You are on page 1of 1

380 Ch.

7 Itinerant Electron Magnetism

selves to the single-orbital Hubbard model, it is justified to use the same


effective Heisenberg model on both sides of the phase boundary, and the
phase transition of the electron system can be modelled as the phase
transition of a spin system25.This approach can be used with confidence
for systems which are wildly Mott insulating like COO. In contrast, the
collapse of a weak-coupling SDW order leaves behind a weakly corre-
lated Fermi sea. At intermediate couplings, we would predict a Fermi
sea whose susceptibilities (especially the staggered susceptibility) are
(perhaps strongly) enhanced. All the same, the SDW-to-paramagnetic
transition is still an insulator-to-metal transition. At a still somewhat
bigger U (around U W ) the scenario changes because we get on the
N

other side of the Mott transition: the PM phase becomes insulating.


Thus there must be a finite-temperature metal-insulator phase bound-
ary within the paramagnetic state, which meets the SDW critical line
in the intermediate coupling regime (Fig. 7.6).
We were arguing that in the ground state, the perfect nesting prop-
erty causes a unified behaviour from U = +O to U = 00; there is
continual change but no sharp boundary. We feel that the weak-U sys-
tem is merely a weak SDW, while the large-U system must be really
a Mott insulator but remaining at T 5 T N ,we cannot pinpoint where
the transition occurs. The situation is quite different at temperatures
T > TN where there is a sharp PM-PI phase boundary. This is a line
of first-order phase transitions26. The PM and PI states have the same
spatial symmetry, do not break the spin-rotational and time-reversal in-
variance, thus they can differ only27 in density-like characteristics: the

25Allowingfor degenerate models may bring extra complications due to the orbital
degrees of freedom. We have seen in Sec. 5.4 that the interplay with orbital ordering
may lead to situations where different Heisenberg models have to be used for the AF
and the P M phases.
“We certainly would expect such a boundary for realistic models which incorporate
long-range Coulomb interaction, and/or electron-phonon coupling. For the pure
Hubbard model, we have to repeat the warning given in the discussion of Fig. 5.5: it
is possible that the model cannot yield discontinuous behaviour.
”If we mean it as a general statement, it is not true that if a transition does
not change symmetries then it cannot be a higher-order transition; the Kosterlitz-
Thouless transition is a counterexample, and even apart from that, we have to worry
about the possibility of hidden characteristics which are not macroscopic order pa-
rameters in the conventional sense. However, our simpleminded argument is good

You might also like