008) 24362442 [27] W. E, B. Engquist, X. Li, W. Ren, E. Vanden-Eijnden, Commun. Comput. Phys. 2 (2007) 367. [28] I.D. Mayergoyz, Mathematical Models of Hysteresis and Their Applications, Elsevier, Amsterdam, 2003. [29] D.C. Jiles, Introduction to Magnetism and Magnetic Materials, Chapman and Hall, London, 1990. [30] L.R. Dupre, R. van Keer, J.A.A. Melkebeek, J. Appl. Phys. 85 (1999) 4376. [31] E.D. Torre, Magnetic Hysteresis, Wiley-IEEE Press, New York, 2000. [32] G. Durin, C. Beatrice, C. Appino, V. Basso, G. Bertotti, J. Appl. Phys. 87 (1999) 4768. [33] E. Cardelli, E.D. Torre, G. Ban, Physica B 275 (2000) 262. [34] L.R. Dupre, R. van Keer, J.A.A. Melkebeek, J. Magn. Magn. Mater. 254255 (2003) 121. [35] I.D. Mayergoyz, Phys. Rev. Lett. 56 (1986) 1518. [36] C.R. Pike, A.P. Roberts, K.L. Verosub, J. Appl. Phys. 85 (1999) 6660. [37] H. Chiriac, N. Lupu, L. Stoleriu, P. Postolache, A. Stancu, J. Magn. Magn. Mater. 316 (2007) 177. [38] A. Stancu, P. Andrei, Physica B 372 (2006) 72. [39] S. Takahashi, J. Echigoya, Z. Motoki, J. Appl. Phys. 87 (2000) 805. [40] M.J. Sablik, T. Yonamine, F.J.G. Landgraf, IEEE Trans. Magn. 40 (2004) 3219. [41] M.J. Sablik, F.J.G. Landgraf, R. Magnabosco, M. Fukuhara, M.F. de Campos, R. Machado, F.P. Missell, J. Magn. Magn. Mater. 304 (2006) 155. [42] G. Bertotti, IEEE Trans. Magn. 24 (1988) 621. [43] F. Fiorillo, A. Novikov, IEEE Trans. Magn. 26 (1990) 2904. [44] G. Bertotti, IEEE Trans. Magn. 28 (1992) 2599. [45] L.R. Dupre, O. Bottauscio, M. Chiampi, M. Repetto, J.A.A. Melkebeek, IEEE Trans. Magn. 35 (1999) 4171. [46] S.E. Zirka, Y.I. Moroz, P. Marketos, A.J. Moses, IEEE Trans. Magn. 42 (2006) 2121. [47] F. Fiorillo, C. Beatrice, O. Bottauscio, A. Manzin, Appl. Phys. Lett. 89 (2006) 122513. [48] C. Beatrice, F. Fiorillo, IEEE Trans. Magn. 42 (2006) 2867. -Piat, M. Codegone, M. [49] O. Bottauscio, A. Manzin, V. Chiado Chiampi, J. Appl. Phys. 100 (2006) 044902. -Piat, M. Chiampi, M. Codegone, A. [50] O. Bottauscio, V. Chiado Manzin, Eur. Phys. J. Appl. Phys. (2007) 2007070 doi:10.1051/epjap. [51] F. Colaiori, G. Durin, S. Zapperi, Phys. Rev. Lett. 97 (2006) 257203. [4] H. Kronmuller, M. Fahnle, Micromagnetism and the Microstructure of Ferromagnetic Solids, Cambridge University Press, Cambridge, 2003. [5] L.D. Landau, E.M. Lifshitz, Physik. Zeits. Sowjetunion 8 (1935) 153. [6] W.F. Brown, Magnetostatic Principles in Ferromagnetism, NorthHolland, Amsterdam, 1962. [7] W.F. Brown, Micromagnetics, Krieger Publishing Co., New York, 1963. [8] B. Hillebrands, A. Thiaville (Eds.), Spin Dynamics in Conned Magnetic Structures III, Springer, Berlin, 2006. [9] A.G. Gurevich, G.A. Melkov, Magnetization Oscillations and Waves, CRC Press, Boca Raton, FL, 1996. [10] D. Suess, T. Schre, J. Fidler, IEEE Trans. Magn. 36 (2000) 3282. [11] F. Fiorillo, Measurement and Characterization of Magnetic Materials, Elsevier, Amsterdam, 2004. [12] S. Takahashi, H. Kikuchi, K. Ara, N. Ebine, Y. Kamada, S. Kobayashi, M. Suzuki, J. Appl. Phys. 100 (2006) 023902. [13] A.D. Simone, R.V. Kohn, S. Mueller, F. Otto, Recent analytical developments in micromagnetics, in: G. Bertotti, I.D. Mayergoyz (Eds.), The Science of Hysteresis, vol. II, Physical Modeling, Micromagnetics, and Magnetization Dynamics, Elsevier, Oxford, 2006, pp. 269381. [14] G. Bertotti, V. Basso, M. LoBue, A. Magni, Thermodynamics, hysteresis, and micromagnetics, in: G. Bertotti, I.D. Mayergoyz (Eds.), The Science of Hysteresis, vol. II: Physical Modeling, Micromagnetics, and Magnetization Dynamics, Elsevier, Oxford, 2006, pp. 1106. [15] F. Fiorillo, L.R. Dupre, C. Appino, A.M. Rietto, IEEE Trans. Magn. 38 (2002) 1467. [16] H.A.M. van den Berg, J. Appl. Phys. 60 (1986) 1104. [17] H.A.M. van den Berg, IBM J. Res. Dev. 33 (1989) 540. [18] P. Bryant, H. Suhl, Appl. Phys. Lett. 54 (1989) 2224. [19] A.D. Simone, Arch. Rat. Mech. Anal. 125 (1993) 99. [20] A.D. Simone, R.V. Kohn, S. Mueller, F. Otto, Commun. Pure Appl. Math. 55 (2002) 1408. [21] J. Sun, P. Monk, IEEE Trans. Magn. 42 (2006) 1643. [22] A. Brandt, Comput. Phys. Commun. 169 (2005) 438. [23] J.P. Eberhard, Phys. Rev. E 72 (2005) 036616. [24] D. Cioranescu, P. Donato, An Introduction to Homogenization, Oxford University Press, Oxford, 1999. [25] G. Pisante, ESAIM: Control, Optimisation and Calculus of Variations 10 (2004) 295. [26] T.Y. Hou, X.H. Wu, Z. Cai, Math. Comput. 68 (1999) 913.

Anda mungkin juga menyukai