HAL Id: jpa-00234402
https://hal.archives-ouvertes.fr/jpa-00234402
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.
Magnetic saturation intensity and some other related
measurements
W. Sucksmith
To cite this version:
MAGNETIC SATURATION INTENSITY AND SOME OTHER RELATED MEASUREMENTS
By
W. SUCKSMITH. Sommaire. - Dans lapremière Partie, l’auteur attire l’attention sur le fait que des mesures
exactes de l’aimantation spontanée n’ont été effectuées que dans un très petit nombre de cas. Les résultats pour le cobalt, qui possède un réseau hexagonal jusqu’au point de Curie et un réseau cubique centré au-dessus, sont donnés et discutés. Une augmentation de 1,5 pour 100 se produit à la tempé-rature de transition.
Dans la deuxième Partie, la variation de l’aimantation à saturation et le degré de l’ordre dans
quelques alliages binaires sont donnés. Dans la dernière Partie, on montre l’importance des mesures
de l’aimantation à saturation pour étudier la détermination de l’équilibre des phases. L’auteur cite
l’application heureuse de cette méthode au cas du système fer-carbone, caractérisé par quelques phases qui ne figurent pas à l’état d’équilibre.
LE JOURNAL DE PHYSIQUE ET LE RADIUM. 12,
1~~~?
I. Saturation
Intensity
Measurements andthe Iaaur of
Corresponding
States. - Of the threeferromagnetic
elements,
the variation of thespontaneous
magnetization
withtemperature,
hasbeen most
thoroughly
investigated
in the case ofnickel,
the most exhaustive and accuratemeasu-rements
having
been carried outby
Weiss and Forrer[1].
Foriron,
the lowertemperatures
werecovered
by
the work of the same two authors[21
whilst theinvestigation by
Potter[3]
deals with the range from roomtemperature
to the Curiepoint.
Cobalt,
which wasinvestigated by
Bloch[4]
presents
additionaldifficulties,
chiefly
on accountof the
large
crystalline
anisotropy.
Bloch’smea-surements in fields up to about i3ooo Oe have been hitherto
regarded
asrepresenting
thetempe-rature variation of
spontaneous
magnetisation,
a conclusion which must beregarded
as open toconsiderable criticism on the
following
grounds.
In the first
place
theabnormally high crystalline
anisotropy
makes theapproach
to saturation of thepolycrystalline
material very slow. This rate ofapproach
decreases as thetemperature
is loweredowing
to the increaseddifficulty
ofattaining
satu-ration in the basal
plane,
i.e. at -i goo C only
80 per 100 saturation is reached in an effective field of io ooo Oe. In addition there is the
diffi-culty
that atordinary temperatures polycrystalline
cobalt exists as a mixture of face-centred cubic and close
packed hexagonal
structures, thepro-portions
of which varyconsiderably according
tothe method of
preparation
and issubsequent
thermal and mechanical treatment and is notfully
control-lable.
Further,
there is aphase
transitionbeginning
at about
’700
C,
above which the metal iswholly
face-centred cubic.
Lastly,
it is notimpossible
that the wide variations in the recorded
density
of cobalt exercise some
effect,
ifonly
inlimi-ting
accuracy of measurement. Theonly
method ofsecuring
moreprecise
data on the variation ofspontaneous
magnetisation
withtemperature
appears to be to usesingle crystals
of themetal,
and to make measurementsalong
the easydirec-tion of
magnetisation
up to the transitiontempe-rature. Above the transition range, the
wholly
face-centred metal hasanisotropy
constants similarto the other cubic
ferromagnetics.
Myers
and Sucksmith havejust
completed
thisinvestigation
with thefollowing
results.By
slowcooling
of the molten metal(99.9
per Ioopurity)
through
themelting point
and the solid metalthus
produced through
the transition range,single
crystals
ofhexagonal
cobalt wereproduced.
The variation of the saturationintensity along
its easy direction ofmagnetisation
shows the usual trend until atemperature
of4ooo
C is reached when thephase
change
takesplace. Up
to thistemperature
quantitative comparison
of the results with otherferromagnetic
materials is notpossible,
other thanon the somewhat
specious assumption
of a Curietemperature
for thehexagonal
material. On theother
hand,
because of the closeagreement
inpart
of this range between our results and those of Guillaud(see below),
it is not without interest toextrapolate
such values as we have. The reducedmagnetisation
decreases fromunity
to a valueof
o.942
at6730
abs. If weextrapolate
thesevalues
according to j
=2 ,
the Curietemperature
2
is about 1150
abs.,
considerably
lower than theCurie
temperature
for the cubic cobalt of1404°
abs.Assuming
a behaviour identical with that of nickel431
(see
below),
leads to a Curietemperature
atabout
1420°
abs. Whilst this closeagreement
is
wholly
fortuitous,
it can be taken topoint
tothe conclusion that the curve for
hexagonal
cobalt lies belowthe j
curve
and close to that for the face centred nickel.At the transition
temperature
themagnetisation
rises
discontinuously
at the transitionby
about1,5
per ioo(see
f g.
I ),
andthereupon
Fig. I.
follows a trend up to the Curie
temperature
identicalwith that of nickel when the reduced
magneti-sation
G
isplotted
against
the reducedtempe-0
rature ( n )’
This means that in this uppertempe-0
rature
region
the values aregreater
than for the theoretical curvefor j
= §.
Subsequent
heating
andcooling through
the transition zone for both thesingle
crystal specimens
and thepolycrystalline
material showed similar behaviour of a rather remarkable nature. With this thermal treatmentthe
intensity
both above and below the transition increasedby
small amounts which decreasedpro-gressively
until the cobalt was 11 stabilized " after about 10repetitions
ofthe
thermalcycle.
Thetotal increase amounts to a
change
at 3000(i.
e. below the transitionregion)
of about u =0.4.
Theonly possible
explanation
of this minoranomaly
appears to be topostulate
11lattice mistakes "
in the structure.
The value of 6 for the
hexagonal phase
is foundto be 162.5 with pn =
1.72, and a Curie
tempe-rature of i 13 1
± 3° C,
thedensity
at roomtempe-rature
being
p =8.84!t
± o.oo5 in all cases. Theseresults are in fair
agreement
with the recentmeasu-rements on a
single crystal
of cobalt of Guillaud and Roux[5],
who obtained from measurements at and below roomtemperature
c -=16o.9
with1.70. Their value for the ratio of the
intensity
at roomtemperature
to that at absolute zero is about 1.005 and compares well with ours ofThe best value for 7 for the cubic cobalt we estimate
to be
16~1.5
(p~ _ ~
thisbeing
obtainedby
extrapolation
from above the transition range andconsequently higher
than values obtainedby
others.Comparison
of all theferromagnetic
elementswith
simple theory gives
the best accordfor j
== ~
7with minor deviations in
general
common to allthreee
elements,
in that the measuredspontaneous
magnetisation
liesslightly
below the theoreticalT
curves for small values of
(j’
, but attemperatures
approaching
the Curietemperature,
experimental
results exceed those demandedby theory.
If one considers the
application
toalloys,
inparticular
intermetalliccompounds,
the work of Guillaud[6]
has addedconsiderably
to theexpe-rimental data available. For the three
compounds
for which data are available up to the Curie
tem-perature,
two ofthese,
MnP andCr02’ approximate
moreclosely to j
= ~
withdepartures
of the sametype
as thoseexperienced
in the pure metals. On the other hand. Cr - Teapproximates
closely
to the
curve j
= 3.The inherent
difficulty
in the utilization of all these measurements of saturationintensity
lies in the determination of thespontaneous
magnetisation.
At lowtemperatures,
the law ofapproach
is ofsecondary
importance
andextrapolation
to infinite fieldgives
anadequate
method in which the errorcannot be
large.
Forhigher
temperatures,
however,
where the
change
oaf " saturation "magnetisation
withtemperature
isgreater,
and theapproach
tosaturation is more
gradual, extrapolation
becomesincreasingly
difficult and uncertain. Thebest,
method for
determining
thespontaneous
magne-tisation comes from themagnetocaloric
effect,
the rise of
temperature
onmagnetisation
isdirectly
related to the variation of the square of the spon-taneousmagnetisation
withtemperature,
so thatT
only
for valuesof 6
above abouto.6,
where there is asufflciently
great change
of cr withT,
is it pos-sible to obtain measurable results.Weiss,
however,
has made use of curves
giving
the field as a function oftemperature
for constant values of themagne-tisation. These curves are linear for the
higher
values of the
field,
andextrapolation
to zero fieldgives
thetemperature
whichcorresponds
to theparticular
value of 6employed.
The two methods have been found to a gree for metal over the limited range oftemperature
where both measurementsare
possible,
but the accuracy is far fromadequate
to enable
satisfactory comparison
withtheory.
There is therefore a need for considerable extension of
magnetocaloric
measurements onferromagnetics.
Accurate measurement of the small
temperature
which accurate
knowledge
of the values of thespontaneous
magnetisation
to beconsiderably,
increased. Asyet,
noattempts
have been made toextend the measurements to any of the
important
alloys
which are thesubject
ofinvestigation
today.
In the absence of thesedata,
one iscompelled
toassume that the saturation
intensity
is anadequate
measure of the
spontaneous
magnetisation,
thisassumption becoming increasingly
hazardous with the slowerapproach
to saturation characteristic ofnearly
allferromagnetic alloys.
Forexample,
the reduced curves
G,
T)
change shape
markedly
in solid solutions of aferromagnetic
and anon-ferromagnetic
element. It is notyet
definitely
known whether thespontaneous magnetisation
curve is of the same
shape,
and more extensivemeasurements of the
magnetocaloric
effect as well ascomprehensive
determinations of.I~
curves over the wholetemperature
range arerequi-red. The writer is
initiating
experimental
work in thehope
ofextending
the range of information of this nature.II. Saturation
Intensity
Measurements and Order-Disorder. - Thewriter,
ininvestigations
on the Fe-Ni-Al
alloy
system,
haspreviously
noted differences in the variation of saturationinten-sity
of ordered and disorderedalloys.
At least three diff erentexamples
arepresent
in thissystem,
two of which have been the
subject
of furtherinves-tigations
and will be dealt with below. Thethird,
which is the
prototype
of thepermanent magnet
precipitation alloys,
showed saturationintensity
data which fit in well with the nowgenerally
accep-ted view of the processes involved. The(g, T)
curve for the annealedalloy (two phase),
showsbehaviour characteristic of the
single phase type
with a Curietemperature
approaching
closely
to that of pure iron and a saturation
intensity
of about 105units,
i. e.slightly
less than half thatof
iron,
thusjustifying
the view that the twophases
are Fe and thenonmagnetic
Ni Al. Thequenched
material, however,
has ahigher
saturationintensity
than the annealedalloy,
and aftertending
to a lower Curietemperature
than pure iron up to4ooo C,
gradually
shows the break up of thesingle phase by
an arrest in the fall of themagnetisation,
andfinally
at about 6ooO C reaches theequilibrium
intensity
of the twophase alloy.
Of other work in this
field,
probably
the mostimportant
is that on the 5o per 10oFe,
5o per I oo Cosuperlattice.
Ellis and Greiner[7]
had shown evidence of the existence of thesuperlattice by
X-rays, supported
by
the additional evidence of the thermal arrest method.Later,
Goldman and Smoluchowski[8]
indealing
withexperimental
and theoretical data on saturationmagnetostriction
and
order,
show that themagnetostriction
atsatu-ration increases
by
4o
per 10o uponordering.
In alater
contribution,
the theoreticalprediction
thatthe saturation moment should be about
4
per 100higher
than in the randomalloy
has been confirmed. Onecontributory
factorleading
to thehigh
permea-bility
of thisalloy (known
asPermendur)
may bedue to the establishment of order
by appropriate
heat treatment. Somesupport
of this view isgiven by
work on theFe,Al
superlattice by
Ben-nett[9].
The(6, T)
curve for thisalloy,
both in thequenched
and annealed states and for othersnear this
composition,
show the usualtrend,
but theintensities at low
temperatures
show notable features. Whilst theintensity
of the orderedalloys
falls offsmoothly
with increased aluminium content at agradually
increasing
rate,
the disorderedalloy
is moremagnetic
up to about 25 per i ooAl,
butdrops
sharply
to a lower value than the ordered forhigher
aluminium content. This effectpersists
up toabout 3 o per 100. The reason for this is
presumably
due to the formation ofnon-magnetic
Fe Al. In addition to this there is aslight
discontinuity
in theintensity-temperature
curve in theneighbour-hood of the
ordering
temperature, though
the Curietemperatures
of bothquenched
and annealedspecimens
do not differappreciably.
The effecton the low field
properties
at roomtemperature
wasinvestigated,
but the increase inpermeability
due toordering,
e. g. from about 2 ooo to 10 000,was not marked. The
coercivity, irrespective
of thedegree
of order was less than i Oe in all cases. On the otherhand,
measurements of the(~, H)
curve athigher
temperatures
showquite
differentresults,
theregion
ofpartial
ordershowing
themost marked effects. Here the
equilibrium
per-meability
fell from thehigh
values of the order of some thousands to a value of about4oo
just
below 5ooo C
rising again
withincreasing
tempe-rature. The
coercivity
reached a maximum of about 20 Oe in the sameregion.
The difference in behaviour betweenhigh
and lowtemperature
measurements can be attributed to the time
required
for the establishement of order. A tentativesuggestion
for thehigher
coercivities in the transitionrange
might
be made on the lines of theexisting
theories,
by postulating
smallregions
of order in a metrix of disorderedmaterial,
with theprobable
accompaniment
of strains in theboundary regions
between order and disorder.
An
investigation
into themagnetic
properties
of the Fe-Cr
system
has been madeby
Ardron[10].
Around thecomposition Fe3Cr
there issatisfactory
evidence of the existence of the
superlattice
both frommagnetic
saturationintensity
and resistancemeasurements.
Alloys
containing
20 and 27 per 100chromium have low
temperature
intensities upto
2,5
per 10o more for the ordered state than for433
to a Curie
temperature
of about8oo-gooo
C,
the Curietemperature being
about65~°
C for the 25 per 100alloy.
The break down of order isaround
600° C,
so that the orderedalloy
does notreach its true Curie
temperature.
Thepermea-bilities do not accord with the views
expressed
above,
the orderedalloys showing
very lowpermea-bilities,
but it has beensatisfactorily
shown that this is due to theprecipitation
of thenon-magnetic
sigma phase
which isresponsible
for thehigh
coer-civity
with theaccompanying
lowpermeability.
The writer haspreviously
drawn attention to themagnetic
evidence for the existence of the super-latticeNi3Fe [11].
The smalldependence
in theato-mic numbers of the two constituents makes
X-ray
evidencedifficult
but themagnetic
saturationintensity
measurements show a similar trend to thosealready
referred to in the case ofFe3Cr.
Fig. 2. - Variation of (c, T)
for ~5 atomic per 100 Ni, after cooling at 23~ : h.
Wakelin
(in progress)
hasrecently
extended theseobservations,
of which the evidence forthe
Ni3Fe
superlattice
is shewn infigure.
The spe-cimens were cooled from about 600~ C to4000 C
at 0.23° per hour to
produce
order,
after which morerapid cooling
to roomtemperature
was pos-sible.Magnetic
measurements were then taken from lowtemperatures
upwards.
Above 5000 C the order isdestroyed
toorapidly
for thehigh
Curietemperature
of thesuperlattice
to bereached,
and the curve falls
rapidly
to meet the lower curve for the disorderedalloy.
Similar measurementshave been carried out on a range of
alloys containing
between45
and 85 atomic per 10oo iron. At boththese limits no difference in
magnetic
behaviour forcooling
rates betweenquenching
and the above slow rate weredetected,
but at 55 and 80 per I o0quite
measurable differences of thetype
shownin
figure
2 wereobserved,
thusindicating
someordering.
Resistancetemperature
measurements of the well-knownshape
confirmed themagnetic
observations,
butonly
for the 25 per i oo Ironalloy
was it
possible
to show the existence of thesuper-lattice
by
directX-ray
measurements. The maxi-mum increase of v onordering
was about4
units in I 15 forNi3Fe,
thisdecreasing
on both sides of thiscomposition.
Theextrapolated
Curietemperature
fort his orderedalloy
isevidently
about 2oo-3ooOhigher
than that for the disordered material. On the basis of such evidence as is available atpresent,
one may conclude that the orderedbinary
alloys
have ingeneral,
intensities around per 100greater,
with Curietemperatures
up to 200-300° Chigher
than for the disorderedalloys.
III. Saturation
Intensity
Measurements andMagnetic
PhaseAnalysis.
- At a conferenceat the Institute of
Physics
in theUniversity
ofStrasbourg
in1939,
the writerpresented
a paperdealing
with anapparatus
for the convenientmeasu-rement of saturation intensities at all
temperatures,
irrespective
of theshape
of thespecimen, together
with results which had been obtained on a suitableternary
alloy
system,
Muchinfor-mation can be derived from
analyses
of thistype.
Fig. 3.
A recent and useful summary of the measurements
and the information to be derived has been made
by Guillaud
[12].
Inparticular,
he has made astudy
of thealloys
of manganese witharsenic,
bismuth,
antimony
and silicon. Twoparameters,
the variation of saturation
intensity,
and thetem-perature
at whichferromagnetism
disappears,
can be utilized toprovide
information. In a solid solution both saturationintensity
and Curietempe-rature
usually
decrease with addition of anon-ferromagnetic
diluent,
initially linearly,
but withincreasing proportions
at a morerapid
rate. At atemperature appearing
if the secondphase
isferro-magnetic.
The saturation intensities at anytem-perature
areproportional
to theproducts
of theintensity
per unit mass and theintensity
at thattemperature,
thusproviding
quantitative
infor-mation.A detailed
example
of thisprocedure
has beengiven
in themagnetic analysis
of the Iron-SiliconSystem by Guggenheimer,
Heitler and Hoselitz[13].
In the
non-equilibrium
condition,
no less than fourphases,
threemagnetic
and onenon-magnetic,
have
been’satisfactorily
isolated(see f g. 3).
In
addition,
however, it has been foundpossible
to follow out rates of reaction of
physical changes
bv means of continuous observations of the
intensity
changes.
By
suitable heat treatment,i. e.
quenching
fromappropriate
temperatures,
the
approach
toequilibrium
can be followedby
means of themagnetic
observations.Fig. 4.
One
example
isgiven
infigure 11,
which showsthe effect of
heating
at 65oo C analloy
which wasquenched
fromI 100° C,
afterbeing
maintainedat that
temperature
a sufficient time toproduce
equilibrium.
Theproportions
of thephases
«" and E are each about o.5(the
Ephase
isnon-magnetic
and the amount
present
determinedby
difference. Continuous measurements of theintensity
show the break down of the allphase
into « and n.Pro-longed heating
over aperiod
of weeks showed thegradual disappearance
of the nphase (which
isonly
stable between about 8500 C and io5oOC),
with theaccompanying
increase of a and s. ’The wholeof the
ferromagnetics regions
of thisalloy
system
. has been examined in thisway. The
accompanying
table
gives typical
examples
of theweight
propor-tions at different
stages
in the reaction. Thecalculated amount of Si is shown in the last column
and agrees well with the chemical
analysis.
The most recent
development
in this field hasbeen carried out
by
Crangle
and Sucksmith(in
progress)
in aninvestigation
into pure Iron-Carbonalloys.
Here aninterpretation
ofequilibrium
condi-tionspresents
littledifficulty .
For low carboncontents, up to about 1.2 per 100
by weight,
thecomponents
are ferrite(a
saturated solution ofcarbon in
body
centred cubic iron which contains~ o.oo5
per Ioo C at room
temperature
and is thereforemagnetically indistinguishable
from pureiron),
and cementiteFe,C,
for which a- = ~ 66 with a Curietemperature
of about 215° C. The dotted curve infigure
showsequilibrium
conditions and the amounts of cementite and ferrite can bedeter-mined from the
intensity-temperature
magneti-sation curve, the results
agreeing
well with the known constitution of thealloys.
The examination ofquenched
material showsclearly
andquanti-tatively
the three knownstages
in thetempering
process.
Measurements of the first
change
up to about 1 50°Cgives (Ci,
T)
curves of thetype
marked ~. infigure 5
Fig. 5.
a, decomposition of martensite; b, extrapolation of reversible
part of graph; c, curve obtained on cooling from This graph is steeper than the equivalent diluted iron
graph, showing that the precipitate is ferromagnetic; d, start of the austenite transformation; e, transformation at constant temperature of non-magnetic retained
aus-tenite ; f , that there is here no marked inflection of the
curve shows that the precipitate is mainly something other than cementite; g, decomposition of the first precipitate;
h, region where there appears to be a precipitated mixture of Fe.,C and Fe20 C9. The approach of the " quenched "
and " annealed "
graphs depends on temperature but not
sharply on time.
which is known to be due to the breakdown of the
martensite solid solution. The identification of the
breakdown as distinct from a Curie
temperature
is indicatedby
theirreversibility,
whilst thepro-duction of a
ferromagnetic component
with a lowerCurie
point
than iron is shownby
thecooling
curve c.being
steeper
than for iron mixed with anon-ferromagnetic
material.Subsequent heating gives
reversible valuesalong
c. The Curietemperature
of the new constituent is not less than about 3ooo C,
and it is
possible
that it may be identified withthe s-carbide
Fe,C,
whichthough
notyet
prepared
435
has a Curie
point
of 380~ C. It is alsonoteworthy
that Jack has obtained
X-ray
reflections of this carbide from ahigh
carbon steeltempered
at I20~ C.The second
stage
shows the transformation of thenon-magnetic
austenite(face-centred
cubiciron)
into a
magnetic
form thusgiving
the increaseindi-cated at e. After this transformation is
completed
the
(c, T)
curve is reversiblealong f .
This issteeper
than that measured before the commencement
of the austenite breakdown and contains an inflexion
in the
slope
at about 2 150C,
i. e. at the Curietempe-rature of cementite. The
implication
is that some cementite isproduced, though
whether this is due to the austenite or the lowertemperature
martensite transformation is notyet
clear.The third
tempering
stage
is characterisedby
asharp
fall inmagnetisation,
and becomesfairly
rapid
above 26oo C. Thechange
isagain
irrever-sible and ispresumably
due to the breakdown of a carbide to aphase,
themagnetic
intensity
of which is low at this
temperature.
At 3ooO C thechange
becomes veryslow,
but themagnetic
intensity
remainshigher
than the annealed stateuntil a
temperature
of about 5oo-55oo C is attained. Theinterpretation
of thisstage presents
considerabledifficulty.
The trend of the(o-, T)
curve up to about 3ooo C can bereproduced by assuming
thatthe three constituents are
ferrite,
cementite andiron
percarbide (Fe2oC9).
This latter has beenisola-ted and measured
magnetically.
The Curietempe-rature is
2 ~o°
C which agrees well with the values obtainedby
Pichler and Merkel[14], though
sligh-tly higher
than the value of2470
Cgiven by
Hofer,
Cohn and Peebles
[15].
Ironpercarbide
however,
decomposes
into cementite above45oo
C,
and thus theonly ferromagnetic
constituent available in thetemperature
range 300°-500°C is theferrite,
thequantity being
in excess of theequilibrium
percen-tage.
Further work is in progress totry
and elucidate thesepoints.
TABLE.
Results
of
Quantitative Magnetic
Analysis of
Some I’hree-Phasecr-1Y Curçes,
andCornparison
with 1-rotal Silicon Content.w = Relative amounts of each phase; £w - i.
It is
hoped
that theexamples
considered aboveindicate the
importance
ofmagnetic
saturationintensity
measurements both in the field ofmagnetic
phase analysis
and that of thephysical
interpre-’
tation of
magnetic phenomena.
Remarque
de M. Goldman. - 11 existe unpr6-cedent pour
l’hypothèse
de Sucksmith relative à la coexistence desphases
ordonn6e etdesordonnee
dans
Fe3Al.
Smoluchowski et Newkirk ont trouv6 que dans CoPt lesphases
ordonn6e etdesordonnee
pouvaient
exister en mêmetemps.
Dans un telsyst6me,
le durcissementmagn6tique peut provenir
de la
precipitation
d’unephase
ordonn6et6tragonale
dans une matrice de
phase
cubique
desordonnee.Remarque
de 11,1. Hoselitz. -During
themagnetic
studies of iron nickelalloys
ofcomposition
approxi-matively Fe3Ni,
I found that in thenon-magnetic
y state there existed asuperlattice
which can be detectedby
magnetic
measurementowing
to thefact that a 25 per 100 nickel
alloy undergoes
irreversible
phase change
to themagnetic
oc form when cooled to thetemperature
ofliquid
air. Thusan
alloy
with 23 per 100 at.nickel,
when annealedat a
temperature
of 5250 C andsubsequently
cooledto the
temperature
ofliquid
air,
showed threemagnetic
constituents. The twoequilibrium
concen-trations at 5250C,
i. e.containing
about gand 27 per 100 nickel
respectively,
and a 25 per 100nickel constituent which must have been formed
during
theannealing
treatmentprior
tocooling
in
liquid
air(HOSELITZ,
J.o f
Iron a SteelInst.,
I9!l4,
Part
I,
p.193).
Remarque
de M.Shockley.
- In connection with the iron aluminiumalloy
discussedby
Prof. Sucks-mith it may beappropriate
to remark that the inimmediate
neigborhood
of the criticaltemperature
for order, there may be ordered and disorderedphases
present
of differentcompositions.
Theobservation
by
Prof. Smoluchowski of ordered andby
Prof. Goodman and thetheory
ofphase
diagrams
requires
ingeneral
that thecompositions
be diffe-rent. The difference in
composition
may lead to stresses and other discontinuities’which
contributeto the coercive force.
-Remarque
de iV. Sfoner. - 11 fautsouligner
l’im-portance
de ces mesurespr6cises
de la variationde 1’aimantation
spontan6e
avec latemperature.
Commeje
1’aideja
dit,
les resultats ne sont pasen accord avec la theorie
classique
pour j
2
ni avec la th6orie des electrons collectifs.Mais,
enprin-cipe
onpeut
d6duire des d6saccords comment les interactionsd’6change
varient avec1’aimantation,
cequi
est tr6simportant
dupoint
de vueth6orique.
Les mesures
magn6to-caloriques
que M. Sucksmith a commene6es a faire sur lesalliages
pourront
aussi donner desrenseignements pr6cieux.
L’analyse
des r6sultats
exp6rimentaux pr6sente beaucoup
de difficultes mais nous pensonspouvoir d6velopper
a Leeds des m6thodes
convenables;
des mesurespr6cises
comme celles de M. Sucksmith sontindis-pensables
pour cela.Remarque
de M. Néel. - Les désaccordssignal6s
par M. Stoner
peuvent
provenir
non seulementde la variation des interactions
d’6change
avec 1’aimantation mais aussi de la variation avec latemperature.
Remarque
de ll4I. Van Vleck. - 11 fautsouligner
le faitqu’a
bassetemperature
le calcul de 1’aiman-tation a saturation est très difficile. Onpeut
com-prendre
pourquoi,
a bassetemperature,
la dimi-nution de 1’aimantation parrapport
a sa valeur pour T = o estplus grande
exp6rimentalement
que la diminution calcul6c avec le mod6le duchamp
mol6culaire;
on doitplut6t
utiliser la m6thode desondes de
spin qui
donne une loid’approche
de laforme a
= o~o ~
Aux bassestemp6ratures,
ils’agit
de fluctuations li6es aux ondes despin
et non
comprises
dans la th6orie duchamp
mole-culaire. Les difficultés relatives aux bassestemp6-ratures se
pr6sentent
d’unefaçon f rappante
dans lam6thode de Bethe-Peierls
d6velopp6e
pour lemagn6-tisme par Peter Weiss. Anderson a observe
qu’avec
cette m6thode il existe une
temperature
au-dessousde
laquelle
il ne devrait pas exister def
erromagne-tisme. On doit attribuer ce r6sultat absurde au faitqu’a
bassetemperature
il existe des corr6lationset des
longues
ondesqu’on
nepeut
pas traiterfacilement par la m6thode de Bethe-Peierls.
REFERENCES.
[1] WEISS and FORRER. 2014 Ann.
Physique, 1926, 5, 153.
[2] WEISS and FORRER. 2014 Ann.
Physique, 1929, 12, 279.
[3] POTTER. - Proc.
Roy. Soc., 1934, 146, 362.
[4] BLOCH. 2014 Z.
Physik, 1930, 64, 817, and 1933, 81, 790.
[5] GUIELLAUD and Roux. - C. R. Acad. Sc., 1949, 229,
1062,
[6] GUILLAUD. - C. R. Acad. Sc., 1946, 222, 1110.
[7] ELLIS and GREINER. - Bell
Telephone Tech. Publ., B 1257.
[8] GOLDMAN and SMOLUCHOWSKI. 2014 Phys. Rev., 1949, 75,
140, 310.
[9] BENNETT. 2014 Thesis. Sheffield, 1949
(unpublished).
[10] ARDRON. 2014 Thesis. Sheffield, 1949 (unpublished).
[11] SUCKSMITH. - Proc.
Roy. Soc., 1939, 171, 525.
[12] GUILLAUD. - Rev. Métall., 1949,
46, 453.
[13] GUGGENHEIMER, HEITLER and HOSELITZ. - J. Iron and
Steel Inst., 1948, p. 192.
[14] PICHLER and MERKEL. 2014 U. S. Bureau
of Mines Techn,
Paper 718, 1949.
[15] HOFER, COHN and PEEBLES. 2014 J. Amer. Chem. Soc.,