You are on page 1of 1

292 Ch.

6 Heisenberg Magnets

It would be difficult to miss observing A S in quasi-two-dimensional


low-spin antiferromagnets. D = 2 is an intermediate case in the sense
that both the vicinity of k = 0 and the outer parts of the Brillouin
zone give comparable contributions to the integral (6.90). Spin wave
theory predicts A S M 0.197 for the square lattice. That leaves us with
only -60% of the classically estimated order parameter for an S = 1/2
antiferromagnet! The case is of great interest because it closely cor-
responds to LaaCu04. However, let us not forget that our spin wave

-
theory amounts to keeping the lowest-order contributions in a l/S ex-
pansion, thus finding AS/S 0.4 in lowest order may have led us to
suspect that in higher orders A S is even larger, or maybe long range
order is destabilized completely. A great variety of theoretical tech-
niques have been brought to bear upon this problem and the upshot is
that the simplest estimate is basically right 1260). It is interesting that
m,t M 0.3 is essentially confirmed by experiments, thanks to the fact
that in LazCuO4, potential disturbing effects are almost entirely absent:
the interlayer coupling is very weak, and anisotropy effects apparently
small. The situation is discussed in [260].
There are two-dimensional lattices with coordination number z = 3
and we expect that for these, fluctuation effects are even stronger. A
honeycomb lattice S = 112 antiferrornagnet looses -56% of its classical
sublattice magnetization: m,t M 0.22f0.03 both from spin wave theory
and Monte Carlo studies [333]. A different z = 3 lattice is the idealized
structure of the quasi-two-dimensional S = 1/2 Mott insulator CaV409,
which we show in Fig. 6.4 (top left) (see p. 327). Notice that inequiv-
alent bonds have the differing coupling constants J1 and Jz. If we take
these as equal16, spin wave theory gives mStx 0.228 [4121. We mention
that for these lattices, the lack of inversion symmetry would make the
formal development of spin wave theory slightly dissimilar to what we
have done before.
We have found that even the most wildly fluctuating bipartite-
latticei7 two-dimensional antiferrornagnets have sublattice polarizations
mSt > 0.2, while (isotropic) D = 1 antiferrornagnets do not order at all.
"But in real life, they are rather different, and CaV409 is not an ordered antifer-
romagnet. The situation is discussed in Sec. 6.3.3.
"The case of non-bipartite lattices is a quite different story. See Sec. 6.3.4.

You might also like