You are on page 1of 11

J. Phys. Chem. Solids Pergamon Press 1971. Vol. 32, pp. 939-949. Printed in Great Britain.

A MODEL FOR THE INVAR ALLOYS AND THE


Fe-Ni SYSTEM

W. F. "SCItLOSSER
Department of Physics, University of Toronto, Toronto 5, Ontario, Canada

(Received I 1 June 1970; in revised form 14 August 1970)

A b s t r a c t - A model is proposed to account for the metallurgical and physical properties of f.c.c.
F e - N i alloys, including the lnvar anomalies around 34 at.% Ni. Microscopic inhomogeneity due to
short range atomic ordering on the basis of the FeNia compound is assumed to be a dominant feature
of the alloy system. Regions of both f.c.c. Fe and FeNi3 are present in alloys in the invar region. The
electronic state of only the Fe atoms, and hence their volume, magnetic moment, and sensitivity to
external variables, is further assumed to be a function of the number of Ni nearest neighbours, which
in turn depends on such variables as cold work. With these assumptions, the properties of the annealed
alloys above the Curie point are developed, the phase diagram and metastability of the alloys is
explained and the lnvar anomalies below the Curie point are accounted for. Furthermore, the model
explains the unusual low temperature properties of these alloys as due to exchange coupling between
ferromagnetic FeNia regions and antiferromagnetic Fe regions. An extension of the model is made to
other binary alloy systems that exhibit Invar-type anomalies:

1. INTRODUCTION small and negative only above 70% Ni.


THE F.C.C. iron-nickel alloys in the Invar Similarly, a negative term proportional to the
concentration range, ~ 35 at.% Ni, are ano- temperature is still present[9] in the experi-
malous in many respects; in particular, the mental fit to the thermal expansion near T ----0
fall of the T --- 0 magnetization with increased for a 50 at.% Ni alloy.
concentration of Fe [1,2], the departure from Attempts to explain these anomalous prop-
Vegard's Law[3] and the associated large erties have often involved the assumption of
spontaneous volume magnetostriction[4] and a random distribution of Fe and Ni atoms
low thermal expansion[5] below the Curie [10-17], with some further assumptions about
temperature, may be cited. The preceding differences in the interatomic interactions for
paper[6] has described measurements of the the different possible pairings[16, 17]. These
large forced magnetostriction below room models do not seem able to account for the
temperature in invar. There is also an ano- persistence of the anomalies to very high Ni
malously large paraprocess susceptibility [7,8] concentrations.
and large temperature variation of the mag- In this paper we propose a new model
netization [ 1,7,8]. applicable not only to the invar alloys, but to
A feature of these anomalies is their per- the whole F e - N i system. The invar anomalies
sistence to quite high Ni concentrations in then are seen to be a consequence of the
somewhat attenuated form. Deviations in the factors that determine the behaviour of this
concentration dependence of the lattice system.
parameter are evident at -- 57 at.% Ni[3], and The existence of microscopic inhomogeneity
in the magnetic moment at - 5 8 at.% Ni[1]. in the alloy is regarded as a premise. Given
The positive volume anomaly [5] that occurs this, the model rests on two main assumptions:
below the Curie temperature of the invar first (a) that the atomic ordering of F e - N i
alloys, although largest at - 3 4 a t . % Ni, is alloys is not random, but is characterized by
still quite prominent at 60% Ni and becomes extremely marked short range atomic order
939
940 W.F. SCHLOSSER

on the basis of the FeNi3 compound with the compound is stressed in this paper. To illus-
tic.c. L 1~(CuAua) structure [ 18], and secondly trate this, in Fig. 1 we have plotted the liquidus
(b) that the number of Ni nearest neighbours for the F e - N i alloy system, the heat of for-
determines the electronic state of a Fe atom mation of these alloys, the coefficient of the
while the Ni atoms are not sensitive to the linear term in the low temperature specific
local atomic environment. heat, and the D e b y e 0. Again we wish to
These assumptions are expanded on and point out that it is not the peculiar behaviour
justified in Section 2. The model is developed of these alloys around 35 at.% Ni that is
from them in Section 3, and applied to dis- significant, but the observation that this
tinguish between the behaviour of well an- peculiar behaviour begins near 75 at.% Ni.
nealed and cold worked alloys. This first assumption will allow us to make
The model is applied to the explanation of semi-quantitative statements about the local
the invar properties in Section 4, with environments of particular atoms.
particular emphasis on the results of the pre- Collins et al.[22] proposed a structure in
ceding paper. Section 5 generalizes the model which the cube corner sites of the f.c.c, cells
to other invar-type alloy systems. are occupied mainly by Fe atoms and the
face-centre sites are randomly occupied by the
2. ASSUMPTIONS remaining Fe atoms and the Ni atoms. The
In the models of several authors[10-15] structure maximizes the number of F e - N i
inhomogeneity has been arbitrarily assumed nearest neighbour pairs. The present assump-
to avoid a singularity in the alloy properties at tion differs from theirs in stressing the impor-
30 at.% Ni. While the introduction of a fine tance of the contribution to the stability of the
scale inhomogeneity certainly has important system of the FeNi3 structure as a whole. It
consequences for such models, it has been is probably not possible to decide between
given no experimental or theoretical justifi- these models on grounds of their relative
cation. The model of Shiga and Nakamura stability by a fundamental calculation; the
[19,20] also assumes a random distribution of analogy with the f.c.c. C u - M n alloys, which
Fe and Ni atoms[21], but with a variable are known to evolve by segregation of Mn
electronic state for the Fe atoms. This atoms [27], would seem to offer some support
random distribution is insufficient to explain for the present model. In any case, the main
the properties of F e - N i alloys over a wide practical difference between these models
range of compositions and temperatures. lies in the existence of a more well defined
Neutron scattering data[22-24] and elec- interface between Fe-rich and Ni-rich regions
trical resistance studies [25,26] on the F e - N i under our assumption. The properties of the
system indicate that extensive short range system are then conveniently discussed in
atomic order is present over wide composition terms of the positions of atoms with respect to
and temperature ranges in supposedly dis- the interface, or transition region, without
ordered alloys. This ordering is based on the reference to the details of the background
FeNi3 compound. This compound is ferro- statistical inhomogeneity which is the impor-
magnetic[18] with a Curie temperature tant feature of the alternative assumption [22].
between 863 and 980 K, probably closer to Our second assumption relates the atomic
863 K. There is considerable uncertainty in behaviour to particular environments. It
this temperature as the compound disorders follows more nearly the idea of Shiga than the
at a lower temperature, 780 K. The Curie 'two states of iron' idea of Weiss [33], in that
temperature of the disordered alloy with we suppose that the electronic configuration
75 at.% Ni is -- 873 K. The importance of the of a particular Fe atom, which determines its
short range atomic ordering based on the FeNi3 interaction with nearby atoms and hence the
A MODEL FOR THE INVAR ALLOYS 941

oF- (

~ 150(? -
I.u
n--

i 147~

~ 1425~

-~~176
I
4Bob- I 9
9 - ,zz

/?
~- ~/ l-I ~~

'~176 7 /-q~

3ao - ~ ~ -~4.o
/
T c rrex ~ 1 1 1 3 . 6
30(
20 30 40 50 60 70 80 90 I00
COMPOSITION ( ATOMIC% Ni )
Fig. l. The liquidus[28], the heat of formation[29, 30] and
the low temperature specific heat 3,[31] and 0[31,32] for the
Fe-Ni alloy system.
942 W.F. SCHLOSSER

nearest neighbour distances and the ordered bour distances, and an expansion of the whole
state magnetic moment on the atom, can vary alloy relative to a linear interpolation
between two limits depending primarily on (Vegard's Law) of the lattice parameters of the
local atomic environment, and on external two components of the alloy, f.c.c. Ni and
variables such as temperature and magnetic f.c.c. Fe. Thus, this assumption alone explains
field. This assumption has theoretical support why the presence of Ni nearest neighbours
in the work of Goodenough[34] and Bailyn produces an increase in volume and a large
[35]. The main experimental justification for positive departure from Vegard's Law above
this assumption is the observation that the Tc. Another related effect appears for those F e
lattice parameter of 3' Fe in its region of atoms with insufficient Ni neighbours to have
stability (1183 to I662 K) linearly extrapol- the maximum atomic volume at To. As the
ates to a 300 K lattice parameter o f - 3.574A alloy is cooled below its Curie temperature
[36,37]. The 300 K lattice parameters[18] of these atoms will increase their moment due to
F e - C , F e - N , and F e - M n alloys against a net increase in the magnetization with
concentration extrapolate also to - 3.57 A for decreasing temperature. The conversion of
pure Fe. However, Fe atoms enter into solu- non-magnetic, bonding, 3d electrons to
tion in F e - N i , F e - P d and F e - P t alloys, for magnetic, non-bonding electrons in the polar-
not too large Fe concentrations, causing a ization process reduces the bonding of this
change in lattice parameter[38] which extra- atom to its neighbours, resulting in a large
polates to a 300 K lattice parameter of - 3.64 positive spontaneous magnetostriction below
A in each case. The small parameter state of the Curie temperature and a corresponding
Fe appears to be antiferromagnetic at low large decrease in the thermal expansion
temperatures with a magnetic moment of coefficient. Furthermore, an applied magnetic
1 /x/~ or less [39-42]. The large parameter field, by increasing the magnetization also
state of Fe contributes a large moment (~ 2.8 converts Fe atoms to high moment states,
/zB) per atom when alloyed with Nil1], Pd producing a large high field susceptibility and
[43-46] or Pt[47] at sufficiently high concen- forced volume magnetostriction. The essential
trations (7 - 10 at.%). An Fe atom with all Fe feature of this paper is the application of these
nearest neighbours will then have a low ideas to the particular constraints imposed on
'volume', (the volume of an atom being a model in which most of the Ni atoms are
defined in terms of the nearest neighbour in the FeNiz compound.
distances), and have a small magnetic moment
oppositely oriented to the moment on the 3. THE MODEL
atoms around it at low temperatures. Above 1. lnhomogeneity and the alloy structure
some minimum effective number of Ni We consider first a well annealed alloy. By
neighbours, an F e atom has essentially its this we mean an alloy annealed at low tem-
maximum moment and volume. Between peratures sufficiently to allow the diameters of
these extremes the moment and volume of the the ordered regions to exceed several lattice
Fe atom in question are strongly dependent on parameters. These regions are single phase
the number of Ni nearest neighbours, as well ordered domains and are generally of irre-
as on external variables. The increase in gular size and shape. An alloy contains three
volume with magnetic moment is a result of a regions of interest due to the existence of
change of bonding electrons with no moment, ordering. These are illustrated in Fig. 2. Type
into non-bonding electrons with magnetic 1 regions consist of FeNi3 of assumed perfect
moment[19,48]. This lower number of bond- order. In this core, each Fe atom has ideally
ing electrons results in a reduction of the 12 Ni nearest neighbours and exhibits its
cohesive force, an increase in nearest neigh- maximum moment and nearest neighbour
A MODEL FOR THE INVAR ALLOYS 943

meter is the volume of the type 2 regions in an


alloy. For well annealed alloys, it may be
assumed that the thickness of type 2 regions
is approximately constant and independent of
the size of type 1 regions.
The general effects of decreasing the nickel
concentration in an alloy from the 75 at.% Ni
ordered compound are then as follows. D o w n
to 65-70 at.% Ni no change in the alloy proper-
ties are apparent. The minimum number of Ni
nearest neighbours has not fallen below the
I I
defining value f o r type 1 regions. At concen-
7.'
I I
FeNi 3
trations below 65 at.% Ni, added Fe atoms
can no longer be accommodated in ordered
I I
I t FeNi3 regions, and begin to appear as isolated
5 TYPE 3 I TYPE 2 TYPE I
type 2 regions, still containing some nickel

'/I
i I which will influence the state of the Fe atoms.
,,=, I I As additional iron is added to the alloy, the
i I
8 i. J total alloy volume in type 2 regions increases.
The presence of ordering in the alloy, obtained
E, Fe i I through atomic diffusion, dictates that the
DISTANCE ALONG SECTION INDICATED IN UPPER DIAGRAM type 2 regions take the form of thin layers
Fig. 2, The three types of regions defined in our model of tending to surround and isolate the ordered
the F e - N i alloys. The upper diagram illustrates the type 1 regions. A maximum volume of type 2
regions for a low nickel concentration alloy. The lower material occurs when the ordered regions are
diagram illustrates the variation of the nickel concentra-
tion in the various regions. just separated from each other by type 2
material. The composition at which this
distance. Type 2 regions form shells around occurs may be estimated by considering the
the regions of perfect order and are the type 1 material to be in spheres. The maximum
regions of decreasing Ni concentration and surface area of the spheres (and therefore
decreasing order that separate the perfectly maximum type 2 volume for a well annealed
ordered type 1 regions from regions designated alloy) will occur ,when the spheres come into
type 3 which are the f:c.c, iron excluded from contact. If the spheres are in a cubic array,
regions 1 and 2 due to the existence of FeNi3 this will be when half the total volume is
order in Fe rich alloys. Type 3 regions contain inside the spheres.
no nickel, but consist of the segregated f.c.c. If the material outside the spheres is pure
Fe atoms with small nearest neighbour dis- Fe and that inside is FeNi~, the maximum
tance ( - 1 . 2 6 A ) and with low moment occurs at - 37.5 at.% Ni. This figure is in good
(~ I/zB or less). agreement with the Ni concentration of the
The boundary of a type 1 region is drawn alloy exhibiting the largest invar anomalies
along iron atoms with sufficient Ni nearest (~ 35 at.% Ni) and with the Ni concentration
neighbours that they exhibit practically their of the alloy exhibiting the maximum deviation
full volume and moment at high temperatures from Vegard's Law . ( - 43 at.% Ni). The
(above ~ 9 0 0 K ) . The material in type 2 difference in these concentrations can be
regions is that which can exhibit large changes understood when the finite thickness of the
in volume or moment in response to changes type 2 regions is considered. The maximum
in external variables. Thus an important para- high temperature deviation_from Vegard's
944 w. F. SCHLOSSER

Law is expecte.d to occur at a higher Ni ordered type 1 regions. The stability of the
concentration than that corresponding to the alloy can be maintained only if a large inter-
maximum in the invar anomalies, because the face area is present. As we have just seen,
former is determined by the amount of type 1 annealing reduces this. The Fe regions grow
material only (by the expansion mechanism until they become unstable, and then trans-
outlined in the previous section), while the form to the b.c.c, phase. The alloys are
anomalies are due to the type 2 material. The metastable up to 65-70 at.% Ni. The time
maximum amounts of type 1 and type 2 will required for equilibrium is, however, extremely
not occur at the same Ni concentration if long below 400~ as the necessary migration
there is a finite thickness of type 2 material. of nickel atoms must occur by diffusion, a very
slow process at low temperatures.
2. Effect of cold working From our model the low temperature
For well annealed alloys the thickness of equilibrium is between b.c.c. Fe and the
the type 2 regions is approximately inde- ordered FeNia compound with some excess of
pendent of further annealing, yet the size of Fe. This is in good agreement with the
the ordered type 1 regions increases with experimental phase diagram of Hansen[52]
further annealing by growth of the more (Fig. 3) and with the more recent delineation
favourable regions at the expense of others. of the phase boundaries of the ordered regions
Therefore the type 3 regions will increase in o f the Fe-Ni system by Kneller[53]. The
size with further annealing, while the volume phase diagram of Ananthanarayanan and
of the type 2 regions decreases. Peavler[54], obtained with very fine particles,
In general, since both quenching and cold extends the region of stable ordered f.c.c.
working result in a reduction of the degree of alloy to - 5 7 a t . % Ni. At concentrations
order in an alloy, both processes increase the below 65 at.% Ni, this model is in accord with
amount of type 2 material at the expense of the work of Kneller in assuming the lack of
material of both t y p e s 1 and 3. Remembering strong FelNil ordering in the Fe-Ni system.
that the amount of type 2 material is a measure 160C
of the magnitude of the anomalies, the model's UID
prediction of an increase of this with increase
i n t h e disordering of the alloy is in agreement 140C
MINI~MUM
with experiment[4,49]. These assumptions as 9"-'67ot % Ni

to the presence of FeNi3 ordering are suffi-


120C
cient to explain the effects of rolling on the
properties of invar single crystals observed by
Chikazumi et al.[49], which they could not IOOC
interpret on the basis of Fe3Ni ordering.
7" Fe Ni3
3. Stability and the phase diagram
:o 80C
I-
!
f.c.c. 3, iron is unstable below 1183 K, and ~ \ ~66~t % Ni

transforms to the b.c.c. ~ form. In finely I- 60C

divided form, f.c.c. Fe can be maintained


below this temperature through coherency
with a suitable lattice[50,51] although the
lattice match is rarely perfect, and there is in ^^:~o%
general a coherency strain at the interface. In 200 :F~uTo~\ I I I I ~l-, I I
I0 20 30 40 50 60 70 80 90 I00
the f.c.c. F e - N i alloys, the f.c.c. Fe regions ATOMIC PER-CENT NIC~C
are stabilized by coherency with nearby F i g . 3. T h e phase d i a g r a m o f t h e F e - N i system [52].
A MODEL FOR THE INVAR ALLOYS 945

4. EXPLANATION OF INVAR PROPERTIES sent: pure f.c.c. Fe, ordered FeNia and the
1. General type 2 transition region, one additional
Turning to application of these ideas to the assumption is necessary: as mentioned, it is
observed properties of the invar alloys, suffi- supposed that the type 3 regions are anti-
cient mention has already been made of those ferromagnetic at low temperatures, for in-
properties which depend on the inhomogen- vestigations[57] have shown that f.c.c. Fe is
eity per se, such as the departure from antiferromagnetic at low temperatures with a
Vegard's Law. Neel point below 100 K. Mossbauer results
A large class of anomalous properties for low Ni concentration alloys[2,58-61]
depend particularly on the amount of type 2 show that a paramagnetic line, which we attri-
material present, since this is the fraction bute to the type 3 regions, is prominent. It
whose behaviour is sensitive to external broadens below 90 K[2,58,59], indicating an
influences. It may be stressed again that this antiferromagnetic transition in type 3 regions
model explains in a natural way the persis- to a magnetic state with low internal field.
tence of anomalies in properties like the Extrapolation of the N6el temperature of
paraprocess susceptibility or thermal expan- antiferromagnetic F e - M n alloys[62-65] to
sion to high Ni alloys. The general nature of pure Fe also indicate that f.c.c. Fe should
the magnetic ordering with decreasing tem- become antiferromagnetic below - 80 K.
perature according to this model is as follows According to this model, the invar alloys
-individual regions of the alloy interact more have a very unusual structure below this
strongly through common type 2 regions or transition. Type 1 and type 2 regions are
type I overlaps to form larger regions of ferromagnetic, type 3 regions are antiferro-
magnetic order, and eventually a macroscopic magnetic, and the two types of magnetic
Curie transition. The Curie transition is thus regions interact at the boundary between the
quite broad, with a high temperature tail, type 2 and type 3 regions. In this temperature
reflecting the inhomogeneity of the alloy range both regions vie for control of the type 2
[4,55,56]. The continual increase of the material and there is a very large area of
magnetic moment of the alloy due to both the interface between the ferromagnetic and
coupling of isolated regions and increased antiferromagnetic regions. Kouvel[66] pro-
induced magnetization in type 2 regions posed a mechanism of very strong coupling
results in a curve of magnetization vs. tem- between ferro- and antiferromagnetic regions
perature that falls considerably below a in f.c.c. Mn alloys to explain their magnetic
Brillouin curve fitted at T = 0 and at To. This properties. These alloys are believed to have a
flattened curve is a well known feature of the similar microstructure to that proposed here
invar alloys[I]. The large spontaneous for the F e - N i alloys. This type of strong
magnetostriction and hence the thermal coupling between ferro- and antiferromagnetic
expansion anomaly below the Curie point are regions has been observed in other materials
a consequence of the increased induction of [67] and is attributed to exchange of magnetic
Fe in type 2 regions due to increased ferro- electrons between atoms on opposite sides of
magnetism in the alloy. the boundaries of the regions. I n an ideal
ferromagnet and an ideal antiferromagnet the
direction of the magnetization at zero field
2. Exchange coupling in the 35% Ni invar relative to the crystal axes is determined by
In order to fully explain the results pre- the magnetic anisotropy of the material. F o r
sented in the preceding paper[6], and other the antiferromagnet this refers to the sub-
data for invar alloys in the vicinity of 35 at.% lattice magnetization. Rotation of antiferro-
Ni, where all three types of regions are pre- magnetic moments is comparatively difficult
946 W. F. SCHLOSSER

and requires large magnetic fields to achieve TE, at which the couplingenergy is equal to
small rotations. Thus, if there exists an the antiferromagnetic anisotropy energy, is
exchange coupling between ferromagnetic and approached, ferromagnetic and anti ferro-
antiferromagnetic atoms at the boundary magnetic regions become more tightly ccupled.
separating them, the antiferromagnetic atoms The antiferromagnetism of the type 3 regions
can seriously affect the behaviour of the increasingly restricts the induction of ferro-
ferromagnetic region. magnetism in adjoining type 2 regions. The
We assume that this coupling energy has a forced magnetostriction thus decreases in this
temperature dependence and magnitude region, and the rigidity of the lattice increases
relative to the anisotropy energy of the anti- as less material can transform between the
ferromagnet as given in Fig. 4. Just below the low volume and the high volume states.

ANTIFERROMAGNETIC
ENERGY

~ISTROPY EXCHANGECOUPLING

tr
i ~ f~ ~ ~ ~ ENERGY
t.~
z

I
I
o rE
TEMPERATURE
Fig. 4. The relation assumedbetweenthe anisotropy energyof the antiferro-
magnetic regions and the exchangeenergy coupling these regions to neigh-
bouring ferromagnetic regions.

N6el temperature, the anisotropy energy of Experimentally an increased rigidity of the


the antiferromagnet is small and the ability of m a t e r i a l - a decrease in the magnetoelastic
the ferromagnetic regions to respond to a coupling-is observed below - 2 0 K. The
magnetic field is only slightly reduced by the forced magnetostriction decreases by - 3 0
constraint of the exchange coupling to the per cent below this temperature[6], the mag-
antiferromagnetic regions. However, losses netic susceptibility exhibits a corresponding
can be expected in this region as the anti- decrease [70], the velocity of sound rises [68],
ferromagnetic moments are rotated against and the compressibility correspondingly falls
their anisotropy energy. A peak in the ultra- below ~ 40 K.
sonic attenuation has been reported for invar According to our model, cold worked and
alloys below 100 K by both Meincke and Litva annealed alloys are distinguished by the
[68] and by Fletcher[69]. As the temperature amount of type 2 material present. As we have
A MODEL FOR THE INVAR ALLOYS 947

seen, increased coupling of the antiferro- a real alloy due to the variable conditions
magnetic regions to the lattice reduces the affecting different regions in the alloy. For a
ability of type 2 regions to respond to changes change in magnetic field, the response of the
in temperature or magnetic field. For very alloy, principally occurring at the boundary
strong coupling, i.e. below TE, cold worked between type 2 and type 3 regions and
and annealed alloys can be expected to behave involving a change in the state of Fe atoms,
identically, as has been observed experimen- relaxes through a volume creep because of the
tally [6]. Despite large differences in the ther- strong magnetoelastic coupling. The slope of
mal expansion coefficient above 2 0 K for the logarithmic creep curve and the extent of
entirely different samples from different the relaxation depend on the existence of a
manufacturers and of different purity, there is well defined boundary between type 2 and
remarkably little difference between the type 3 regions, which is favoured by annealing.
values for these samples below 20 K. Similarly The temperature dependence of the rate of the
the forced magnetostriction of annealed and logarithmic creep thus reflects the shape of the
cold worked samples, from the same rod, anomaly at 90 K [6].
begin to converge to a common value at 20 K. The decrease in magnetoelastic coupling
The response of such exchange coupled observed below 20 K is a common feature of a
magnetic systems to a rapid change in external wide range of alloys, (up to 45% Ni). In the
conditions is generally logarithmic in time. For alloy Fe65(Ni10.TMn0.3)35 there is a sharp drop
example, experiments on exchange coupled in the magnetic moment of the alloy below
C o - C o O and N i - N i O films by Schlenker [71] 20 K[19]. The magnetization results of
have shown this time dependence of the Cochrane and Graham[18] also show an
response of the magnetization to a change in unexpected fall of the moment from the high
magnetic field. A logarithmic time dependence temperature extrapolated values below 20 K
of the rotational torque at high magnetic fields (1 in 104). This similarity of the temperature
was found in Fe65 (Nil-xMnx)35 alloys ex- dependence of the magnetoelastic coupling,
hibiting exchange coupling by Nakamura and despite a wide range of magnitude and alloy
Miyata [21]. It is expected in view of the compositions can be accounted for quite
observation of rotational torque hysteresis in readily by our model where the temperature
29-32 at.% Ni invar alloys by Nakamura, dependence of the coupling exhibits different
Takeda and Shiga[59], which correlated with behaviour above and below a temperature
the widening of the paramagnetic Mossbauer determined only by the antiferromagnetic
line in these alloys at low temperatures, that anisotropy energy of the pure Fe type 3
exchange coupling time effects would be found regions, which we assume occur in all of these
in the higher Ni invar alloys. The observation alloys, and the exchange coupling between
of a logarithmic time dependence of the type 3 and type 2 regions, which does not
volume of a 33 at.% invar alloy following a vary greatly with alloy composition.
change in the magnetic field (above saturation
only) seems to confirm its presence. The CONCLUSIONS
temperature dependence of this length re- In summary, we have proposed a model for
laxation is similar to the temperature depen- the F e - N i alloy system, which stresses the
dence of the torque relaxation reported for the atomic inhomogeneity. By assuming that this
Fees(Nil-xMnx)a5 alloys[21]. This tempera- inhomogeneity is due to short range atomic
ture dependence follows from Fig. 3, where order based on the FeNi3 compound, and that
relaxation can only occur by thermal excita- the state of an Fe atom depends on the number
tion for temperatures within kT of TE (the of Ni nearest neighbours, the model is made
hatched regions) but with some broadening in sufficiently simple that it can be used to make
948 W. F. S C H L O S S E R

predictions of a wide variety of physical invar and related alloys should provide
properties that turn out to be reasonably in further correlation of the physical and
accord with experiment. metallurgical properties of these alloy systems.
It is possible to generalize this model to
Acknowledgements-The author wishes to thank Dr. G.
other systems under the following conditions M. Graham, Dr. P. P. M. Meincke, and Dr. E. Fawcett
for the appearance of invar properties. for their assistance in the preparation of this paper.
(1) The alloy system must exhibit consider- This work was supported by the National Research
Council of Canada.
able short range atomic ordering based on a
magnetic compound in a paramagnetic or REFERENCES
diamagnetic matrix. The compound involved I. C R A N G L E J. and H A L L A M G. C., Proc. R. Soc.
in the invar properties will be the most stable 272, 119 (1963).
compound in an alloy system. 2. A S A N O H.,J. phys. Soc.Japan27, 542 (1969).
3. O W E N E. A., Y A T E S E. L. and S U L L Y A. H.,
(2) The magnetic compound and matrix Proc. Phys. Soc. 49, 315 (1937).
must have the same or a similar crystal 4. T I N O Y. and M A E D A T., J. phys. Soc. Japan 24,
structure, similar, though not equal, lattice 729 (1968).
5. C H E V E N A R D P., Rev. Metall. Paris, 25, 14 (1928).
parameters and some degree of coherence 6. S C H L O S S E R W. F., G R A H A M G. M. and
across the compound-matrix interfaces. M E I N C K E P. P. M., (Preceding paper).
The invar anomalies will b e m o s t intense at 7. T I N O Y. and M A E D A T., J. phys. Soc. Japan 18,
955 (1963).
a composition approximately half-way be- 8. C O C H R A N E R. W. and G R A H A M G. M., Can. J.
tween the composition of the matrix and the Phys, 48, 264 (1970).
stoichiometric composition of the compound 9. W H I T E G. K., Proc. Phys. Soc. London 86, 159
(1965). "
responsible for the short range order. 10. S H I M I Z U M. and H I R O O K A S., Phys. Lett. 27A,
There are two other binary alloy systems 530(1968).
that exhibit invar anomalies: F e - P d and 11. S H I M I Z U M. and H I R O O K A S., Phys. Lett. 30A,
133 (1968).
Fe-Pt. They differ from the F e - N i system in 12. K A T S U K ! A. and T E R A O K.,J. phys. Soc. Japan
easily forming stable ordered I:1 ferromag- 26, 1109 (1969).
netic compounds FePd[72] and FePt[73]. 13. T E R A O K. and K A T S U K I A., J. phys. Soc. Japan
27, 321 (1969).
The invar anomalies are found in the iron-rich 14. T E R A O K. and K A T S U K i A., J. phys. Soc. Japan
disordered alloys of these systems, are most 27,826 (1969).
intense in the region 2 5 - 2 7 at.% Pd or Pt 15. K A C H I S. and A S A N O H., J. phys. Soc. Japan
27,536 (1969).
[74-78] and are destroyed by the Fe3Pt order- 16. K O N D O R S K l l E. I., Soviet Phys. JETP 10, 1284
ing in the F e - P t system[79,80]. Also, the (1960).
concentration dependence of the invar 17. S I D O R O V S. K. and D O R O S H E N K O A. V.,Phys.
Status Solidi 16, 737 (1966).
anomalies is sharper for the F e - P d and F e - P t 18. K O U V E L J. S., In lntermetallic Compounds (Edited
systems, and the anomalies are stronger by J. H. Westbrook), p. 540. J. Wiley, New York
[5,74-80]. (I 967).
19. S H I G A M.,J. phys. Soc.Japan 22, 539 (1967).
Partial satisfaction of the above conditions 20. S H I G A M. and N A K A M U R A Y., J. phys. Soc.
in other alloy systems may result in the Japan 26, 24 (1969).
appearance of weak invar-like anomalies. We 21. N A K A M U R A Y., and M I Y A T A M.,J. phys. Soc.
Japan 23,223 (1967).
have tried to show that the effects of inhomo- 22. C O L L I N S M. F . , J O N E S R. V. and L O W D E R. D.,
geneous structure of the alloy must be con- J. phys. Soc.Japan 17, Suppl. B-Ill, 19 (1962).
sidered as an essential part of the analysis of 23. G O M A N ' K O V V. 1., P U Z E I 1. M. and R U K O -
S U E V M. N., Soviet Phys. crystallogr. 13, 449
data on such systems. These effects may even (1968).
dominate the behaviour of the material. 24. G O M A N ' K O V V. i., P U Z E I 1. M. and M A L T S E V
Although present knowledge of the con- E. I., Urk. Fiz. Zh. 14, 1716 (1969).
25. M A R C H A N D A., BROSSON P. and D A U -
sequences of these structural features is T R E P P E D., C.R. Acad. Sci. Paris 262C, 1770
limited, a re-examination of existing data on (1966).
A MODEL FOR THE INVAR ALLOYS 949

26. C H A M B E R O D A. and P A U L E V E J., J. Phys. 54. A N A N T H A N A R A Y A N A N N. I. and P E A V L E R


Chem. Solids 29, 1683 (1968). R. J., In Advances in X-Ray Analysis (Edited by
27. S M I T H J. H. and V A N C E E. R., J. appl. Phys. 40, J. B. Newkirk and G. R. Mallet), Vol. 10, p. 240.
4853 (1969). Plenum Press, New York (1967).
28. H A N S O N D. and F R E E M A N J. R., J. Iron Steel 55. T I N O Y. and M A E D A T., J. phys. Soc. Japan 18,
Inst. 107, I, 307 (1923). 955 (1963).
29. S T E 1 N E R W. and K R I S E M E N T 0., Arch. Eisen- 56. BELOV K. P., In Magnetic Transitions, p. 74.
huttenw. 32, 701 (1961 ). Consultants Bureau, New York ( 1961 ).
30. D E N C H W. A., Trans. Faraday Soc. 59, 1279 57. K O H L H A A S R. and L A N G E H., Z. angew. Phys.
(1963). 18,453 (1965).
31. D I X O N M., H O A R E E. E. and H O L D E N T. M., 58. A S A N O H., J. phys. Soc. Japan 25,286 (1968).
Proc. R. Soc. A303, 339 (1968). 59. N A K A M U R A Y., T A K E D A Y. and S H I G A M.,
32. BOWER D. I . , C L A R I D G E E. a n d T S O N G I. S. T., J. phys. Soc. Japan 25,287 (1968).
Phys. Status Solidi 29, 617 (1968). 60. N A K A M U R A Y., S H I G A M. and S H I K A Z O N O
33. WEISS R. J., Proc. Phys. Soc. 82,281 (1963). N., J. phys. Soc. Japan 19, 1177 (1964).
34. G O O D E N O U G H J. B.,J. appl. Phys. 39,403 (1968). 61. LIN A., Ph.D. Thesis, U. of Toronto, Toronto,
35. B A I L Y N M.,Phys. Rev. 139, 1905 (1965). Ontario, Canada (1967).
36. B A S I N S K I Z. S., H U M E - R O T H E R Y W. and 62. S E D O V V. L., Soviet Phys. JETP 15, 88 (1962).
S U T T O N A. L., Proc. R. Soc. A229, 459 (1955). 63. U M E B A Y A S H I H. and I S H I K A W A Y., J. phys.
37. K O H L H A A S R., D U N N E R P. H. and S C H M I T Z - Soc.Japan 21, 1281 (1966).
P RAN G H E N., Z. angew. Phys. 23,245 (1967). 64. I S H I K A W A Y. and E N D O H Y., J. phys. Soc.
38. P E A R S O N W. B., In A Handbook of Lattice Spac- Japan 23,205 (1967).
ings and Structures of Metals and Alloys, V. 1 (1958), 65. H A S H I M O T O T. and I S H I K A W A Y., J. phys.
V.2 (1967). Pergamon Press, Oxford. Soc. Japan 23, 213 (1967).
39. A B R A H A M S S. C., G U T T M A N L. and K A S P E R 66. KOU V EL J. S., J. Phys. Chem. Solids 24,795 (1963).
J. S., Phys. Rev. 127, 2052 (1962). 67. J A C O B S 1. S. and BEAN C. P., In Magnetism
40. N A T H A N S R. and P I C K A R T S. J.,J. Pizys. Chem. (Edited by G. T. Rado and H. Suhl), Vol. 3. Academic
Solids 25, 183 (1964). Press, New York (1963).
41. K O U V E L J. S. and K A S P E R J. S., J. Phys. Chem. 68. M E I N C K E P. P. M. and L I T V A J., Phys. Lett. 29A,
Solids 24,529 (1963). 390 (1969).
42. ISH1KAWA Y. and E N D O H Y., J. phys. Soc. 69. F L E T C H E R R., J. Phys. C., Ser. 2, 2, 2107 (1969).
Japan 23,205 (1967). 70. K O N D O R S K I I E. I. and S E D O V V. L., Soviet
43. F U J I M O R I H. and S A I T O H., J. phys. Soc. Japan Phys. JETP 35,586 (1959).
20,293 (1965). 7 I. SCH LEN KER C., Phys. Status Solidi 28,507 (1968).
44. F A L L O T M.,Ann. Phys. Paris 10, 291 (1938). 72. H A N S E N M., In Constitution of Binary Alloys, 2rid
45. C R A N G L E - I . , Phil. Mag. 5, 335 (1960). Ed. p. 696, McGraw-Hill, New York (1958).
46. C A B L E J. W., W O L L A N E. O. and K O E H L E R 73. H A N S E N M., In Constitution of Binat3, Alloys, 2rid
W. C., Phys. R ev. 138, A755 (1965). Ed. p. 698, McGraw-Hill, New York (1958).
47. K U S S M A N A. and G R A F I N v. R I T T B E R G G., 74. K U S S M A N N A., A U W A R T E R M. and G R A F I N
Ann. Phys. (Leipzig) 7, 173 (1950). v. R I T T B E R G G.,Ann. Phys. Leipzig 6, 174 (1948).
48. G O O D E N O U G H J. B., in Magnetism and the 75. M A T S U M O T O H. and K O B A Y A S H I T., Sci.
Chemical Bond, Ch. 111, Section 111, Interscience Repts, RITU 2,856 (1950).
and John Wiley, New York (1963). 76. M A T S U M O T O H. and K O B A Y A S H I T., Trans.
49. C H I K A Z U M I S., M I Z O G U C H I T., Y A M A - JIM6, 113 (1965).
G U C H I N. and B E C K W l T H P., J. appl. Phys. 39, 77. M A T S U M O T O H., S A I T O H. and K O B A Y A S H I
939 (1968). T., TransJIM4, 114(1963).
50. N E W K I R K J . B., Trans. AIME209, 1214(1957). 78. K U S S M A N N A. and JESSEN K., J. phys. Soc.
51. G O N S E R U., M E E C H A N C. J., M U I R A. H. and Japan 17, Suppl. B-I, 136 (1962).
W l E D E R S I C H H.,J. appl. Phys. 34, 2373 (1963). 79. BELOV K. P. and S I R O T A Z. D., Soviet Phys.
52. H A N S E N M., In Constitution of Binary Alloys 2nd JETP 36, 752 (1959).
Ed. p. 677, McGraw-Hill, New York (1958). 80. T A K A K I H., N A K A M U R A Y. and N A K A J I M A
53. K N E L L E R E., In Magnetism and Metallurgy T.,J. phys. Soc. Japan 17, Suppl. B-I, 349 (1962).
(Edited by A. E. Berkowitz and E. Kneller), Vol. 2,
p. 549. Academic Press, New York (1969).

You might also like