7.

5 Generalized Susceptibility 357

ferromagnetism at sufficiently large U. Clearly, this cannot be the whole

story because we have already convinced ourselves that near half-filling,
the Hubbard model prefers antiferromagnetism. We have t o go on to
formulate instability criteria for the other kinds of magnetic ordering.
Problem 7.1. Temperature effects in the spin susceptibility:
a) calculate the lowest-order finite-temperature correction to the Pauli sus-
ceptibility. Does it decrease or increase as you increase the temperature from
T = O?
b) Discuss the large-T behaviour of the spin susceptibility for a half-filled
band with a symmetrical density of states.
c) Devise a DOS form for which the susceptibility has a marked maximum
as a function of temperature. Discuss the effect of switching on an interaction.
Problem 7.2. Finite-m predictions of the Stoner theory:
a) Derive the T = 0 Stoner criterion for finite m.
b) Consider the double-peak model DOS used in the previous problem (see
p. 407). Increasing U from zero, which kind of Stoner instability do you find:
is m increasing continuously from m = 0, or does it jump immediately to a
finite value?
Problem 7.3 Stoner theory should work best in the limit of very weak itinerant
ferromagnetism where Tc << TF (ZrZnz is thought to realize this limit; see
p. 396 for discussion). Here we work out some of the finite-tempereture results:
a) Use the result of Problem 7.1 to calculate the Curie temperature.
b) Assume that p(e) = po - m2,and the band is half filled. Derive the
temperature dependence of the order parameter.
c) For the same model, derive the susceptibility for T > Tc.

7.5 Generalized Susceptibility

Now we want to find out whether the Hubbard model is unstable against
ordering into a spin density wave of the form (7.7). The instability is
signalled by a divergent response t o an external magnetic field with
wavevector q. The full Hamiltonian reads

(7.28)

In first-quantized form, the coupling of the spins to the field is