J. Phys. Chem. Solids Perpamon Press 1959. Vol. 11. pp. 73-77. Printed in Great Britain.

PIEZOMAGNETISM Ir\; CoF,

T. MORIYA*
Bell Telephone Laboratories, Murray Hill, New Jersey

(Rectived 11 February 1959; e-wised 13 March 19593

Abstract-It is pointed out that in antiferromagnetic iron-moup difluorides, MnFz, FeFs and CoFz,
a linear compression in the &-plane can produce magnetization along the c-axis. The magnitude of
this piezomagnetic moment is evaluated theoretically for CoF2, in which the largest effect is expected.
The estimated order of magnitude of the magnetization is a few tenths cgs under a pressure of 100
kR.‘cm2 along the [l lo] direction.

1. INTRODUCTION of site are no longer the same, even if we allow for a

THE possible existence of piezomagnetism in some rotation around the c-axis by 90”. The magnetic
antiferromagneticcrystals has recently been pointed moments of the two kinds of spin in the exchange
out by DZIALOSHIKSKI~~) from considerations of field should in general differ owing to the different
crystal symmetry. He mentioned as examples anisotropic fine-structure couplings and the differ-
a-Fez03 and MnFz, FeF2 and CoF2. In these ent g-factors. Therefore, net magnetization will
fluorides, he states, shear strains uZt, uyZ will be
accompanied by magnetization along the y, x
directions respectively, though he does not give
their orders of magnitude.
The purpose of the present note is to point out
that in MnF2, FeF2 and CoFz even a linear com-
pression in the &plane, say along the [l lo]
direction, can produce magnetization along the
c-axis, and to estimate its order of magnitude in
the case of CoF2, in which the piezomagnetic
moment is expected to be especially large.
The crystal structures of iron-group difluorides
are of the rutile type.@) There are two kinds of ca-
tion sites, the corner and the body-center sites, in
FIG. 1 The rutile-type crystal structure of CoFs. Solid
these crystals, as shown in Fig. 1. The crystalline
circles are cations, open circles anions.
fields around these two kinds of site differ only in
that their principal axes, x and y, are interchanged.
arise iust as in ferrimagnets, provided that the lifc-
In their antiferromagnetic states, the ions on one
time-for the interchange oi the + and - sub-
kind of site form one sublattice, say of up spins,
lattices is sufficiently long, as is actually the case.
and those on the other kind of site form another
This effect will be most significant in CoF2, less
sublattice of down spins. The sublattice magneti-
so in FeF2, and least in MnF2. In MnF2 magnetic
zations are along the f c-directions.
anisotropy is much smaller than in the other fluor-
When the linear compression in the &plane is
ides, and it arises mainly from dipole-dipole inter-
applied, the crystalline fields around the two kinds
action between spins; the crystalline electrical
-- ._
l On leave of absence from Tokyo Metropolitan Uni-
field contributes only 10 per cent to the aniso-
versity. tropy.(s) Piezomagnetism will thus be smallest in
73
74 T. MORIYA

this substance. Tn FeFz the electric splitting of spin where i = 1,2 represents ions on the + and -
levels is comparable with the exchange interaction, sublattices, respectively, and
and in CoFa the former is even larger than the
hl -- 2J_~jSs))
latter. In this situation, the sublattice magnetiza- ) (21
tion depends to a considerable extent on the elec- h2 z-72J,_ (Sl ),
tric splitting. In fact, theoretically the sublattice
J being the exchange integral, z the number of
spin value in CoFe is l-19 at O°K,f~) which is con-
nearest magnetic neighbors. The definition of the
siderably smaller than 32. On the other hand the
co-ordinate axes are shown in Fig. 2 for i 7 I. The
sublattice spin value in FeFs is estimated to bc
eigenvalues of equation (1) arc easily obtained as
1.99, very near to 2, at O”Ii.@) This fact seems to
follows :
suggest that piezomagnetism of the type WC are
considering here is much larger in CoF;? than in &Ci)=. - ~oi+~hl-{((Di-/li)“+3E*‘)’
FcFs.
&:,cO._z - ;ui -3frt+((r)i+hf)‘+3E~2~~
_
I
(3)
2. PIEZOMAGNETISM IN ANTIFERROMAGNETIC f:‘,(i) - :Dt+~lfj+{(Dj--lri)“+3~~z3a
CoFe
‘1theoretical study of the magnetic anisotropy
[,;(a - -,~l)l-g/zj-~(Dl+hi)“+3~~~~~. I
of CoFa has been made by NAKAMURA and ‘Ili’e know that L)l > 0, De > 0 and we assume
TAKETA~~).The experimental results@) seem to be without loss of generality hj < 0, hz > 0, El :* 0,
we11explained, at least semiquantitative~y, by their Ez < 0. Then we see from inspection of equation
theory.* Here we shalI adopt their model, i.e. we (3) that the lowest state energies of ions 1 and 2 are
shall assume that we can safely take a spin Hamil- respectively
tonian that implies that the lowest excited orbital
state lies well above the ground orbital state, which El(l) -2 - ~D1+~I~1-{(f)l-hl):+3~~2}t \
j (‘+)
itself is not degenerate. Adopting the molecular- j32) _- - P~8-~h2--(1)2+IiZ):!+3~~:‘ze3:.
field approximatio~l, we have the following Hamil-
tonian for the spins on the two sublattices: The sublattice magnetizations at 0°K are obtained
as follows :
H(i) .-.: --nrSi,2-~~(S~,z--Siy2)+Sr
* hi, (1) .VJ’ ) Xl”’ VA--h)
-.: f+---____
.w I: 8JlII ((D1-/z1)*+3E,2}~
I
(9
fiffi, 8&Q, (&+k)
-- = -=:- .;- --- .--
:plf it/I2 ” {(/)z+/Z#+3&2fP i
In the normal state under no compression, where
Dl = D2, L:‘l . -& =. I:', (6)

we get

M~O'l) M~O'~'
- =; - ~ ;.
SO
$$I3 ‘wu

FIG. 2. Definition of co-ordinate axes for i = 1 sites.

For i = 2, the x- and y-axes are interchanged.
.- . .~__. -..---- ._._
* Paramagnetic resonance of Co2* in ZnFs has been Under a compression, the relations (6) no longer
observed by TINKHAM(?) and by STOUT@). They have
gixzn the g-values and the hyperfinc coupling constants
hold; we write instead
within the lowest Kramcrs doublet (s’ = l/2 manifold). <S$l)) :‘- so+ss,, (Sz(2)) = --sfJ+ss2,
XAKAMURAand TAKETA’S theory seems to be consistent
with these measured values.@) (8)
 PIEZOMAGNETISM IN CoF2 75

D1 = D+6D1, D2 = D+6D2, and partial covalent bonds. Though these iron-

(9) group difluorides may be regarded as more ionic
El = E+6E1, E2 = -E+6E2. 1
than oxides or hydrated salts, where the covalent-
6S~, SD,, 6Ef are regarded as linear functions of bond contribution seems to be predominant, the
strain or stress. Inserting equations (8) and (9) numerical calculation of the crystalline electrical
into (S), we get

3B2D( SD1;GD2)
-3E2(D+2JzSo)~sE1;s”)
ss,+ss2 = (10)
{(D+~JzS,,)~+~E”}~+~E~J~

There is another contribution to the piezomagnetic field based on the point-charge model did not give
moment, i.e. the change of g-factors due to com- a result which could explain the magnetic aniso-
pression. Writing tropy of this salt.(rs)
We are here compelled to content ourselves with
$32 = gz+& the following simple reasoning. Equation (14) con-
(11)
g2t = gz+Qz, 1 sists of three terms, of which the first one seems to
be the largest, since the orthorhombic character
we have as a net magnetic moment per pair of of the crystalline electrical field is essential in this
spins problem. The second and the third terms might
(%9-~g2)PR~O.
roughly be of the same order of magnitude, as we
(‘2)
see from the following considerations. According
Summing up the contributions (10) and (12) and to the crystalline-field theory, D, E and g are
using the following numerical values of parameters written as follows:
estimated by ~AKAMUR.4 and TAKETA:

Di= 17*9k, E = 36.2k, 2Jz = 34+k, D= 32*.--n.-*,),

so = 1.19, (13)
we get E=

SEI+SE~
6l%'f~+6i%f2 = g#LB -0.3 + 2(1-&),
E
6D1--SD2 (M~)(~lLlg), etc
+ 0.09 +1*19 y@] (14) *,
D
c
d
Ei-Eg

at 0°K. We can easily show that the influence of where A is the spin-orbit coupling constant, and g
the change of J, the exchange interaction, due to and i represent the ground and the excited orbital
compression is not important; it vanishes in the states, respectively. If we tentatively assume that
first-order effect. 6AZ z --6A,, we get
We have now only to estimate 6E, SD and Sg as
a function of strain or stress. However, to do this %z -Ai, SD
_,_._x?.;. (15)
is not an easy task at present because of the absence -l-M, I) gz
gr
of any clear understanding of the origin of the
crystalline electrical field in this crystal. We can The numerical value for CoFs is (gt-2)/gZ z 0.2.
consider the contributions to the crystalline elec- In order to get some idea on the order of mag-
trical field from distant charge, overlapping charge nitude, we shall take only the first term of equation
76 T. MORIYA

(14) and further assume bonds may make contributions, may show
stronger dependence on the interionic distance
6Ei 6Ai
than that calculated on the point-charge assump-
- =-=-, (16)
Ei Ai tion.

where Ai is defined by the following expression for

3. DISCUSSION
the crystalline field potential around the ion of the
We have shown in this note that in antiferro-
ith kind :
magnetic MnFz, FeFz and CoFz even a linear
V,(r) r Ai(~z-y*)+B~(3z*-v*)+ compression in the &-plane produces a magnetic
moment along the c-direction. This effect is con-
+(fourth-order terms).
sidered to be especially significant in CoFz. :\n
If we take account of the contributions to At from order estimation for CoFz shows that the piczo-
the six fluorine ions nearest to the Coz+ ion and re- magnetic moment of this type is considerable.
gard them as point charges, we can easily cal- This effect will be observed by direct measurc-
culate GAJAf as a function of strain or stress. We ments of the piezomagnetic moment if the crystal
shall show the result when a linear stress is applied has only one domain or a small number of do-
along the [l lo] d irection, the x-axis in Fig. 2. mains. Fluorine and/or cobalt nuclear resonanceor)
Assuming uniform deformation in the crystal, WC will provide a sensitive method for detecting the
get possible unbalance of the sublattice magnetizations.
This method will not be affected by the cxistencc
SAi
,1- = (-15~.~-0.3sr~+13ss~)7’, (17) of domains. Optical Faraday rotation(t*) and the
L method of observing domains by using super-
conducting (niobium) powderus)+ may make it
where sij’s are elastic compliance factors. Further
possible to see antiferromagnetic domain pattcrns,
assuming, though not necessarily correctly from
or to see whether domains really exist or not in
the microscopic point of view,
antiferromagnetic CoFs. In conclusion, we would
sr* = szr, s31 = s32, (18) stress that the piezomagnetic moment will be re-
versed in sign when the antiferromagnetic sub-
lattice magnetizations are reversed in sign. This
sAl+W? suggests the possibility of measuring the relaxation
- ..----. .- 15(Srr-Srs)T, (19)
A time for the interchange of the sublatticc mag-
netizations in an antiferromagnet.
where the relation Al = - Az = A was used.* We
estimate therefore, taking srr N 10-s cm*/kg,
Acknoaledgenzenfs-The writer wishes to thank Dr.
P. W. XNI)ERSON for reading the manuscript and for kind
advice. IIe has also benefited from conversations with
z -5 x lo-sg,pBT(in kg/ems). (20) Drs. R. M. BOZOHTH, A. M. CLOGSTON, J. E‘. DILLOS,
V. JACCAHISO and A. I,. SCHAWL~W.
Under a compression of 100 kg/ems, wc get a mag-
netic moment of 0.6~ 10-s Bohr magnetons per
ions, which corresponds to a magnetization of REFERENCES
M = 0.2 cgs. 1. DZIALOSHINSKI I. E., Zh. Eksper. Teor. Fiz. 33, 807
Finally, we should mention that the actual mag- (1957); (translation) Sov. Phys. JETP 6, 621
(1958).
nitude may be larger than the above-estimated
2. KACAWYA T., YOSIDA K. and KUBO R., Adrtances
value, because the actual crystalline field potential, in Phys. 4, 1 (1955).
to which overlapping charges and partial covalent 3. &FFER F., Phys. Rev. 88, 608 (1952).
-_
* (6Bi - SBz)!B was calculated in the same way, and t By applying an external magnetic field of adequate
the value is -6(srr --s12)T. M’e see, therefore, from equa- magnitude, we can make the magnetic field on the sur-
tion (14) that the second term is one order of magnitude face of one kind of domains larger than H,, while that on
smaller than the first term and of the same sign. the surface of another kind of domains smaller than If?.
 PIEZOMAGNETISM
4. NAKAMURAT. and TAKETA H., Pmgr. Theor. Phys.
13, 129 (1955).
5. HOMMA A., Busseirm-kmkyu 4, 358 (1958).
6. STOUT J. W. and ~~ATARRESE
L. M., Rev. Mod. Phys.
25, 338 (1953).
7. TINKHAM M., Proc. Roy. Sot. A236, 53.5 (1956).
8. STOUT J. W., private communication.
IN CoFa 77
9. TOSHIMA S. and NAKAMURAT., to be published.
10. MORIYA T., MOTIZUKI K., KANAMORI J. and
NAGAMIYAT., J. Phys. Sot. Japan 11, 211 (1956).
11. JACCARINOV., Phys. Rev. 107, 1196 (1957).
12. DILLON J. F., J. Appl. Phys. 29, 539 (1958).
13. SCHAWLOWA. L., MA~HIAS B. T., LEWIS H. W.
and DEVELIPI;G. E., Phys. Rev. 95, 1344 (1954).