Quantum treatment

Excitations- lower the spin once

Si
|i =
2S

e.g.
S=1/2

|Sjz

= S =

|Siz

= S 1

j=i

|Sjz = S

Quantum treatment
Hamiltonian

H = J

Siz Sjz

ij

1 + 1 +
+ Si Sj + Si Sj
2
2

H z + H

zz terms

Hz |i = E0 Jz[S(S 1) S ] |i = (E0 + JSz) |i

Quantum treatment
+- terms

J +
+
H =
Sj Sk + Sj Sk
2

jk

J +
H |i =
Sj Si |i
2
ji

J
= 2S
|j = JS
|i +
2

ji

Quantum treatment
All together

H|i = (E0 + JSz) |i JS

Looks like a hopping Hamiltonian

Bloch:
1
|k =
N

H|k =

|i +

eikri |i

E0 + JSz JS

E0 + (k)

eike

|k

Quantum treatment
These are spin waves or magnons

(k) = JSz JS

= 2JS

JSa2 k 2

Magnons are gapless

eike

=1

cos k a

quadratic for small k

Magnons

Magnons are gapless

Energy vanishes as k 0
This is because

k/a
3

|k = 0 STOT
|0

This is an example of a Goldstone mode

Goldstones theorem: if H has a

continuous symmetry that is broken by
the ground state, there will be a gapless
mode

Magnons
Spin wave/magnon can be regarded as a

quantized precession wave of slightly tilted

spins

~ k2 behavior, a general property for

ferromagnets, can be understood this way

Magnons
Think of effective field due to other spins
h(r) c0 m + c1 2 m +

Local spins precess in this field

t m = h m c 1 2 m m

Magnons
t m = h m c 1 2 m m
m = (mx , my ,

mx
my

m20 m2x m2y )

= c1 m0

(mx , my , m0 )

0
k2

= c1 m0 k 2

k
0

mx
my

Neutron scattering
Neutron has a S=1/2 similar to an electron,

and its own dipole moment, which interacts

with magnetic dipoles in materials
Hdd

=
[3(m

r)(m

r)

m
]
3
4r

Consequently, a neutron can exchange

energy and momentum with electronic

spins
k=ki-kf,=i-f
ki,i

kf,f

CrBrs

IN
q =, I043

Xb

46

=. 2589

I20
O
O

40-

80

161

"

viously been observed in the ferrimagnet Fe304'

n-Fe, O, . One should
and the antiferromagnet
notice the way in which the optical-branch data in
Fig. 6arepresented, ~amely, at the points in reciprocal space relative to 111 where they were
actually observed. The reason is that, if folded
back to the 111 BZ, they would not include the
points l" and Z. The optical branch is not directly
observable in the BZ surrounding 111 because of
an almost vanishing structure factor, and when
observed at the 001 or 110 reciprocal-lattice points
'*
with our samples, the peaks are not "clean, as a
consequence at the a, b twinning.
No second branch due to domain II was seen. Calculations show that its dynamical structure factor
nearly vanishes along the whole b, direction. The
two branches, in fact, cross at two-thirds distance
to the zone boundary, but the structure factor allows only one continuous branch to be seen.

Neutron scattering
\

40

00

20

.2

300

20'K

O~

200

IQQ

Integrate over all energies: get total scattering

0

.2

.2

.3

ENERGV, meV

FIG. 4. Intensity profiles for low-energy spin waves

of the same energy along Z, 6 and along A at 4. 6 'K and
20'K. The Z, 4 peaks are well focused, giving rise to
a high accuracy of the spin-wave energy. The renormalization of energy with temperature is demonstrated.
Calculated profiles are shown, and arrows indicate the
nominal energy of the spin waves entering the folding
program (Ref. 16).

VI. LOW-ENERGY SPIN WAVES

tool for detecting magnetic ordering in solids

The spin waves that are important thermodynamically in determining the magnetization deviation at
a given temperature are, of course, those with an
energy comparable to the thermal energy of the
system. For temperatures below 20 0K this will
imply spin waves of energy less than about 3 meV.
The observed dispersion relation for some of those

Resolve both momentum and energy of

neutrons: measure spin waves
tion associated with the low incident energy of 5. 2
MeV.
Some corrections due to instrumental resolution,
effects are sometimes required of the neutron
spectrometer data. ' The corrections are normally
of significance only for systems with steep dispersion relations and for nonfocused peaks. Some
calculations were performed using a previously
described program" to fold the instrumental resolution function with the scattering cross section. They
revealed that for the high-energy end of the spectrum
some minor corrections were needed for a few
cases, the corrections always amounting to less
than the experimental uncertainty.
The curves
shown in Figs. 3 and 4 are the calculated line

shapes.

6.
Figs.
example of spin
waves in Yb2Ti2O7
(2012) refers

The observed dispersion relations are shown in

important7'
The thermodynamically
5 and
low-energy parts are given in Fig. 5, including
the entire acoustical branch in the A direction.
Figure 6 shows the data taken in directions normal
to A. The labeling of the directions and points
shown
to Fig. 2. The three branches shown
were assigned to domains I and II as indicated,
based upon their apparent periodicity. Since CrBr3
contains two Cr ' ions per unit cell, an optical

3.0

2.0
E
IJJ

LLJ

I.O

.2

.3

.5

0
WAYEYECTOR IN A

first measurement in
an insulating FM:
CrBr3, 1971

FIG. 5. Dispersion relation of spin waves of low energy

as measured in various directions at 6 'K, all shown on
the same abscissa. The labeling corresponds to that of
Fig. 2. Where no error bars are given, the uncertainties
are less than the size of the symbols. Notice that the
F, B direction has its origin at the Z point of the Brillouin
zone. This way of presenting the data is chosen (i) to
show the large difference of stiffness along the c axis
and in the basal plane and (ii) to demonstrate the difference between
&

Antiferromagnets
Actually it is much more common to have
interactions that favor anti-aligned spins
H = +|J|

ij

Si Sj

This is trickier, because even classically

spins take on different orientations

Nel state

unfrustrated lattice:
can minimize all pairs of
interactions simultaneously

Antiferromagnets
Actually it is much more common to have
interactions that favor anti-aligned spins
H = +|J|

ij

Si Sj

This is trickier, because even classically

spins take on different orientations

frustrated lattice:
spins must compromise

Unfrustrated case
For example, square lattice
n = Si (1)xi +yi

In MFT, find staggered magnetization n

behaves the same as m for a ferromagnet

But quantum situation is different!

Unfrustrated case
Guess the ground state?
| =

NO!!
Is it an eigenstate?

|J|
H =
2

|Siz = (1)xi +yi S

jk

= J

+ ...

Sj+ Sk

Sj Sk+

Singlets

For two spins, the ground state is actually a

singlet
1
|s = (| | )
2
This can be viewed as a quantum superposition of
two classical Nel states
Consquently, in a real antiferromagnet, the
magnetic order will be reduced or even removed
by quantum fluctuations below that of MFT

However, it turns out that most

antiferromagnets still manage to order

time

A brief history of
magnetism
~500BC: Ferromagnetism
documented in
Greece, India, used in
China

1949AD:Antiferromagnetism
proven experimentally

sinan, ~200BC

time

A brief history of
magnetism
~500BC: Ferromagnetism
documented in
Greece, India, used in
China
Why so long???
1949AD:Antiferromagnetism
proven experimentally

sinan, ~200BC

A debate
H = |J|

Nel

ij

Si Sj

Landau

antiferromagnet

1
2

singlets

The right tool...

Neutron scattering
LETTERS

TO THE

sitive. This last condition is required in order that spin

gh multiplicity, which favor ferromagnetism, have the
gy. It seems certain that for many of the non-ferroubstances containing a high concentration of magnetic
xchange integrals are negative. In such cases the lowest
e is the one in which the maximum number of antirs occur. An. approximate theory of such substances
developed by Neel, I Bitter, and Van Vleck3 for one
e and the results are briefly described below.
a crystalline structure which can be divided into two
lattices such that atoms on one lattice have
ing
ghbors only on the other lattice. Examples are simple
body-centered cubic structures. Let the exchange
r nearest neighbors be negative and consider only
ighbor interactions. Theory then predicts that the
will exhibit a Curie temperature. Below the Curie temhe spontaneous magnetization vs. temperature curve
the sub-lattices is that for an ordinary ferromagnetic
directions for the two
However, the magnetization
antiparallel so that no net spontaneous magnetization
absolute zero all of the atoms on one lattice have their
magnetic moments aligned in the same direction and
atoms on the other lattice have their moments antithe first. Above the Curie temperature the thermal
sufficient to overcome the tendency of the atoms to
allel and the behavior is that of a normal paramagnetic

EDITOR

(3I I)

(33I)

IOO

125'7

(Sl l)(333)
NETIC UNIT CELL

e, = 885K

Shull and Smart, 1949

40

20
0
t
(IOO)

IOO

(IIO)

(III} (200) (2IO} (2ll)

(22C)

tttt
(310)

N22)

(300) (3I I)

CHEMCAL UNIT CELL,

80

a, 4. 434
=

80

20

M SAMPI

MPURITY
I

30'

20'
COUNTER

40'

50'

ANGLE

. 1. Neutron

Now we know antiferromagnetism is

diffraction patterns for MnO at room
temperature and at 80~K.

In conclusion it appears that neutron diffraction studies of antiferromagnetic materials should provide a new and important
method of investigating the exchange coupling of magnetic ions.

commonplace

exhibiting the characteristics described above have

ated "antiferromagnetic.
Up to the present time the
experimentally
ods of detecting antiferromagnetism
indirect, e.g. , determination of Curie points by suscepspecific heat anomalies. It has occurred to one of us
at neutron diKraction experiments might provide a
In an antiferrons of detecting antiferromagnetism.
material below the Curie temperature a rigid lattice of
ons is formed and the interaction of the neutron magent with this lattice should result in measurable co-

"

Fi(

* This

work was supported in part by the ONR.

physique l7, 5 (1932).
Rev. 54, ?9 (1938).
' J. H. Van Phys.
Vleck, J. Chem. Phys. 9, 85 (1941).
4 O. Halpern
and M. H. Johnson, Phys. Rev. 55, 898 (1939).
' Whittaker, Beyer,
and Dunning, Phys. Rev. 54, 771 (1938); Ruderman,
Havens, Taylor, and Rainwater, Phys. Rev. 75, 895 (1949); and also
work
Oak
unpublished
at
Ridge National Laboratory.
II Bizette,
Squire, and Tsai, Comptes Rendus 207, 449 (1938).
' B. Ruhemann,
Physik. Zeits. Sowjetunion 7, 590 (1935).
~

L. Noel, Ann. de

F. Bitter,

Landaus dream
People are still looking for the quantum

disordered antiferromagnet, with a singlet

ground state instead of a Nel one