General Properties of Power Losses in Soft
Ferromagnetic Materials
GIORGIO BERTOTTI
Aburec—Measurements re reported of the ass pe ecle (ina
at ax waveform) vers mugneving frequency fx fo = 100 or
Predict the dependence of dn
ith an sppresieation within 10 percent
casts, The bear of graimrieted Se which
resents some spec complentes, dscused in particular et
1. Intropuerion
HE AIM of the present paper is to provide a contti-
Dution to the understanding of magnetic losses in soft
metallic ferromagnetic materials that may be useful both
to the fundamental physicist interested in microscopic
magnetization processes and to the engineer involved in
the electrotechnical applications of soft steels. Basically,
We shall examine the dependence of power losses on mag
netizing frequency f, and peak magnetization Jax. tying
to clarify how this dependence is related tothe parameters
like grain size that define the microstructure of a given
material, as well as to those like domain size that char-
acterize its magnetic domain structure. Our starting point
will be a simple and important phenomenological princi
ple, which has been commonly adopted in loss invest
gations for & long time, the so-called separation of losses
[1], 2). According to it, the average power loss per unit
volume P of any material is decomposed into the sum of
hysteresis and a dynamic contribution, which are sepa-
rately investigated,
P= pie) 4 pide
a
Manuscript received December 10,1985; revised My 23,1987. Thie
snort was soporte in at by te lian Mine of clon apd the
Frog ia Ei eat Gat CXR NEA
IEEE Log Numer #717388,
(0018-9464/88/0100-0621801.00
with PO” equal tothe area of the quasi-static hysteresis
Joop times fa. The physical reason for such a decompo-
sition is that P°™ originates from the discontinuous
character of the magnetization process at a very micro.
scopic scale, whereas P'” is associated with the mac-
roscopic large-scale behavior of the magnetic domain
structure, In fact, a proper statistical treatment of the loss
problem [3]. [4] shows that (1) is naturally obtained
‘whenever the characteristic time seales ofthese two pro-
cesses do not overlap.
‘The important conceptual simplification introduced by
the separation of losses is readily appreciated, because the
interpretation of P'"”—the quantity of principal interest
in this paper—can then be based on simple macroscopic
domain models, paying no heed to the intricate fine-scale
details of wall motion. In this connection, the so-called
‘classical model (2), [5] even disregards the very pres-
cence of magnetic domains and assumes a magnetization
process perfectly homogeneous in space. In te case of a
Tamination of thickness d and inthe range of magnetizing
frequencies where the skin effect is negligible, this model
predicts (sinusoidal flux waveform)
POM PS = aod f/6 (2)
‘where o is the electrical conductivity ofthe material. The
classical model is, however, too gross a simplification
‘As a consequence of domain effects, the dynamic loss
Pls. generally found to be definitely larger than
PIS), the diference between them, called excess loss
PO, being in many cases larger than P™) itself. By
expressing the total loss as
P= pi 4 pla 4 plow, 8)
the central problem becomes then to understand the origin
and the properties ofthe excess term P'*’.
‘The guideline in the interpretation of excess losses has
been, for a long time, the model proposed by Pry and
Bean [6], in which the dynamic loss of an infinite lami-
nation containing a periodic array of longitudinal domains
of width 2L is calculated from Maxwell's equations, This
model indicates that the fundamental parameter control-
ling excess losses should be the ratio 2L/d between the
domain size and the lamination thickness and predicts
Pie ce peta,
when 2L/d <1
1988 IEEEpiss (: 632-1) pow,
when 26/d >> 1 @)
In spite ofits remarkable achievements, however, even
the Pry and Bean model turns out tobe of limited validity,
essentially because of its highly idealized character, In
microerystalline materials, infact, itis commonly found
that PS! ~ PS even if 2L/d << | [7]-I9]. On the
‘other hand, P'“/f, shows in general a nonlinear depen-
{dence on f,, which cannot be easily explained in terms of
G) [7]. Im the case of single crystals or grain-oriented
laminations having 2L/d >> 1, the departure of mea
sured losses from Pry and Bean predictions has been re
lated to various effects, like the presence of imegularities
in the wall motion, domain multiplication, or wall bowing.
[10)-[16]. As to ‘fine-grained materials, attempts have
‘been made to connect excess losses with Barkhausen noise
171, oF to attribute them to continuous rearrangements of
the domain configuration [18]. However, no general
agreement as tothe role of all these mechanisms in deter
‘mining the actual loss features of soft materials has yet
been achieved.
‘A new conceptual framework, capable of including
most ofthese effects in a unified treatment has been pro
posed recently by the author, on the basis of a statistical
Approach to the loss phenomenology. The theory, devel-
‘oped in [3], [4], 19]-[22] and here briefly reviewed in
Section II, leads to the conclusion thatthe large-scale be-
havior of magnetic domains can be described in terms of
the dynamics of # statistically independent magnetic ob-
jects (MO), each corresponding to group of neighboring,
interacting domain walls, and reduces the loss problem to
the investigation of the main physical properties of as a
function oF fn, Apa: and the material microstructure. This
approach can be successfully applied 10 a variety of dif.
ferent material, by investigating the physical meaning of
the concept of MO in each case. In particular, it has been
shown that a single MO can be identified with a single
Bloch wall in grain-oriented materials with large domains
[20], whereas, in microcrystalline materials, the whole
domain structure inside a single grain plays the role of a
single MO [21), [22]
Starting from these premises, in this paper we present
the results of « twofold investigation. Inthe first place,
‘we have carried out a systematic experimental study of
the loss behavior versus Zn. and fin various soft mag-
netic materials, consisting of measurements of the loss
per eycle versus fy in the range 0 = fy = 100 or 400
Hz, depending on the material) performed—at several
values of peak magnetization Jqu.—on SiFe alloys of dif-
forent texture and grain size, Atmeo iron, NiFe alloys,
and amorphous ribbons. By these results, reported in Sec=
tion TIL, we have tried to provide a source of information
‘on the properties of eddy current losses in soft materials,
‘as well as a basis for @ proper testing of theoretical loss
models. In this connection, our second goal has been to
interpret these experimental results in terms of the men
tioned dynamics of MO's, and to relate the observed
properties ofthe parameter f tothe microstructure and the
Somain structure of the investigated materials. As fully
discussed in Sections TIT and TV, several interesting con-
clusions have been obtained by’ this analysis. First, the
theory gives a natural interpretation of the general fact
thatthe dynamic loss per cycle is a nonlinear function of
Jay and shows, in particular, that this nonlinearity is ap=
‘proximately consistent with a simple law PO /f, =
Aavfj, for a wide variety of materials, Secondly, the the-
‘ory is also applicable to nonoriented, microcrystalline
laminations having a fine domain structure (22 /d << 1)
In this case, the absolute value of the constant k; can be
calculated, under reasonable assumptions, directly in
terms of the values of the hysteresis loss and the average
grain size, obtaining good agreement with experimental
results, Finally, the theory reduces the dependence of ex
cess losses on Both Ips and fa t0 & common mechanism,
the competition between the applied field and highly in:
homogeneous internal counterfelds governing the dy:
namics ofthe single MO’s. This feature ofthe theory has
important practical consequences, since it permits one t0
extract the dependence of dynamic losses on both Igy and
fa from a single loss curve ata given value of fey. and 10
‘express this dependence in closed analytic form. In par-
ticular, it will be shown in Section IV that, in the case of
grain-oriented three-percent SiFe, the knowledge of only
{oo points of a single loss curve permits one to predict
the loss behavior Versus Jay, and f With an approximation
that, for the data presented in this paper and in the range
Jy % 20 He, is within +10 percent
I. Sranisticat. InreRPRerarion oF Eooy CuRRENT
Losses
According to the statistical theory of eddy current losses
0, and by the magnetic
field Vo. Equations (7) and (8) lead tothe nonlinear excess
loss expression
Hog = PY}
Agha (VI-¥ 40GB —
Meas ha (9)
{In the following sections, we shall undertake a general
study of the properties of the function (Haq: {P}) in
different materials to clarify the real value and applica-
bility of the proposed theoretical approach to magnetic
losses, and, in particular, the physical meaning and range
‘of validity ‘of the linear law (8). Before passing t0 the
Presentation of the experimental results, however, we
Want to recall a simple and useful procedure, already pro
posed in [20], by which the function A(Hac: {P}) can be
directly obtained from experiments. According to (5),
HO /Fios is determined by the values of H™ and
Hig. Both of these quantities can be directly evaluated
from experiments, as can be seen from their very defini-
tion. Therefore, if we plot a generic loss curve in terms
OF O/H Ves Hege—instead Of P fy Versus fy a8
y/2
usual—we shall directly obtain the behavior of the func-
tion (Hey {P}). It is interesting to notice that a quite
similar loss representation has also been proposed and
used, although on more empirical bases, in [14], [15].
‘According {0 the physical interpretation summarized in
2), is expected to depend explicitly on H,x. only, while
‘other quantities (€.8., fna.) should have only an indirect
influence, through induced changes of the parameters
{P}. This means that we may obtain valuable informa-
tion on the microstructural properties of different mate-
rials, or on the modifications of the domain structure be-
havior upon changing Jma, in a same material, simply by
using the a versus Hox. representation. For example, if 10
changes occur in the magnetic domain structure of a given
material as a function Of Zyc, We expect that the loss
curves referring to different Values of fan, Should all re-
duce to a single curve, when represented in terms of
versus Hac. This isthe point of view that will be adopted
in the discussion of the experimental results reporeed in
the following section.
IL, Experimenta RESULTS AND Discussion
Losses were measured on grain oriented (GO) and now-
oriented (NO) three-percent SiFe alloys, Armco iron, mu
metal, 50-50 NiFe alloys, and amorphous materials, as
listed in Table I. Two sample geometries were employed
in the experiments: standard Epstein frames and toroi
dally wound ribbons. The power loss was obtained by
micans of a high precision analogic wattmeter, permitting
measurements at controlled sinusoidal magnetic lux
waveform in the range of magnetizing frequencies 1-400
Hz. The hysteresis loss was measured in general through
the ballistic method. In the case of NO SiFe and Armco
iron, however, where the error in the ballistic evaluation
of the hysteresis loss was found to be comparable withthe
variations of dynamic loss at low f, the bysteresis loss
‘was obtained by directly extrapolating the dynamic loss
ccurve down tof ~ 0. Far each sample, the loss behavior
in the range 0 = fy = 100 Hz (0 fy =< 400 He for the
amorphous ribbons), and at several (four to five) values
of peak magnetization Ing, Was determined.
‘The oblained P /fy verss fq curves are reported in Figs.
1-10 (left-hand portion of each figure). In the right-hand
part of the figures, the very same results are reported in
ferms of (He; {P}) versus He, according to the rep-
resentation discussed in Section If. We shall now examine
these results in more detail, focusing our atention, in the
first place, on GO and NO SiFe, because these two cases
exhibit some basic features of power losses common, to
some extent, to all the investigated material.
Grain Oriented Three-Percent SiFe
Fig. 1(@) reports the behavior of the loss per cycle and
unit mass P/df, (8 is the density of the material) versus
Js measured at five different values Of Iyuy. The corre
sponding representation in the plane (#7... 7) is given in
Fig. 1(b). This representation clearly shows the fact al=
ready stressed in previous works [20], that GO SiFe is aresist
— cP GRE Gy WES oe ~
lef eacrenres sean free Peano oss curves are easily obtained by a least-square linear fit-
Soa | SESE yy | ting of the experimental point in the (gc) pane,
Fig. 1 (a) Los pry and uit mass sol lax waveform verses
‘Segvuzing equeny in sin onened tree percent Sie, at erat
‘ales of peak mapatzon fa, se Tle I for fer infomation).
‘Vanou sols potent experimen loss pir ir sae age
than uncer of messre aly ken ines ar eit ces
‘ial om (2,3), (ah valus ol parameters hy and Vo
‘ner cenaitlow. (0) Represtion of ste expeimesal pms
{nem of paracetf Jelied Oy
Tae en diayee je ican he seen th th representation ingle
Ieee tee nt agi ines of fom (a). Thee slope deer
Pines eld yw sven hy ther intersection mith # ai
eral valet of and Yyste ely shamed est ate me
Tis Yound tat Yar always coma vaue Vy = 0.1 Am ine
Mem Ibi y= Toten gs = 17. = iD we fog = 1S
Fin 228 hen ge 23
hd (ves amie eld
cease where the linear law (8) for the dependence of on
Hog is tvicty followed, We point out that this result c
not be merely interpreted in terms of a first-order expan-
sion of (ix; {P}) around He = 0, since the linear
term Hy. /Vo is by no means small with respect 10 Ny, a8
can be easily checked directly from Fig. 1(b). The values
of the parameters iy and Vy associated with the different
shown by the broken straight lines in Fig, 1(b). We obtain
Vo = 0.15 A/m, independent of Igy. Whereas Ap in-
ceases fom fy = 1, when Ing, = 1.7 T, up to fo = 25
When Jay ST. The knowledge of my and Yy permits
‘one to predict, through (2), (3), (9) the behavior of P/Sfe
versus fat Fay. The broken lines in Fig. 1(a) have been
‘obtained just by this procedure. One can notice the ex
cellent agreement with experimental points, and, in par-
ticular, the ability of (9) to describe the observed loss
nonlinearities. This result leads toa first, interesting con-
clusion about the behavior of excess losses in GO SiFe.
Equation (9) implies infact, that, for sufficiently high val:
ues of f_ (commonly above a few Hz) the excess loss per
cycle can be approximated as
PN fy = Slaw (YOO SVoleasta — Ro¥/4) (10)
land therefore becomes essentially proportional to Vf.
‘Thus the statement, often found in the literature [2], 24},
that dynamic loss nonlinearities are limited to low mag
netizing frequencies, seems somewhat inaccurate. Out re=
sults show, on the contrary, that such nonlinearities ap-
proximately obey @ simple vj, law in the whole
investigated, range, so that the dependence of the total
loss per cycle P/fy on fy is essentially of the form
Pf = bo + iba + bn ay
where the meaning of ky, ky, A: can be easily deduced
from (2), (3), (10). It can be checked from (8), that,
in the f, range where P/f, 2 Vfy. the number m of
active MO’s increases essentially as vf. This conclusion
is in good agreement with the results reported in [14],
US}, (25).‘The fact that the field Vy is found to be independent of
Togs makes it an intrinsic parameter of the material, con:
‘nected to its microstructural properties, rather than a sim-
ple fitting parameter. We shall see that this interesting
Conclusion applies to most of the materials fellowing the
linear law (8). It has been suggested in a previous paper
[20] that the microstructural interpretation of Vp should
involve a relation ofthe form Vp ~ Hyy/Ny between Vi,
the quasistatic coercive field Hy, and the number Ny of
'MO’s available for magnetization in a given cross section
of the material. While the meaning and validity of this
connection in GO SiFe needs further study and will be
analyzed in a future paper, a possible interpretation ofits
origin in microcrystalline materials like NO SiFe has been
recently proposed (21], [22] and will be discussed in the
next subsection,
{AS to the dependence of Ay OM Imus, We recall that fy
represents, acconding to (8), the number of MO's which
are simultaneously active in the limit of low magnetizing
frequencies, and that, in the material under consideration,
‘each MO essentially coincides with a single Bloch wal.
AAs previously pointed out, ly = 25 at 1OW Ina & Value
“which is ofthe order ofthe number of walls present inthe
cross section of the investigated sample in the demagne
tized state. This indicates that a faitly regular and coher-
cent motion of all walls takes place at 1oW Igy. The pro=
‘gressive onset of irregularities in the wall motion with
increasing Jy, is then clearly pointed out by the steady
decrease of fy, down 10 ~ 1 (one active MO at a time),
which represents a quite general lower limit for the values
‘flyin any material [20]. An interesting test of the phys-
ical meaning of ry can be made by considering the mod-
fications in the loss behavior consequent to dierent pro-
cedures of sample demagnetization. Each of the loss
‘curves reported in Fig. 1(a) refers to a sample prelimi:
narily demagnetized at a frequency of 1 Hz. If the same
‘measurements are performed after $0 Hz demagnetiza-
tion, different results are obtained. The example reported
in Fig. 2(2) refers t0 Im, = 0.7 T and shows a definite
dynamic loss decrease after 50 Hz demagnetization. This
effect progressively disappears when higher values of
nay are considered. These results can be explained qual-
itaively by noticing that, by demagnetizing the sample at
50 Hz, we activate domain multiplication processes that
introduce a higher number of domain walls in the sample
cross section, with a consequent reduction of the excess
loss. The newly generated walls, however, are easily an-
niilated at high inductions, and this explains why the loss
behavior is afected by demagnetization only when Imac =
1 T. However, this interpretation does become really ev-
ident and expressible in quantitative terms when the same
loss data are represented inthe plane (Hx, #) Fig. 2(b).
‘The difference between the two loss curves results now in
an increase of Rp, from My = 25 up toy ~ 40, indicating
the higher number of domain walls introduced in the sam-
ple cross section, while the slope 1/V of the loss lines
has not changed at all, as expected from the mentioned
‘iicrostructural character of the field Va,
Fig. 2. Modifications in os behavior fein one thee percot SF
Tag = 0.9 T, consequent dierent procedures of ple demap
eibion. See capon af Fig. for mening af employed Fepeeas
‘As a final remark, we point out that the linear law (8)
is strictly followed by GO SiFe even after extensive plas-
tic straining or upon the application of extemal tensile
stresses [20] and well describes also the behavior of three
percent SiFe (110) {001 single crystals [19], [20}.
Nonoriented Three-Percent SiFe and Other Materials
Following the Linear Lav Mt = Ay + Hlx./Vo
Figs. 3 and 4 repor the loss behavior measured on two
laminations of NO three-percent SiFe, characterized by
different grain size and coercivity, In these material, the
average size of magnetic domains is certainly smaller than
the grain size, which, in turn, is smaller than the lami-
nation thickness (see Table 1). As discussed in the Inro-
duction, any model similar to that of Pry and Bean would
predict, inthis case, no excess loss at all (See (4), in clear
contradiction with the experimental data reported in Figs,
3¢a) and 4(@). We will show now how this contradiction
can be resolved by making use of the theory discussed in
Section I,
A fist result of interest is that, as shown by Figs. 3(b)
and 4(b), the linear law (8) is stll fulfilled. However,
contrary to the ease of GO SiFe, the term Hox / Vo is now
predominant at all values of Inn, and f, of interest. The
parameter fy does not play any significant role and can be
neglected. Physically, this simply means thatthe number
of new MO's becoming active in dynamic conditions soon
destroys any memory of the quasi-static state described
by My, Equations (8) and (9) can then be conveniently ap-
proximated as
(Hoes {P}) = Hexe/ Yo (2)
PI) = 1 VAM, = 8VoG™ IV (Inacha) (13)
Equations (12) and (13) have a number of remarkable
consequences. Assuming, in analogy with the results ob-
tained in GO SiFe, that Vis an intrinsic parameter ofthe
‘material, independent of ae, We expect from (12) that all,“Prevent sre gs]
®
teanalopous to that of Fi 1. Forths mtral (canbe apronimate
by (2) which doce mt comin parameter My Value of bel Yo deter
‘lad rom (Pi A opeetticn of expetoetal ports Va B07
[A/m Two coninuous straight Segnees mig Na} epee hysteresis
th iss coobutons
Pg 4 Sam tn Pig 3 for sample wr a = 0.12 A/
loss points at different values of Ing. should reduce 19 @
single straight line, when represened inthe plane Hoc
2). This prediction is well confirmed, apart from some
scattering of the data related othe precision of the mea-
surements, by Figs. (6) and 4(b). In terms of the behav-
ior of P/f versus ja, this implies that the single param
eter Vo determines, through @), 3), (13), the complete
‘dependence of dyramic losses on Igy ad fay nd, i pa
ticular, that the excess loss per cycle follows a simple
Jf law. The broken lines in Figs. 3a) and 4(a) were
calculated from (2), (3), and (13), assuming in each case
the fixed value of Vp obtained from the corresponding
(Hc) representation: Vg = 0.07 A/m in Fig. 3) and
Vy ='0.12 A/min Fig. 4a)
is evident thatthe fundamental problem at this point
isto justify, on physical grounds, the absolute value of
the field Vp. so as to explain the microstructural reason
for the presence of an excess loss contribution in magnetic
laminations having domains definitely smaller than their
thickness. A possible key to this problem has been pro
posed recently in [21], [22], on the basi of the followit
Schematic description ofthe magnetization process in fine-
grained materials. Adopting a point of view slightly dif-
ferent from that of Section Il, we can equivalently inter-
pret a single MO as a region of the material cross section
characterized by a fairly homogeneous local coercive
field, so that it will become active as a whole when the
applied field overcomes a given threshold. In an annealed,
fine grained lamination, where the main fluctuations of
local coercive field are expected to take place from grain
to grain, a single MO is then naturally identified with the
whole domain structure delimited by the cross section of
a single grain, so that the total number of MO's present
‘a given eross section of area of the lamination will
be
N= s/t as)
with s equal to the average grain size. The statistical dis
{ebution of the values of local coercive field st which f=
ferent MO's will become aetive can be assumed to be, to
4 first approximation, flat, with constant density 1/¥q (he
field Vp tepresents therefore the average minimum sepa-
ration between different local coercive field values). AS
fully discussed in (22), these assumptions imply that in
dynamic conditions, when the applied field is greater than
its comesponding quasi-static value by the amount He,
all the MO's inside a portion of width Ha. ofthe men-
tioned distribution of local coercive fields willbe simul
taneously active. Since this number is simply =
Hgc/Vo, (8) 8 staightforwardly obtained. On the other
hand, a8 a consequence ofits very definition, the field Vy
rust be directly related 10 the macroscopic quasi-static
coercive field Hy, of the material. In fact, as shown in
{21}, (22) the value of Hye in the loop of peak magnet-
Jaation fy is elated to Vp by the expression
Hy = NoWoloas 21.) (as)
where Ny is given by (14) and ¢ 1, is the value of mac
roscopic magnetization when all grains are saturated, each
along its most favored easy magnetization axis (
1.$°T) are due tothe fat thatthe expression we have used
for the hysteresis loss becomes a bad approximation, while
those a 1OW Inu arise from neglecting the term MV /4 of
(10), Actually, this term becomes more and more impor-
{ant with decreasing In, since Mp progressively increases,
‘while, at the same time, the first term of (10) decreases
48 View. Incidentally, these considerations show that both
these sourees of error should be small around 1.5 7, 0
thatthe values of Cp and C; are indeed best determined
by considering, as we have done, the loss curve corre-
sponding t0 Inay = 1.5 T. Of course, (18) could be ap:
plied not only to GO Sie, but in general to any material
‘obeying the linear law (8) or (12), with Vp independent of
Tuas td Hy appFOXimately proportional 10 Imus (See Figs
3% and 11).
Equation (18) gives evidence of the existence of wo
basic contributions tothe dependence of the total loss P/6
on Jpg at a fixed magnetizing frequeney, one proportional
to Fag the other to Ii. The average exponent expressing
the law of dependence of P/5 on Ima is therefore ex:
pected 10 be included between 1.5 and 2, depending on
the values of Cy and C; in (18). This conclusion should
‘be compared with old empirical rues, offen mentioned in
the literature [5], expressing Pas P ~ I. Our treatment
provides a physical interpretation to these rules and shows,
how they may arse from the interplay of the three terms
which, according to (3), contribute to power losses.
V. Conctusion
‘The theoretical interpretation of the experimental re-
Sults on power losses in soft materials presented in this
paper has shown that the function Ff; (P}) defined
in Section I provides a promising convenient too! to look
into the connection between dynamic losses and micro~structure of soft materials In several cases, a single fune-
tion MCP: (PJ) can be associated witha given mate-
ial, even when loss cues at ferent peak
‘magnetization are considered and when in GO SiFe—
{dependence of #07 Ina Hs observed, it provides adie
tional information on the magnetic. domain stracture
‘These results suggest focusing theoretical efforts on the
study of the properies of # in diferent physical situa
tion. Adopting this point of view has permite us ive
4 natural interpetation ofthe general presence of nolin-
Cartes inthe behavior ofthe loss per eyele of any ma-
teria, and fo achieve quantitative prediction of excess
losses in fine grained laminations (see (12)-(17). How.
ver. even inthe cases where we ae not yet able to pre-
fice the properties of from a microscopic mode, the
fact that a single function ACHng: {P}) can Deas
sociated with a given material as relevant practical con-
Sequences. Infact, can then be obtained from a single
loss curve measured at a given vale of Toy, and, though
(the complete dependence of dynamic losses on both
I a fy can then be worked out. By making use of his
result and of some addtional information on the depen-
dence of hysteresis loses on fay, we have shown in See-
tion TV that, in GO SiFe, iis possible to predict the be-
havior of the total loss on Inay and fy from the knowledge
‘of only two points ofa single loss curve, with an approx-
imation that, fr the data presented inthis paper and in
the range fy = 20 He, is within +10 percent. Developing
fenerl microstructural models ofthe bchavior of Fat
{2}) in various materials appears now asthe rman ob
jective of future investigations
AckNOWLEDGMENT
‘The author is deeply indebted to Prof. G. Montalenti,
Prof. A. Ferro Milone, and Dr. F. Fiorillo for their eon”
stant encouragement and interest in this work, a number
of fruitful discussions, and valuable comments and sug-
pestions after their critical reading of the manuscript. He
also wishes to thank B. Genova, A. Gobetto, S. Rocco,
and C. Visca for their technical assistance in loss mea”
REFERENCES
(0) 4. W, Sing and 6. L, Howe, Jr. “Magnet properties ant do
‘mun suc ingrid 9 Sc-es"TEEE Pras Mage 0
MAG-10, pp 198-225, 975
[21 6°. Graham, Ir," Bysclorigin of lous in conducting fee
foapetie materials” Japp Phat 33 pp 82768280, 1982,
10] G Manse “Space ine corlaton proper ofthe mageization
one de ai: Toy Ap Ps 5
(4) 2 general sats ppoach te problem of ey cues
tous Mapm: Magn Marr ttl pp. 255-20, 1964
15) 8M Bozo, Feromagnelm. "New Yor: Van Nosnd. 1981
16) RH. Pry and CP. Reus, “Caleultion ofthe ney loss mag
ese het materia wing adnan el" J Ap Phy 29.
pp 32-338, 1938
in Berry, Moment an GP Sard, Non arty aomay
1 pws er ve egoncy in wamaus sot aga mater
EBs Tre Map, vol MAGLI pp. 1341-1348, 1978
{a1 K. Macumarand'B.Fakad, “Recent developens of 908-0
Chie lectin se sheet" IEEE Tra. Magn vol MAG0. Pp.
isan se
{91 6 Berot G-DiSchine, A. Feo, and F,Foilo, “On the eet
‘fg sis on mage ues of 3 tom one Se," J de
‘Prosoue, vl. 46C6, pp. 388-388, 1985.
11011, W. Shilling, “Domain sracre in 3% Sue single cyst with
‘wenn er (110) 00) TEBE Trane. Mage, vol. MAG. Pp.
Sains, 913
uM] G Bene, FFhoillo, and M.P. Sas, “A new approach othe
‘Say of lose anomaly Per" IEEE Trane, Magn. se, MAG
fp 282-2886, 1981
a2] R Nata and. Irnra
Fregueny dependence of to esses
{Sperone single etl wt 100) OI} craton IEEE
Thon Magn ol MAG 13, pp. 981-988-1973
103] MCancor AT Masooo, Pr Marri and 8. Stpaneca, “Stay
‘of Block wal jramic ad losses in monoerysalne, poly sa
the td nrorpoe merle by meas of opie fecagues
doy Phytol 3, pp 428-2) 8S.
114) Y7Sata An approach estimating the amber of dora walls an
tau curren lose In rasorened 9% Se ape wound cores,
EEE Trt Mage, wo MAG 16 pp 562-572. 198.
lus) ¥ Soaks ind $1 Inaph "Reliecip aston ey cuen ss,
frequency. maxim Ba density and x new preter concerning
the mmber of domain wall i plyrstaline and amos wot
‘magnese materi,” EEE Trane. Mapa Yl MAG.IT, pp. 78-
1490, 198,
(16) 'B'L Bion, dy cereetdominted magoctization roceses
fn fas ented soos kon." IEEE Dron, Mop, vole NAG20,
pest? ist, 1084
a7) Pstacret, Bloch wats comeltion and magnetic tos in feo-
Imago," VEEE Trunt Mage. val MAG: 16 p.756-763. 1978
114) PEL ihop Enhanced ey cuore onda to domain displce
roe J Ma. Magn Mater vol 3. p. 241-249, 1988
119) 6 Bena “Spacetime sorelaon properties of he mageiaion
Proves an ey sure lense Apes Appl Phot. 99
Sslpp 9.4355, 1984
[20] "yaya iteration of ey cae oss i frome
‘anne. app Pwo 9) pp 10-2156, 18S.
(21) in cordetns oe te pyc nerptaton of ey
‘orn nts in femorupnetie tesla" J) Mapm Magn. Maer
Sa 8°37, p. 1396-150, 198,
(22) “Dies rein between hysteresis ad dynamic sie i aft
anc mers" de Py qu, a 46C6, np. 389-392. 1988.
(23) HS Witham, W. Shockey. aed Cite, "Sas of he prope
{on velchy of a feremogoeic domain bound.” Py Re
124) BD. Caliy, raduction to Magnetic Materials
‘adison- Wesley 1972 ch 13,
(25) 2 Beoord. “Sepriton of ose in aria sicon set rom
0.8 wb. mm ick" IEEE Trane: Magn sol MAG20, pp.
Psascisiz, ae
eating, MA:
Giorgio Mertolti was bom in Vora Hay, in 1982. He reeived the
pits degre frm the Unveay of Tino Hay 197.
“Ae id of teo years spe m FIAT Reset Cea, Toni. He
199" Hs rin Bld of teri te ivetgation of hysteresis peso
Sh and magneton proces in soft mage mate.