HAL Id: jpa-00234390
https://hal.archives-ouvertes.fr/jpa-00234390
Submitted on 1 Jan 1951
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.
Experiments on a specimen with large domains
K.H. Stewart
To cite this version:
EXPERIMENTS ON A SPECIMEN WITH LARGE
DOMAINS
By
K. H. STEWART.Sommaire. 2014 Un monocristal de ferro-silicium était
façonné de manière à faciliter la formation de grands domaines. Son cycle d’hystérèse était composé, en grande partie, d’un seul saut de Barkhausen
qu’on a étudié en détail et discuté en termes de mouvements des parois des domaines.
LE JOURNAL DE PHYSIQUE ET LE RADIUM. TOME 12, MARS
1951,
PAGE 32ö.1. Introduction. - Most
ferromagnetic
specimens
contain a very
large
number of domainsarranged
in a
complicated
three-dimensionalpattern.
Itis difficult to find
simple
lawsconnecting
themagnetic
behaviour of suchspecimens
with thebehaviour of the idealized isolated domains which
we can treat
theoretically.
There is thereforespecial
interest inexperiments
withspecimens
whose domain structure issimple.
There are twoknown ways of
obtaining really simple
structuresFig. I.
The first is the one discovered about 20 years ago and used
by
Preisach[1],
Sixtus and Tonks[2]
and
others;
this is the method where a wire of amaterial with
positive magnetostriction
isplaced
under tension so that the wire isalways magnetized
along
itslength
eitherpositively
ornegatively
and the
change
from one state to the other takesplace by
the passagealong
the wire of asingle 180°
domain wall.
By studying
the movement of this wall agood
deal of information has been obtained about theproperties
of the wall and the factorscontrolling
its movement.The second
simple
type
of domain structurewas found
recently by
Williams andShockley [3]
(1949)’
They
used asingle crystal
of silicon-ronthat was
carefully
cut into arectangular
blockwith a
rectangular
holethrough
the centre, all thefaces
being parallel
to(100) planes
of thecrystal.
It was found that the
specimen
wasnormally
divided into
only
8domains,
shown inplan
infigure
r,each domain
extending right through
thecrystal.
Changes
inmagnetization
occurredby
the movementof the central 180° wall
dividing
the domainsfor-,
ming
a clockwisemagnetic
circuit from thoseforming
an anticlockwise one. An exactcorres-pondance
was found between the movement ofthis
wall,
made visibleby
the Bitterpattern
tech-nique,
and thechanges
in flux measuredelectrically.
It is this second
experimental
arrangement
forobtaining
simple
domain structures that wasused in the work to be described.
2.
Preparation
of thespecimen. -
Thespe-cimen was a
single crystal
of 3 per 100 silicon-iron of external dimensions about I cm X0.75
cm Xo.!~
cmpierced
by
a hole o.5 cm X o.2 ~ cm. The faces were allcarefully
worked untilthey
were within onedegree
of(100) planes.
In theearly
stages
the
optical
reflections from etchpits
were used as aguide
to the orientation. A more exact test for the finalstages
wasprovided by
the Bitterfigures
formed when colloidalmagnetite
wasapplied
to the
surfaces;
if the orientation was notperfect,
" fir tree "
patterns
were formed as describedby
Williams,
Bozorth andShockley
[4]
and their direction indicated the way in which thecrystal
had to be worked in order to correct the orientation. In most of this work Bitter
patterns
were obtained afterpreparing
the surfaces in the usual waybey
electrolytic polishing,
but an alternativemethod,
which may have usefulapplications
in otherwork,
was also
developed.
In this method the surfaceswere
prepared
by
mechanicalpolishing
and theconsequent
strains,
whichnormally
prevent
obser-vation of
satisfactory
patterns,
were removedby
annealing
thespecimen.
Mostannealing
methods326
produce
a film of tarnish on thepolished
surface,
butby annealing
in a veryhigh
vacuum,provided
by
a diffusion pump, thisdifficulty
was avoidedand
well-polished,
strain-free surfaces werepro-duced. The
patterns
obtainedby
this methodwere as
sharp
and clear as those obtained afterelectrolytic polishing.
After
working
all the faces to the desiredorien-tation,
coils were wound on thespecimen
to measureits
magnetic
properties.
The field wasprovided
by
means of an 18 turn coil wounduniformly
roundthe
specimen
and fluxchanges
were measuredby
a r o turn coil on onelimb,
connected to a ballisticgalvanometer.
w3. ~
Experimental
study
of thehysteresis loop
of the
specimen. -
Thehysteresis loop
measuredin the way
just
described is shown infigure
2. TheFig. 2. - 1-H, Curve of the specimen.
most
striking
feature is thelarge instability
in fieldsof about o.o5 Oe. This can be
regarded
as asingle
Barkhausen
jump corresponding
to reversal of themagnetization
in about half the total volume ofthe
crystal.
We may assume, for purposes ofdiscussion,
that theinstability
in themagnetization
curve
corresponds
to the passage of a domain wallsuch as that shown in
figure I
from oneposition
of stable
equilibrium
toanother,
so that astudy
of the 11
jump
"will
give
information about the factorscontrolling
wall movement.Unfortunately
the Bitter
patterns
obtained on thespecimen
invarious states of
magnetization
did not reveal anydefinite domain structure; all that could be seen was a number of isolated 11 fir trees "
on
parts
ofthe surface where the orientation was not
quite
perfect.
It is therefore notpossible
to sayeactly
what
rearrangement
of domainscorresponded
to thelarge
Barkhausenjump
offigure
2,though
thesize of the
jump
makes it clear that verylarge
domains,
with volumes of the order of1/10
cc. wereinvolved. It is
possible
that,
since thisspecimen
was much thicker than that used
by
Williams andShockley,
the main domainboundary
was in aplane parallel
to the surface shown infigure
Irather than
perpendicular
to this surface as inWilliams and
Shockley’s
experiments.
The
change
inmagnetization
between A and B infigure
2 was found to takeplace
quite slowly.
It
could,
infact,
be arrested after it had startedby switching
off themagnetic
field. If the fieldwas then increased
again
thejump
would re-start at a lower field than theoriginal starting
field,
sothat the true
magnetization
curve isre-entrant,
as shownby
the full line infigure
2. ~ The re-entrantcurve could
plotted roughly by
manual control of thefield,
asjust
described,
but a better idea ofthe form of the curve could be obtained
by
using
an automatic controlapparatus
shown infigure
3.Fig. 3. - Circuit of field control device.
H1, main field coil; H2, auxiliary field coil; B, flux coil; G, galvanometer; L, lamp; P, photocell; A, d. c. amplifier.
A flux
measuring
coil on one limb of thecrystal
was connected to a
sensitive,
shortperiod
galva-nometer.
Light
reflected from thegalvanometer
mirror fell on a
photocell
which controlled the currentflowing
in aauxiliary
fieldwinding
on thespecimen.
The wholearrangement
behavedpractically linearly
so that the field
provided by
theauxiliary
coil wasproportional
to thegalvanometer
deflection and henceto the rate of
change
ofmagnetization
in thecrystal.
327
such a direction as to oppose the
change.
Thus,
if the main field was
adjusted
tobring
thecrystal
to
point
Afigure
2 and cause thejump
to start, thegalavanometer
deflected,
andby altering
theauxiliary
field,
changed
the total field in thedirections
required
to slow down orstop
thejump. Large
amplification
was used so that a small rate ofchange
of
magnetization produced
alarge
change
in theauxiliary
field and the latter was thusalways
kpet
very near to the value
required
forequilibrium
ofthe domain wall. In this way it was
possible
tofollow the " true "
magnetization
curve shownby
the full line infigure
2. With the field control device inoperation
the wall movement was veryslow,
the whole unstablechange
from A to B infigure
2taking
about 5 min.The value of the field at any instant could be
obtained from the currents
flowing
in the main andauxiliary
coils and themagnetization
could be foundby applying
a fieldlarge enough
to causesaturation in one direction or the other and
observing
the ballisticgalvanometer
throw. A continuous record of thechanges
in field andmagnetization
was obtained
by
recording
photographically
the deflection of thegalvanometer
of thefield-controlling
device. This deflection wasproportional
to theauxiliary
field and alsoapproximately proportional
to the rate of
change
ofmagnetization.
The areabetween a trace of
galvanometer
deflection and the time axis thus measured the totalchange
in magne-tization while theheight
of the trace measured the field. Threetypical
traces are shown infigure
4.
Fig. 4.
They
were all taken under the same conditions but it will be seen thatthey
are notidentical,
although
their main features are the same.4.
Interpretation
of thehysteresis
loop.
-The
irregular
traces shown infigure 4
and thecom-plete
hysteresis
loop
offigure
2, which can bederived from
them,
provide
agood
illustration of the processes ofhysteresis,
which in normal materialsare on too small a scale to be accessible to
expe-riment.
According
to theaccepted
account ofhysteresis
the movement of domain walls isopposed
by
« obstacles » which can be caused in various waysI
by irregularities
of stress or structure in the material.
,
The energy of the material can thus be
represented,
as a function of the
position
of thewall,
by
anirre-gular
curve such as that shown infigure
5. IfFig. 5.
an
increasing magnetic
field isapplied,
the wall will move from itsoriginal equilibrium position
at o
along
the dottedpath
shown in thefigure,
the horizontalportions
of thepath
representing
irreversible movement from one obstacle to thenext. Reversal of the field
will,
of course, takethe wall back
along
a differentpath
as shown infigure
5. In most materials the scale of the obs-tacles is so small that apractically
smoothhysteresis
curve is obtained but in the
single crystal specimen,
with itslarge
domains,
the scale is so muchenlarged
that we can observe the
irregular
motion of the walls in some detail.It is not
suggested
that the traces infigure 4
show the full detail of the obstacles the wall hasto overcome. The limitations of the
apparatus,
particularly
the inertia of thegalvanometer,
wouldprevent
anyirregularities
of smallwavelength
frombeing
observed. Theirregular
traces shownare, in
fact,
only
smoothed versions of the still morecomplicated
curves that would be obtained in anideal
experiment.
The fact that the traces obtainedare not
quite reproducible,
aftertaking
moderateprecautions
to ensureconstancy
ofconditions,
suggests,
however,
that the smaller obstacles arevariable,
presumably
becausethey
are verysensi-tive to local stress or
temperature
fluctuations,
and hence that an
experiment
withhigher
«resolving
power » would
give
further informationonly
in statistical form.The energy lost in a
hysteresis cycle
is,
according
to the scheme of
figure
5,
the energysupplied by
the field
during
the various irreversible movementsof the wall and is measured
by
the area betweenthe full curve,
corresponding
toequilibrium positions
328
energy lost in the
jump
from obstacle A to obstacle B is shownby shading
infigure
5. For acomplete
cycle
the energy lostis,
of course,represented
by
the area within the
complete
hysteresis loop.
If itwere
possible
toadjust
the fieldquickly enough
to maintain the wall inequilibrium
at alltimes,
i.e. to follow
exactly
the full curve offigure
5,
thearea of the
hysteresis loop
would shrink to zeroand the total energy
supplied by
the field inperfor-ming
acycle
would also be zero. In theexperiments
with the
single crystal
this process wasactually
carried out to a limited extent. The
hysteresis
loss of the
crystal
measured in the usual way isrepresented by
the area of the outerloop
offigure
2,but when the
cycle
wasperformed
with thefield-control device in
operation
thehysteresis
loss,
the energy transferred from field to
specimen,
was reduced
by
about 12 per 100 because there-entrant curve of
figure
2 was followed. We maysuppose that if an
apparatus
with aquicker
responsehad been
used,
a moredeeply
indented curve would heve been followed and thehysteresis
loss still further reduced. It doesnot,
however,
appear verylikely
from the curvesalready
obtained thatthe area of the
loop
couldactually
be reduced tozero
by
this process.Two comments may be made on this. In the
first
place
the scheme shown infigure
5implies
that for every
position
of the wall there isonly
one value of the
magnetic
field that cangive
equi-librium ;
this is true for most of the mechanisms which aresupposed
toproduce
obstacles to wallmovement so
long
as mechanicalhysteresis
effects in the obstacles areignored.
If, however,
theseare taken into account then the curve
showing
equilibrium
positions
will not be asingle
line asin
figure
5,
but will itself be a closedhysteresis
loop
whose area willrepresent
a minimum below which themagnetic hysteresis
loss canot be reduced no matter howcarefully
the field isadjusted.
Itis
possible
that the re-entrant curve offigure
2 isan
approximate
delineation of this minimumloop.
A second
possibility
is that the obstaclechiefly
responsible
for thehysteresis
loss,
of thetypes
consideredby
Becker, Kersten,
Néel andothers,
are on a very much smaller scale that those revealed infigure
2. In this casefigure
5 should be drawnas a curve
oscillating
in random manner with avery short
wavelength
and the observedhysteresis
loop
would be theenvelope
of this curve. It would then be true that if therapidly oscillating
curve could be followed
exactly,
thehysteresis
loss would become zero, but it would bequite
beyond
the power of thepresent apparatus
tofollow these small-scale oscillations. The
irregu-larities that were
actually
observedmight
then beattributed to some other
hysteresis
mechanismcapable
ofproducing
a fewrelatively large-scale
obstacles. One
simple
type
of obstacle thatmight
account for the main feature of the
hysteresis
loop
infigure
2, thebending
back below the rema-nencepoint,
is thedifficulty
ofintroducing
a 180°wall at all. If an ideal
crystal
of the form shownin
figure
I were firstmagnetized
to saturation in onedirection,
a veryhigh
coercive field would have to beapplied
before a 180° wallrunning
round thecrystal
could appear. It is wellknown,
of course,that the coercive field
actually
needed in a realcrystal
isconsiderably
reduced because walls cangrow out from
regions
where there areslight
defectsin the
crystal.
Even so, a finite field wouldcer-tainly
be needed for the introduction of a new walland in the
particular single crystal
under conside-ration this field may well have the value of about 0.02 Oerequired
toexplain
the «hump
»at A in
figure
2.5. The
speed
of wall movement. 2013 Theinter-pretation
ofhysteresis
effects in terms of obstaclesto wall movement leaves
unexplained
the actual mechanismby
which thehysteresis
energy isdissi-pated
in the material. It is shown that as a wallmoves, energy is stored up in various
f orms
untilan unstable
position
is reached and the wall movesforward to a new stable
position
with the release of acertain amount of energy. The
speed
with which the wall moves will be determinedby
the mechanism whichdissipates
this energy. If no other mechanism wereavailable,
the wall would move veryquickly
and the energy would be emitted as
electromagnetic
radiation,
but inpractice
other mechanisms arepresent
which restrict thespeed
of wall movement to much lower values and absorb thegreater part
of thehysteresis
energy. We mayhope
to findout the chief
energy-absorbing
mechanism in anyparticular
caseby
astudy
of thespeed
of wallmovement. Thus Sixtus and Tonks’ classic expe-riments on the
speed
of 180° walls in nickel-iron wires can beinterpreted,
as Snoek hasshown,
asindicating
that the inducededdy
currents control thespeed
and convert the energy into heat inthick wires but that in thin ones this effect is
less
important
and someextra,
less wellunderstood,
«
frictipnal
)) mechanism must be assumed to beacting.
The
velocity
of wall movement was measured in thesingle crystal
specimen by recording
photo-graphically
thereadings
of aheavily damped
galvanometer
connected to the flux coil on thespecimen.
Measurements were made at thebegin-ning
of the movementonly,
and in these conditions thegalvanometer
acted as an almostperfect
flux-meter. The wall
velocity
was measured as afunc-tion of
applied
fieldby
taking
thecrystal
to apoint
just
short of A infigure
2, and thensuddenly
applying
alarger
field. Atypical
fluxmeter recordis shown in
figure
6. The initial rate ofchange
329 curve, as
shown,
and the wallvelocity
was deducedfrom this on the
assumption
that the fluxchange
was due to the movement of asingle
domain wallFig. 6.
of the form shown in
figure I .
The results obtained for various values of theapplied
field aresshown
in
figure
7. Within the limits ofexperimental
Flgn 7.
error there is a linear relation between
velocity
and field which can be written
V= A(H-Ho)
where A and
Ho are
constants with the values6.3 cm : s : Oe and
o.o4g
Oerespectively.
The lower line infigure 7
was obtainedby repeating
the
experiment
with thespecimen
immersed inliquid
oxygen(goo K);
the value of A andHo
werethen
5.7
cm : s : Oe and o.o56 Oe.The linear V - H relation
implies
that the forcecontrolling
thespeed
of wall movement is oneincreasing linearly
with wallvelocity.
The velo-cities in thesingle crystal
are much lower thanthos that must occur in normal
polycrystalline
material. These latter velocities have not been observed
directly
but it can be inferred frommagnetic
viscosity experiments
thatthey
are at least several hundred centimetres a second. Of the variousmechanisms that have been
suggested
ascontrolling
wall velocities the
only
one that would beexpected
to
give
very different velocities inpolycrystalline
iron and in the
present
single crystal specimen
is the controlby eddy
currents.Eddy
current effectsdepend strongly
on dimensions and should be muchgreater
for thelarge
domains of thesingle crystal
than for the small ones inpolycrystals.
It is difficult to calculate the exact effects of
eddy
currents on the movement of 180~ a wall of the form shown in
figure
I, but an order ofmagnitude
estimate can be madeby
assuming
the wall to mave as arigid plane
andignoring edge
effects. Wethen have the situation shown in
figure
8,
the wallFig. 8.
and the
crystal being supposed
to extendindefi-nitely
in a vertical direction. Movement of the wall will induceeddy
currents as indicatedby
thearrows and if we suppose that the
speed
of the wall is determinedby
the condition that the field of theseeddy
currents should beequal
andopposite
to the
applied
field,
we obtain V = AH with .Awhere p
is theresistivity
of thematerial,
Is
its saturationmagnetization
and a the thickness of thespecimen
perpendicular
to the wall. Theappropriate
values for thesingle
crystal specimen
are p = 5o x 10-6 Q cm,I5 =
160gauss, a =
o.2Q
cm so that we obtain A = 1.6 cm : s : Oe.Although
this is of the same order as theexperimental
value A = 6.3 cm : s:
Oe,
it is several times smaller.The difference can
probably
be attributed to the crudepicture
of the motion of the wall used in thecalculation. In
practice
theeddy
current field will be much less than the calculated value nearthe
edges
of thespecimen
and the wall will thusadvance more
rapidly
near theedges
and sogive
agreater
meanspeed.
The
temperature
dependence
ofvelocity provides
furthersupport
for the idea that thevelocity
is330
effects
depending
on mechanical relaxation of strains. The latter would bestrongly
temperature-dependent
whereas
eddy
current effectsdepend
chiefly
on theresistivity,
which,
insilicon-iron,
does not verygreatly
withtemperature.
The measured decrease inresistivity
of astrip sample
of the same materialon
cooling
from roomtemperature
togo~
K was18 per oo, while the observed decrease in the
velocity
constant A for thesingle crystal specimen
was 1 o per 100. In view of the rather
large
experi-mental errors invelocity
measurement,
these valuesare in fair
agreement
with each other.6.
Energy
losses inalternating
fields.--We may conclude from the evidence of the
previous
section that the
speed
of the main wall movementin the
single crystal specimen
was controlledby
the inducededdy
currents and hence deduce thatthese currents were the means of
converting
intoheat an
appreciable
fraction of the wholehysteresis
loss of thespecimen-certainly
the fraction repre-sentedby
the « unstable » area between the fulland dotted curves of
figure
2 andprobably
alsoa considerable additional area
corresponding
toinstabilities too small to be revealed
by
thepresent
experimental technique.
If it isaccepted
that atleast a
part
of thehysteresis
energy of a material isdisposed
ofby eddy
currents inducedby
domain wall movement, it isinteresting
toenquire
how the energy losses observed inalternating
fields are to beinterpreted.
It is usual to divide these losses into threeindependent
components,
hysteresis
loss,
eddy
current loss andviscosity
loss,
but ifhysteresis
loss is to be attributedpartly
toeddy
currents and if theviscosity
effect is alsosupposed
to be dueto
eddy
current control of wall movement, it is nolonger
obvious that there are threeseparate
compo-nents of loss which can be added
together.
A
satisfactory interpretation
is, however,
stillpossible
in a materialcontaining
many domains. When such a material is taken round itshysteresis
loop slowly
we may suppose that each domain wallpasses
through
its unstableregion
at a differentinstant. As it does so, it will set up, over a limited
region
of space andtime,
asystem
ofeddy
currentswhich
dissipates
thehysteresis
energy of the indi-vidual domain. The total energy sodissipated
may be written
where
wH
is thehysteresis
loss of thematerial, p
itsresistivity,
In
theeddy
current inducedby
the nthdomain in the volume d V at time t.
If the
cycle
isperformed
morequickly
there aretwo differences. In the first
place
if theapplied
fieldchanges appreciably
during
the movementof any domain
wall,
thespeed
of that wall will beaffected and
consequently
theeddy
currents it induces willchange.
Each of theterms,
I;, J2 "
...,I ~
may therfore bechanged by performing
thehysteresis
cycle
morerapidly
and theconsequent
increase in
WH
can be identified with the «visco-sity
loss )). In most materials the movements ofwalls are so
quick
that this effectonly
becomesappreciable
at radiofrequencies.
The second effect of
rapidly varying
fields is toreduce the interval between the wall movements
of
separate
domains until actualoverlapping
occursand the
eddy
current at anypoint
is the sum of contributions from more than one domain. Becausethe energy converted into heat
depends
on thesquare of the current
density,
thisoverlapping
willcause an increase in the total energy lost. We can
write.
where the domains I and
2, l~
and I,
aresupposed
to pass
through
these unstableregions
practically
simultaneously,
while other domains remain dis-tinct. The second term,We,
due to« overlapping »
of
eddy
currents can be identified with theeddy
current loss calculated
according
toordinary
macros-copic electromagnetic theory,
as can be verifiedby
detailed calculation for
particular
models.It thus appears
justifiable,
in material with manydomains,
to treathysteresis, large
scaleeddy
currentsand
viscosity
asproducing
there distinctcomponents
of loss which can be added togive
the total loss inan
alternating
magnetic
field,
despite
the fact thatall three
components
mayfundamentally
be due toeddy
currents.(Other
causes ofhysteresis
andviscosity
may, of course, bepresent
in additionto the
eddy currents.).
In material with small numbers ofdomains, however,
such as thesingle
crystal
specimen
describedabove,
the distinction between the «microscopic » eddy
currentsrespon-sible for the
hysteresis
component
of loss and the«
macroscopic »
onesresponsible
for the normaleddy
currentcomponent
can nolonger
be made andspecial
calculations would have to be madeto
predict
the energy loss in analternating
field.There are
clearly
many ways in which theexpe-riments that have been described could be
improved
and extended.They
were not intended as muchmore than a
preliminary investigation,
made in a rather limitedtime,
of thepossibilities
ofexperi-ments ou
specimens
of thistype.
At the end of theexperiments
thecrystal
was annealedagain
inan
attempt
toimprove
itspropreties.
331
number of small
crystals
within the main one andthe unusual
properties
of theoriginal specimen
were lost.
7.
Achnowledgements. -
The author would like to thank Dr H. Lawton and Dr D.Shoenberg
for their advice and assistance and the British Elec-trical and Allied Industries Research
Association,
which
sponsored
the programme of research.Remarque
de M.Shockley.
- Dr Stewart hasproposed
that if themagnetic
field wereproperly
adjusted
during
changes
inmagnetization,
themagnetization
processmight
becomeentirely
rever-sible.
Although
such a situationmight
prevail
if the field were
ajusted locally
within thespecimen.
For the
specimens
of thetype
measured atCam-bridge
and at BellTelephone
Laboratories there willcertainly
be irreversible energy losses due tothe fracture of the Néel
spikes.
The studies madeby
Williams are believed torepresent
veryclosely
the
limiting
behaviour that would be obtained forvanishing
small rates ofchange,
and these studies show two forms of energy losses : onehaving
amemory and associated with the Néel
spikes
and the otherbehaving
like adynamic
friction.Dr Stewart has also discussed the
relationship
between domain behaviour andeddy
current losses. It may be worthwile topoint
out a conclusion reachedby
Dr Kittel and thespeaker
based on aconsideration of the
collapsing cylinder
model :If the domain wall is
just
inside the surface of thespecimen,
then theeddy
current losses arenegli-gible
for agiven
value of on the other hand ifdt
the radius of the
collapsing cylinder
is madearbi-trarly
small,
the energy loss increases without limit for agiven
value off .
Dr Kittel hasrecently
dt
carried out an
analysis covering
more realistic cases which he uses indiscussing
anomalouseddy
current losses and related
problems.
Remarque
de M. Snoek. - I should like topoint
out that the mechanism
by
which energy istrans-mitted from one
spin
to another in the course of arotation process
involving
adisplacement
of aBloch
boundary
has notyet
been discussedtheore-tically.
Itobviously
has abearing
too on thephenomenon
offerromagnetic
resonance and thedamping
observed therein.The
experiments
by
Stewart onboundary
dis- .placements
in asingle crystal
should be discussed in connection with theexperiments
by
Dijkstra
et Sncek on
boundary
shift inelastically
strained wires of nickel iron.Remarque
de M. Roberts. - If there ismagneto-striction,
then there will be another mechanismenabling
the directdissipation
ofmagnetic
energy of thespins
in the domain wall in the form of thermal lattice vibration i.e. without the intervention ofeddy
currents.REFERENCES.
[1] PREISACH F. - Physik. Z., 1932, 33
905.
[2] SIXTUS K. and TONKS L. 2014 Phys. Rev., 1933, 43, 931.
[3] WILLIAMS H. J. and SHOCKLEY W. - Phys. Rev., 1949,
75, 178.
[4] WILLIAMS H. J., BOZORTH R. M. and SHOCKLEY W. 2014