Irreversible effects in Aug81Fe19 and Ni79Mn 21 below the de Almeida-Thouless temperature

HAL Id: jpa-00209758

https://hal.archives-ouvertes.fr/jpa-00209758

Submitted on 1 Jan 1984

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.

Irreversible effects in Aug81Fe19 and Ni79Mn 21 below the de Almeida-Thouless temperature

S. Senoussi

To cite this version:

S. Senoussi. Irreversible effects in Aug81Fe19 and Ni79Mn 21 below the de Almeida-Thouless tem- perature. Journal de Physique, 1984, 45 (2), pp.315-322. �10.1051/jphys:01984004502031500�. �jpa- 00209758�

Irreversible effects in Aug81Fe19 ^and Ni79Mn21

below the de Almeida-Thouless temperature

S. Senoussi

Laboratoire de Physique des Solides (*), Université de Paris-Sud, ^Bâtiment 510, ⁹¹⁴⁰⁵ Orsay, ^France (Reçu ^{le 13} juillet ^1983, accepté ^{le 11 octobre} 1983)

Résumé. ²⁰¹⁴ Deux alliages désordonnés, Au81Fe19 ^et Ni79Mn21 ^ont été refroidis dans des champs extérieurs fixes allant de zéro à 28 kG depuis~ ^{100 K} jusqu’a ^diverses températures ^{de mesures} comprises ^entre 1,5 ^{et 40 K.}

Leurs courbes d’hystérésis, ^{à faible champ, ont été ensuite mesurées.}

Au voisinage ^{et au-dessus d’une certaine} température caractéristique, TAT, identifiée comme celle de de Almeida-

Thouless, chacun des deux alliages ^se comporte ^{comme un} ferromagnétique conventionnel. Cependant, pour T TAT, ^{les courbes} d’hystérésis ^{des deux} systèmes deviennent extrêmement larges ^{et en même} temps ^assez différentes en forme de celles des ferromagnétiques classiques ^et très différentes entre elles. De plus, ^dans le ^cas de

Au81Fe19, ^{les effets} de mémoire induits par le champ ^de refroidissement sont très instables et effaçables par des

champs négatifs ^de quelques ^{centaines de Gauss} appliqués ^à 1,5 ^K par exemple. ^Au contraire, ^{dans le cas de} Ni79Mn21’ ^{les effets} de mémoire sont extrêmement stables quelle que soit la valeur du champ négatif, compris ^entre

0 et - 28 kG, appliqué à 1,5 ^K.

En dessous de TAT, ^des pertes d’hystérésis apparaissent ^et augmentent ^très rapidement ^{dans une} région ^de température ^{définie comme la} région ^de cross-over de de Almeida- Thouless. On suggère que les propriétés magné- tiques ^de Ni79Mn21, refroidi dans H ⁼ 0 pourraient être dues à l’apparition ^de champs d’anisotropie ^unidirec-

tionnelle pendant le refroidissement. Ces champs seraient créés par le champ moléculaire associé avec l’aimantation

spontanée ^{et auraient la même structure} spatiale que ^cette dernière.

Abstract ²⁰¹⁴ Two disordered alloys, Au81Fe19 ^and Ni79Mn21, ^{were cooled from ~} 100 K down to different temperatures ^{in the} range 1.2 K T 40 K, ^{either in zero field or in a field} ranging ^{from 50 G to} 28 kG. Their low-field hysteresis loops ^{were then measured}

Near and above some characteristic temperature, TAT, identified as the de Almeida-Thouless temperature, the hysteresis loops ^{of both} alloys present essentially ^{the same} shape ^as for standard ferromagnets. ^{Well below} TAT, ^the loops ^become quite large ^{and are very different in} shape both from the standard ferromagnetic ^{case and}

from one alloy ^{to another.} Memory effects, ^induced by ^field cooling, ^{are found to have} very different stabilities in the two alloys. ^In Au81Fe19, they ^are highly unstable and easily ^{washed out} by cycling ^an applied ^{field of a few}

hundreds gauss ^at 1.5 K. In the Ni79Mn21 alloy, they ^are remarkably ^stable whatever the cycling ^field (T ^{= 1.5} K)

in the available _range ± 28 kG. Large hysteresis ^{losses set in below} TAT and increase very rapidly ^{in a narrow} temperature region ^{defined as the de} Almeida-Thouless cross-over region.

It is suggested that the low field behaviour of Ni79Mn21 ^{cooled in zero field is} determined by the appearance of unidirectional anisotropy ^fields during cooling. ^These fields would have approximately ^{the same domain structure}

as the spontaneous magnetization and would be created by the molecular field of this magnetization.

Classification

Physics ^Abstracts

75.20E - 75.30K

1. Introduction.

At present time, ^{there is a} considerable interest in disordered

magnetic

materials in which both ferro-

magnetic

^and

spin-glass orderings

coexist Such an

interest stemmed some time _ago from the

pioneering

work of a number of

investigators

ⁱⁿ concentrated

gold-iron alloys [1-3].

^It has been found that, ^{above a}

critical concentration of about 15 at

%

Fe, the latter

alloys

^{exhibit a}

spin-glass-ferromagnetic-like

^tran-

sition. Since then, ^a number of other disordered

magnetic

materials, ^{such as NiMn} [4, 5] ^{near the}

composition

^{20 at}

%

Mn, have been found to exhibit the same sort of behaviour ^as that of concentrated AuFe. A new upsurge of activity ^{for these} alloy systems seemed to stem from recent theoretical developments

concerning

^{the real nature of the}

magnetic

^transition

just

^{mentioned Two} theoretical studies

[6,

7] ^based

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01984004502031500

316

on

Ising

^mean field model for

spins

have shown that

magnetic alloys in which the exchange interactions

are

randomly

distributed in

signs,

but based ferro-

magnetically

^{to some} extent, ^would undergo ^two magnetic transitions as a function of temperature.

The alloy which first _passes from paramagnetic ^to ferromagnetic, ^{at a Curie} temperature

Tc,

^would

exhibit a second and less conventional transition at some temperature T(H) much lower than

Tc.

^The

latter temperature ^would depend ^{on the}

applied

^field

so that the corresponding transition would be defined

by

^{a line in the}

(T,

H) plane, called de Almeida- Thouless

(A-T)

line. Since Ising spins ^{have no trans-}

verse components, the A-T line is associated with the

freezing

^{of the}

longitudinal

component ^{of the} magne- tization.

Subsequent

work based on a

Heisenberg

mean field model for

spins

[8, 9] have shown that the A-T line would not, in fact,

correspond

^{to a} coopera- tive transition but to a cross-over

region (in

^the (T, H)

plane)

^marked

by

^{the onset of} strong irreversibilities associated with the _response of the

longitudinal magnetization

^{to an}

applied

^held

According

^{to the same}

Heisenberg

^{mean field} model, ^the cross-over

regime

of de Almeida-Thouless would be preceded

(at

^a higher T)

by

^a well defined transition associated with a

canting

^{of the transverse}

components ^{of the} spins. ^This canting is defined in the

(T, H) plane by

^a

boundary

called the

Gabay-Tou-

louse

(GT)

line and is

thought

^{to be also}

accompanied by

^{some kind of}

irreversibility (though

very low).

The

canting

^{of the} transverse component ^{of the}

magnetization,

M, ^{seems to} have been

recognized

recently ^{in an}

Ni79Mn21

^{and an}

AU8,Felg alloy

[10-13]

by

^{means of both} transport ^{and Mossbauer measure-} ments, around 130 and 50 K, respectively. ^{On the}

other hand, characteristic manifestations of the cross- over

regime

of de Almeida-Thouless have also been seen, in the magnetic properties ^{of the same}

alloys

[5, 14]. However, ^{at low}

enough

temperatures, ^below the cross-over

region

which is situated

(as

^{will be seen}

later)

^{around 14 K for}

AUSIFe19

^{and - 30 K for}

Ni74Mn21,

^the

relationship

^{between M and H beco-}

mes

strikingly

different for the two

alloy

systems

just

mentioned This difference seems to subsist whatever the usual

magnetothermal

history

experienced by

these

alloys.

In this _paper, we report ^{an extensive}

study

^{of the}

relationship (M

^vs. H) ^{as well as a detailed} comparison

between the low temperature behaviours of the two

systems

AUSIFe19

^and

Ni79Mn2l’

^In

particular,

^we

will examine the

following points :

(1)

^How the whole shape ^{of the}

hysteresis loop (M

^vs. H, ^{for low}

H),

for the ZFC

sample

evolves with temperature.

(2)

How, ^{at a chosen} temperature, ^{such a}

shape

^is

modified when the

loop

is traced not

just

^{after ZF}

cooling

^{but after}

exposing

^the

alloy

^{to a field}

Hexp

^of

increasing strength,

^{the field}

Hexp ^having

^been

^applied

either during ^the

cooling

stage or, isothermally, ^at

the temperature ^of measurements.

It is to be noted that, usually, ^the transition towards

a state of strong irreversibility

(commonly

^called spin- glass

ferromagnetic

transition) [15] ^{is revealed} by ^a drop ^{of the ac} susceptibility [1-3] ^{below the}

1/N

^limit

set

by

^the value, N, ^{of the} reciprocal

demagnetization

factor. However, ^{the ac} susceptibility ^{is known to be} intrinsically less sensitive to

hysteresis

^{effects than}

are the dc measurements [16] ^and depends critically

on the

geometrical

form of the

alloy [17].

^{As a} result, the start of the fall off in the ac

susceptibility, generally,

occurs somewhere between

T GT

^and

T AT at an

^ill-

defined temperature

[17]

which is close to

TAT

^{for a}

spherical sample

^and

approaches TGT

^{for a needle} shaped sample. ^It is, therefore,

hoped

^{that the}

study

of the whole

hysteresis

loop ^{of the sort}

reported

^{in this}

article will give complementary information on the low temperature ^{state of} strong irreversibility.

2.

Experimental.

To minimize the effect of the

demagnetizing

^field,

the

Au8lFeig

^and

Ni79Mn21 specimens investigated

here were

rectangular

plates (- ^needle

shape-like) having

the dimensions - 10 x 1 x 0.08 mm’.

They

were enclosed in a quartz ^tube (in vacuum) ^annealed

~ 4 hours at 900 OC and water

quenched

^{The two} samples ^{were then}

kept

ⁱⁿ

liquid nitrogen

^{for a few}

days

until the

magnetizations

^were measured with a

home made magnetometer

(Foner-type)

^at tempera-

tures

varying

^{from - 1.2 to 40 K and under}

applied

fields _up to 28 kG. The sensitivity ^{of the} magnetometer

was about 10 - 5 emu and the field was defined to within - + 10 G.

3. Results and discussion

3.1 GOLD-IRON ALLOY. ^- The results of the low field

magnetization

measurements

(M

^vs.

H)

^{on an}

Au81Fe19

^{alloy cooled in zero field to} various fixed temperatures ^{are shown in}

figure

^1. Each isotherm has been recorded

following

^the sequence of the arrows,

starting

^{from the}

origin

^and

increasing

^the applied ^field up ^{to -} 1 kG. The field ^was then

always cycled

between about ± ¹ kG. This field value was

chosen because it was sufficient to line _up the _spon- taneous

magnetization

^at any temperature

(it

^{is of}

the order of the technical saturation field,

Hg$t,

^at

1.2 K).

As in standard

ferromagnets,

the initial magne- tization curve

(dashed branches)

^{is not}

reproducible,

whereas the rest of the

hysteresis

loop

(full branches)

is fairly

reproducible (to

^{within our}

experimental precision -

± ¹⁰ G). By

analogy

with conventional

ferromagnets

^{we can refer to} the initial

magnetization

branch

(dashed)

^{as the}

Virgin

^{curve and to the} repro- ducible

loop (full)

^{as the}

cyclic

^curve.

The most typical features of the set of

loops

ⁱⁿ

figure

^{1 are the}

following :

^{first of} all, the forms of the

Fig. ^{1. - A set of} hysteresis loops ^for Au81Fe19 ^{cooled in}

zero field (ZFC) ^{down to the} temperatures ^of measurement indicated in the figure. ^{Note the} position ^{of the} Virgin ^curve (dashed) ^with respect ^{to the} cyclic ^one (full) and its evolution with temperature.

Virgin

^{Curves are}

strikingly

different from those of

typical ferromagnets. Indeed,

ⁱⁿ the latter _case, the initial

magnetization

^{curve is}

generally

situated well inside the

cyclic

^one

presenting

^a

quite regular

^beha-

viour of

Rayleigh

type. ^In

particular,

^{near the}

origin,

it is concave

upward

^Its form, here,

depends critically

on the temperature : ^{at T 10 K}

(see

the 4.2 and 9 K

isotherms) the initial curve starts from the

origin

^with

a very low

slope dM/dH

^{= x,}

varying fairly linearly

with the

applied

^{field H} up ^to about

H,12,

^where

Hc

is the coercive field It then intersects the

cyclic

loop

at a field value close to, ^but

slightly

greater than,

H,.

After that, ^{the curve} increases rather

abruptly

^towards

the technical saturation limit. For T > 14 K, which,

as will be seen later

(see

^{also Ref}

14), approximately

represents ^{the de} Almeida-Thouless temperature

TAT,

the

virgin

^{curve becomes} quite ^{similar to} that observed with more conventional

ferromagnets.

^{An abnormal}

virgin

^curve of the kind

displayed

ⁱⁿ

figure

^{I has been}

seen in other

magnetic alloys

^{at low} temperatures

[18].

The second

interesting

feature of the curves dis-

played

ⁱⁿ

figure

^{1 concerns the evolution of the}

cycle

(in

particular

^its

width)

^with temperature. ^{As can} be seen the

cycle

^is

extremely large

^at the lowest temperature (4.2 K)

but

^{its width} drops extremely

rapidly

^{in a} temperature range centred around 9 K.

This is, perhaps, best illustrated

by

^the temperature variation of the _energy losses

(which

^{are measured}

by

the area of the

cyclic loop) displayed

ⁱⁿ

figure

^2.

As can be seen there, the losses are very large ^and approach saturation for T in the

liquid

^{helium tem-}

perature range. Upon

warming, they

^then drop ^rather

suddenly

^{in an} interval of temperature

extending roughly

^{from 6 to 14 K. Such an} interval could be taken as the cross-over

region

considered above.

These results are consistent with the behaviour of the coercive field as a function of the temperature

[14].

We now want to focus our attention on the effect of

exposing

^{the AuFe} sample ^to various fields of

increasing

strengths (applied

^either

during cooling

or at the

measuring

temperature ^{after ZF}

cooling).

The results are summarized in

figures

^{3 and 4}

by

^a

series of curves which are self explanatory. ^{We first}

stress some unusual aspects concerning ^the

stability

(or ^the

instability)

^{of these curves} upon

cycling

^several

times the field between about ±

Hsat.

^{For this we}

first consider the 1.2 K isotherm

displayed

ⁱⁿ

figure

3a, which has been recorded as follows :

The sample ^was cooled in 25 kG down to 1.2 K and the dashed curve

(labelled He

^{= 25}

kG)

^of

figure

^{3a was}

subsequently

^{recorded at this} tempera-

ture. As can be _seen, the curve intersects the field axis at a

negative

value of about - 720 G

(point

D) ^which

is about twice the coercive field

(- -

³⁶⁰

G)

^obtained

after ZF

cooling (Fig

la).

(II)

However, ^{when the}

Fig. ^{2. -} Magnetic energy losses in Au81Fe19 (ZFC) ^{as a}

functior of temperature.

318

Fig. ^{3. -} Hysteresis loop ^for Au81Fe19 ^after cooling ^{in 25 kG}

down to the indicated temperatures. The dashed branch

was traced first (cycle ⁺ 1.5 kG -+ - 1.5 kG) ^{and this was} immediately ^followed by ^the plot ^{of the full curve} (cycle

- 1.5 kG -+ + 1.5 kG - - 1.5 kG). ^{The full curve is} found to be nearly ^{identical to the} cyclic ^{curve in} figure ^1.

field is

cycled again (between

^{about ±}

HsaJ

^the

dashed curve described

just

^{above is not followed but}

there is a new

path represented by

^{the full} loop ^of

figure

3a. The latter

loop

^{is found to be almost iden-}

tical to that obtained after ZF

cooling (Fig.

la) ^and

is labelled

H,

⁼ 0 for this reason. The curves of

figures

^{3b and 3c}

(corresponding

^{to T = 4 and 8 K,}

respectively)

^were traced under the same conditions

as described

just

^{above and were found to exhibit the}

same kind of instabilities

(for

the dashed section)

as in

figure

^3a.

Fig. ^{4. -} Magnetization ^{vs. H curve for} AUs1Fe19 ^cooled

in zero field down to 1.2 K and then exposed ^{to different} high fields,

Hexp :

^full,

Hexp

^{= 0; dotted,}

Hexp

^{= 7.5 kG;}

dashed,

Hexp

^{= 15} kG; dotted-dashed,

Hexp

^{= 28 kG.}

Consider now

figure

^{4 which was}

plotted

^{after zero}

field

cooling

^{under the}

following

circumstances :

’(1)

^the

cyclic loop (full

section) ^was

plotted

^imme-

diately

^{after zero field}

cooling;

(2)

^{the field was} then increased to + 7.5 kG and reduced back to N + 1.5 kG. After that, the dotted

curve was

plottedfollowingthe cycle

^{+ 1.5 kG}

- 1.5 kG - + 1.5 kG.

Finally

the dashed and the dotted dashed curves were

plotted

^{in the same} way

as

just

^above but after the

sample

^{had now been}

exposed,

^{to 15 and 28} kG,

respectively.

^{As in the case}

of

figure

^{3 discussed}

previously,

^{we find that the curves}

recorded

just

after the

exposing

^fields

(Hexp,

^{= 7.5,}

15 and 28 kG) had been removed are never followed if the field is

cycled again

^{between ± 1.5 kG.} Instead,

we

always

follow the

cyclic

^{curve defined}

by

^{the full}

branches.

Therefore, ^{we can} say that the fact of

cycling

^the

field one or several times between about ±

Hsat

^erases

the _memory

previously acquired by

^the

Au81Fe19 alloy during

^the

application

^of any field

Hexp

>

Hsat.

The same kind of

magnetic

behaviour has been observed with more diluted AuFe

alloys [19] (pure spin glass alloys).

On the other hand, similar irre- versibilities have been ^seen in torque measurements

[20].

^{This is an}

important

property in which the present

alloy

^differs

fundamentally

^{from the}

Ni79Mn21

alloy

^{which will be}

investigated

later in this article.

We consider

again figure

3, ^{in order to} examine the modification of the

shape

of the low field loop ^{due to}

the

cooling

held Like in _pure

spin glasses

^{the modi-}

fications of the

AU81Fe19 cycle

^concerns

principally

the thermo-remanent

magnetization (TRM)

^{and the}

width of the

cycle

(or equivalently the coercive field).

More

precisely :

^{on the one hand, we observe that at}

the lowest temperature ^of measurements

(T

^{= 1.2} K,

Fig. 3a),

the residual >> remanent

magnetization (RRM

⁼ OA,

Fig. 3a)

^{is increased}

by

^{an « incre-}

mental

^{» TRM} term, AB, ^{of more} than 20

% (passing

from OA N 7.5

emu/g

^{to OB = 9.4}

emu/g),

^{due to}

the

cooling

field effect. Upon

warming

^the

sample,

the incremental TRM decreases

progressively

and,

finally,

^{seems to} vanish around 10 K, ^{that is near the}

centre of the A-T transitional

region

(because ^{of the} strong TRM, ^{it is difficult to}

study

the incremental TRM in detail as a function of

T).

The creation of such an incremental TRM, ^{and its}

(qualitative)

variation with T are in _very

good

agre- ement with what is

generally

observed in canonical

spin glasses

^{such as CuMn and AuFe}

(dilute).

^On

the other hand, ^we observe that the

magnetic cycle

^is

enlarged by

^the

cooling

field process. ^For

example,

at T = 1.2 K

(Fig. 3a),

^the

negative

coercive field is increased

by

^a factor of the order of two

(passing

^from

OC ⁼ 360 G to OD ⁼ 720

G).

^{Such an} incremental

coercivity drops

^{then with}

increasing

^{T and}

finally

seems to

disappear

around 10 K. In the same way

as above we find

(see Fig. 4)

^{that the} coercive field is also increased

by submitting

^the

sample (ZFC)

^{for a}

short time at 1.2 K to a

high

^field

(IRM process).

The

important point

^to be stressed here is that the increase of the coercive field

strength

and its variation with T, ^{are in} complete opposition ^{to the behaviour}

of the canonical

spin glasses;

^at least, ^those

having

^a

well defined

cycle

like CuMn and NiMn systems.

Indeed, it is well known that for the latter systems the width of the

cycle

^{decreases as a} function of the

cooling

field, whereas, ^{for a}

given cooling

field, ^it

increases with

increasing

temperature.

In order to find out the

possible origin

of the above

discrepancies

between the

Au81Fe19 alloy

^{and the}

canonical

spin-glasses

^mentioned

previously,

^{we recall}

that in CuMn for example, the coercive field is

thought

to be

approximately

^{the same as the} macroscopic .anisotropy field,

H k

^{which is}

given by

^the

expression

below :

Here, ^{K is a} unidirectional anisotropy energy which has been shown

(by

^{means of usual} transverse

suscepti- bility measurements) [21]

^to

depend only

^{on the tem-}

perature ^{and not on the}

strength,

mr of the remanent

magnetization, ^{nor on the} process

(TRM

^or

IRM) by

which that _mr had been created This

explains why,

ⁱⁿ

CuMn for example,

HA

^{decreases with} mr and, thus, Wth the field

Hexp ^having

^{served to create mr. As already}

noted, the behaviour of the coercive field of

AU81Fe19

is

quite

opposite ^{to the above}

predictions

^{for both}

TRM (Fig. 3) ^{and IRM} (Fig. 4) processes. However, ^it has been shown very recently [22] ^that

equation

¹

connecting H k

^K

and mr

^{is not at all} univoque, ^{so that}

the anisotropy energy K ^can be reduced and even

completely

^removed

(by

^some

appropriate preparation

of the

magnetic

state), ^at any temperature, ^without

changing

mr. Moreover, ^{it turns out} that the inde-

pendence

^{of K with} respect ^to

mr is principally

^{due to}

the homogeneous ^{nature of the} spin glass ^state.

Therefore, ^{it does not seem} easy, ^at the present time,

to

predict

^{how the} anisotropy energy and the related

anisotropy field will depend ^{on the field}

Hexp ^(applied during

^{the TRM or IRM}

processes) previously

defined We feel that it is _necessary to understand first the

properties

of K in pure

spin glasses.

3.2 NICKEL-MANGANESE ALLOY. ^- It is ^now worth

comparing

the characteristic

properties

^{of the}

Au81Fe19 alloy

studied above with those

of Ni79Mn2l

which is also

thought

^{to be a mixed}

phase alloy

^{at the}

lowest temperatures.

First, ^{the M vs. H curve}

ofNi79Mn21

^{cooled in zero}

field down to 1.2 K is shown in

figure

^{5a. As can be}

seen there the curve is

quasi

reversible with in

parti-

cular no

significant

^remanence

(m(0) - 0)

^{and no}

Virgin

^{curve of the}

king

observed in

Au81Fe19 (Fig. la)

at the same temperature.

Secondly,

^the

shape

of the 1.2 K isotherm in

figure

^{5a does not}

depend

^on

exposing

^the

Ni79Mn21 alloy

^to any field

strength

^{in our}

experimental

range

(-

28, + 28

kG);

whereas in the case of

AUSIFe19

the

cycle

^is

significantly enlarged by

^{a field as low as a}

few kG.

Strikingly enough,

all the above differences between the two

alloys practically

disappear ^{when the} temperature exceeds - 10 K as can be seen in

figures

¹

and 5 : for instance, ^{the 20 K} isotherm in

figure

¹

and the 15 K isotherm in

figure

^{5 have} very

nearly

the same shape. ^{It is} also instructive to look at the variation with T of the _energy losses in the two systems.

It is clear that at low

enough

temperatures

(liquid

helium

temperatures)

the losses are almost non-existent in NiMn

(Fig. 6)

^while

they

^{show a} maximum and are

extremely high

^{in AuFe}

(Fig.

2).

Upon

warming,

^we observe that the losses decrease rather

abruptly

^{in the case of AuFe}

alloy (Fig. 2)

whereas

they

first increase and _pass

by

^a maximum,

at about 13 K, ^{in the case of NiMn}

(Fig. 6).

^For

higher

T, above - 13 K, ^the losses in the two

alloys

^become quite

comparable (when

considered at some reduced temperature

equal

say ^{to -}

TpTI2)

^{and exhibit,}

qualitatively,

^{the same} form of variation with T.

Finally,

it is worth

noting

^{that the curve}

displayed

in

figure

⁶ presents very much the same

shape

^{as those}

associated with rotational losses

(deduced

^from torque measurements) ^{in a} somewhat similar NiMn

alloy, reported

^a

long

^time ago

by

Kouvel and Graham

[23]

(see Fig.

^{5 of Ref.}

23).

To complete ^our comparison between the two mixed phase systems, ^we present ⁱⁿ

figure

^{7, three}

320

Fig. ^{5. - A set of} hysteresis loops ^of Ni79Mn2l ^{cooled in}

zero field down to the temperatures ^{indicated Note that} above - 15 K, ^{the curve becomes} very similar to that of AuFe in figure ^1.

curves, M vs. H,

plotted

after the

Ni79Mn21 sample

had been cooled

(from

^{40 to 1.2} K),

respectively,

ⁱⁿ

H ⁼ 0, ⁴⁵ G, ¹ kG and 28 kG. As noted earlier,

by

Fig. ^{6. -} Magnetic energy losses in Ni79Mn21 (ZFC) ^{as a}

function of temperature. Note the maximum at about 13 K and the ressemblance above 14 K with the energy losses in AuFe (Fig. 2). ^{We have checked} (see ^Ref. 25) ^{that the}

maximum does not seem to correspond ^to any transition.

Fig. ^{7. - A set of} magnetization ^vs. field data for Ni79Mn21

after cooling in various fields (indicated) ^{down to 1.2 K.}

Note that the displaced magnetic loops obtained after

cooling either in 1 kG or in 28 kG are practically ^{the same.}

Aitken et al.

[4],

^the

magnetic cycle

^is

critically

^affected

by extremely

^weak

cooling

^fields

(as

^{low as 45} G,

Fig.

7),

leading

^{to a}

cycle displaced

^towards

negative

values. Such a

displaced cycle

^{reaches a stable}

position

as soon as the

cooling

^{field exceeds - 1} kG, ^{so that}

there _are,

practically,

^no differences between the

cycle

obtained after

cooling

^{in 1 kG} and that obtained after

cooling

^{in our} maximum, ^{available field of 28 kG.}

Moreover, ^we have checked that the

displaced cycle (whatever

^the

cooling field)

^was

exactly reproducible (to

^{within our}

experimental

errors) upon

cycling

^the

field

isothermally

^{at T = 1.2} K, between the _upper limits

(± 28

kG) ^{of our}

experimental

^{fields. Therefore}

the ensemble of these properties

(connected

^with

Fig.

7) ^is

highly contrasting

^{with the case}

of AUS1 Fe 19

for which, ^for instance, the

displaced cycle

^{and more}

generally

^the memory

resulting

^from any

positive cooling

^{field was found to be}

easily destroyed by

^a

negative

field of less than 2 kG.

Strikingly enough,

^all

the differences in behaviour discussed above in connec-