HAL Id: jpa-00224927
https://hal.archives-ouvertes.fr/jpa-00224927
Submitted on 1 Jan 1985
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
DIRECT RELATION BETWEEN HYSTERESIS AND DYNAMIC LOSSES IN SOFT MAGNETIC
MATERIALS
G. Bertotti
To cite this version:
G. Bertotti. DIRECT RELATION BETWEEN HYSTERESIS AND DYNAMIC LOSSES IN SOFT MAGNETIC MATERIALS. Journal de Physique Colloques, 1985, 46 (C6), pp.C6-389-C6-392.
�10.1051/jphyscol:1985672�. �jpa-00224927�
Colloque C6, supplCment au n09, Tome 46, septembre 1985 page C6-389
DIRECT RELATION BETWEEN HYSTERESIS AND DYNAMIC LOSSES I N SOFT MAGNETIC MATERIALS
G . B e r t o t t i
I s t i t u t o EZettroteenico NazionaZe GaZiZeo Ferraris and G.N. S.M. -C. I.S.M., 1-20125 To'orino, I t a l y
Rksumk - La perte dynamique en exchs d e s matkriaux polycristallins magnb- tiques doux est exprimCe en fonction de l a perte par hystCrCsis et l a dimen-
sion moyenne d e s grains, sans introduire aucun autre parametre a r b i t r a i r e . L'Cquation obtenue permet de calculer correctement soit l a valeur absolue, soit l a dgpendance de l a dimension des grains des pertes en exc&s mesurkes s u r des t a l e s de FeSi 3% non-orientkes.
Abstract - On the b a s i s of a statistical theory of l o s s e s recently developed by the author, the excess dynamic loss of polycrystalline soft materials i s expressed directly in t e r m s of the hysteresis l o s s and the average grain size, with no adjustable parameters. The obtained l o s s equation correctly predicts both the absolute value and the dependence on grain size of excess l o s s e s measured in nonoriented 3% SiFe laminations.
I - INTRODUCTION
In a recent s e r i e s of papers / 1-41, the author has proposed a new, general inter- pretation of eddy current losses, based on the use of statistical theoretical concepts.
It is well known that statistical methods prove particularly powerful when the inves- tigated process can be decomposed into a sum of statistically independent contribu- tions. In magnetization processes, these independent contributions cannot evidently be found at the level of single spins, and neither a t the one of single Bloch walls , which show, in general, strong mutual interactions 1 3 1 . Nevertheless, the author has proposed that such independence exists at the level of what he has called I t mag- netic objectstt (MO). A MO corresponds to a group of neighbouring walls whose evolution, just due t o the presence of local wall-wall interactions, is s o strongly correlated that can be treated in t e r m s of the dynamic properties of an equivalent, single object. The introduction of t h e concept of MO leads t o a number of important simplifications in the physical description of the magnetization process, which a r e discussed in detail in Ref. / 4 / and can be briefly summarized a s follows.
i ) - The dynamic behaviour of a single MO is, t o a good approximation, the s a m e a s the one of the plane 180' wall considered by Williams, Shockley and Kittel in t h e i r classical paper / 51. T h e r e exists a simple proportionality b?tween the excess dy- namic field Hexc acting on a MO and the magnetic flux r a t e @ correspondingly ob- tained, expressed by the equation
Hex, = OG(W) 6 (1)
where a is the electrical conductivity of the material and, according t o paper / 5 / , G(W) = 4 [ x 1 / ( 2 k + 1 ) 3 ] / ~ 3 = 0.1356 ... .
ii) - At each magnetization r a t e i = 4 ImaX f m s the domain structure dynamics can be described in t e r m s of ii simultaneously active MOs randomly placed in the sample Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1985672
C6-390 JOURNAL DE PHYSIQUE
c r o s s section S. A s a consequence, the dynamic l o s s splits into the classical l o s s and an e x c e s s contribution related t o the individual behaviour o f each MO / 4 /
p ( e x c ) = ij=~,,, = a G ( W ) s i2 / E , (2)
where Hex, i s the same as i n eq. ( 1 ) and it has been taken into account that the flux rate 6 provided b y each active M 0 must be a fraction l/i? o f the total flux rate S i .
i i i ) - i s expected t o be an increasing function o f the field H e x c , expressible, t o a f i r s t order approximation, i n the f o r m
Essentially, assuming this direct dependence o f ?i on He,, i s based on the idea that each MO experiences a d i f f e r e n t , local coercive field, so that, at each magnetizing frequency f m , only the MOs for which the applied field i s greater than tpe local coercive field will be active. When fm i s increased, a higher flux rate @ i s demand- ed f r o m the active MOs, and this implies an increase o f the applied field ( e q . ( 1 ) ).
T h i s increased magnetic pressure, acting uniformly in the sample c r o s s section, will overcome the local coercive fields o f further MOs, thus inevitably producing also an increase o f "n ( eq. ( 3 ) 1. Equations ( 2 ) and ( 3 ) lead t o the l o s s expression
~ ( e x c ) = i ~ o V o [ ~ 1 + 4 o G ( W ) S i / ^ n ~ v ~ - l ] / 2 = t 8 1 m a x f m ] o ~ . (4) T h e last approxiKate f o r m becomes in gef?eral quite accu'kte when f m >, 10 Hz, and points at the field V o a s the fundamental parameter o f the theory.
A systematic comparison with experiments / 4 / shows that eq. ( 4 ) has a wide range o f validity, f r o m v e r y fine grained nonoriented steels t o SiFe single crystals. In R e f . 141, however, V was essentially treated a s an arbitrary, adjustable parame- t e r , used t o fit experimental data. Actually, V o has an important physical meaning, 0 since, a s previously mentioned, it i s related t o the distribution o f the local coercive fields o f d i f f e r e n t MOs. In the present paper, the basic physical consequences o f t h i s fact will be discussed i n t h e case of fine grained materials. An equation will be derived which e x p r e s s e s the e x c e s s dynamic l o s s directly i n t e r m s o f the h y s t e r e s i s l o s s and t h e average grain s i z e , with no adjustable parameters. I t will be shown that t h i s equation correctly predicts both the absolute value and the dependence on grain size o f the e x c e s s l o s s measured in nonoriented 3% SiFe laminations.
I1 - THE MODEL
Let u s consider a polycrysta_lline magpetic lamination. W e shall assume that it con- tains, in i t s c r o s s section, No available MOs and that the distribution o f the local coercive fieids o f d i f f e r e n t MOs i s flat, with constant density l / v o . In other words, the number N o f MOs with coercive field lower than H i s simply
~ = H / v , . ( 5 )
In quasi-static conditions, t h e value 0f.H at each point o f the magnetization loop will be determined b y the requirement o f having at least one active MO ( t h a t i s
- no- 1 ) / 4 / .Conversely, i n dynamic conditions, H will be, at each point o f the loop, greater than i t s static value b y an amount AH=HeXc , which, according t o eq. (5), corresponds t o an increase AN =&Eo =H,,</ V o o f the number o f active MOs, i n agreement with eq. ( 3 ) . From t h i s point o f m e w , t h e r e f o r e , the fact that eqs. ( 2 ) - ( 4 ) are capable o f correctly predicting a great variety o f d i f f e r e n t experimental results c o n f i r m s the general existence, i n soft materials, o f a f a i r l y flat distribution o f local coercive fields. On the other hand, the approximate validity of eq. ( 5 ) i s also supported b y the investigations o f Del Vecchio on low carbon steels / 6 1 .
T h e magnetization reversal i n each MO will be described, for simplicity, a s a 180' reversal, and the spread o f the directions o f t h e e a s y magnetization a x e s i n d i f f e r e n t
Is with its weighted average <Is> = Is<cosO > over a l l the easy axes closest to the applied field direction ( < I ~ > r 0.85 Is r 1. 7 T in isotropic F e and SiFe samples 1. a Thus, when a M0 r e v e r s e s completely its flux, it gives a contribution A1 = 2 < 1,s /No to the total magnetization, and the number of MOs involved in a cycle of peak mag- netization Imax will be
On the other hand, eq. (5) implies that the maximum local coercive field reached in each magnetization half cycle will be 2 Hhyst, with Hhyst = P ( ~ Y s ~ ) / i ( P ( ~ Y s ~ ) is the hysteresis l o s s 1. We therefore obtain, taking account of eqs. 15) and (6),
Equations (4) and (7) give evidence of a direct physical connection between dynamic and hysteresis l o s s e s in soft magn;lle_tic materials.
What can we say about the number N, of available MOs appearing in eq. ( 7 ) ? Here, we shall explore the co;vnsequences o f assuming that each MO actually coincides with a single grain, that is No = S / s 2 , with s equal to the average grain size. This con- dition i s equivalent to assuming that each grain is characterized by a fairly uniform local coercive field, and that its domain structure behaves in a highly coherent fash- ion, while different grains in a same c r o s s section evolve quite independently from each other. All these conditions a r e expected t o hold in annealed samples character;
ized by a grain size much smaller than the sample thickness d. Conversely, when s + d , the inhomogeneities of coercive field inside each grain begin to be important and the number of MOs becomes definitely l a r g e r than the number of grains. In fact, the material tends to approach the limiting condition, met in monoc~ystalline Sam- ples, where each MO coincides with a single Bloch wall 141. With No = S / s 2 , we obtain from eas. (4) and (7)
which expresses the excess dynamic l o s s in t e r m s of the hysteresis l o s s and of the average grain size. We s t r e s s the fact that eq. (8) contains no adjustable parameters.
We have compared the predictions of eq. (8) with two s e t s of available experimental results, both referring to nonoriented SiFe laminations of variable grain size / 7,8/.
Figure l a reports the measured excess l o s s e s together with the predictions (broken lines) obtained from eq. (8) on the basis of the average behaviour of the measured
hysteresis l o s s ( Fig. l b ) . In spite of the simplicity of the proposed model and the absence, in eq. (a), of any a r b i t r a r y parameter, we obtain an essentially correct prediction of the absolute value of the excess loss, a s well a s of its dependence on both grain s i z e and peak magnetization. On the other side, the fact that the theoret- ical curve becomes substantially higher than the measured l o s s when s 3200 pm simply points, a s previously discussed, to a progressive splitting of each grain into several independent MOs.
I11 - CONCLUSIONS
The previous results show that the concept of MO, f a r from having a purely math- ematical character, is instead of relevant physical importance in the description of the magnetization process. The fact that the proposed model can provide a correct quantitative interpretation of the dynamic loss behaviour in c a s e s where previous models, a s the one of P r y and Bean 191, completely fail is the consequence of two basic assumptions : i ) - the single grains, rather than the single Bloch walls, a r e the fundamental objects providing coherent magnetization changes in fine grained materials ; ii) - the spatial fluctuations of local coercive field strongly reduce the
C6-392 JOURNAL DE PHYSIQUE
number of simultaneously active regions of the material c r o s s section, thus giving r i s e to a magnetization process much more inhomogeneous than expected on the ba- s i s of the geometrical arrangement of magnetic domains, and to a consequent large increase of the excess l o s s contribution.
Finally, we wish to point out that, although in this paper we have only discussed the l o s s behaviour a t fm = 50 Hz, eq. (8) provides indeed the complete dependence of
~ ( e x c ) on fm . We have verified that even the prediction of a linear dependence of the excess l o s s per cycle on rm is in excellent agreement with experiments.
Fig. 1 - Dependence of excess dynamic l o s s e s ( a ) and hysteresis l o s s e s ( b ) on av- erage grain size in nonoriented 370 SiFe laminations of thickness d = O . 35mm (open c i r c l e s after 171, I,, = 1 T ; other symbols after / 8 / , Imax = 1 . 5 T 1. The broken lines in Fig. l a represent the theoretical predictions obtained from eq. (8) on the basis of the average behaviour of the measured hysteresis loss.
REFERENCES
"
n cn
Y
/ I / Bertotti, G., J. Magn. Magn. Niater. 41 (1984) 253.
/ 2 / Bertotti, G., J. Appl. Phys. 54 (1983) 5293.
/ 3 / Bertotti, G., J. Appl. Phys. 3 (1984) 4339, 3 (1984) 4348.
/ 4 / Bertotti, G . , J. Appl. Phys. 57 (1985) 2110, 57 (1985) 2118.
/5/ Williams, H. J.. Shockley, W. and Kittel, C., Phys. Rev. 80 (1950) 1090.
/ 6 / Del Vecchio, R. M. and Charap, S. H., IEEE Trans. Magn. MAG-20 (1984) 143 7.
/7/ Matsumura, K. and Fukuda, B., IEEE Trans. Magn. MAG-20 (1984) 1533.
181 Bertotti, G . , Di Schino, G., F e r r o , A. and Fiorillo, F . , this Conference.
/ 9 / Pry, R.H. and Bean,C. P., J. Appl. Phys. 2 (1958) 532.
f m = 50Hz I I
I I I
fm=50Hz
*
*
-4
n cn
Y
I I
2 - I \
w n U % -a
0.5
0
0 150 300 0 150 300
Grain size ( p m )
w 3
h C m
2
2-a
0
I -
I
I 1.5 T
I I
j .
I I
I / *
I
I *
- I
/ I v .* -
+.a
0:
C.fiO- 0 v
f
09P0 0*, \
M /k.Q I I & @ 1T
'
OoO0 O 00
I , (a) ( b ) , I