You are on page 1of 1

8.

4 Instability of the Nagaoka State 443

empty. The energy density is simply

(8.10)

The reason why the Nagaoka state may become unstable, strikes
the eye: it appears that a substantial energy lowering could be achieved
by transferring an electron from the top of the ?-spin Fermi sea to the
bottom of the &spin band, because this is accompanied by the kinetic
energy gain AEo = f& - EF. The corresponding uncorrelated (Hartree-
Fock) single-spin-flip state is

(8.11)

where the wavevector q belongs to the band minimum2*. There is,


however, a substantial energy penalty coming from the Hubbard term,
because in an uncorrelated state, the $-electron cannot avoid the -f-
electrons, thus the spin-flip energy at the Hartree-Fock level is

- ENag = Eb - €F
AEHF= (@I%[@) +U n , (8.12)
wherewith we rederived (8.9). We can introduce the Hartree-Fock esti-
mate for the Nagaoka threshold

(8.13)

The ground state is fully polarized if U 2 Ul:g. Let us observe that


in the independent electron approximation, the U = 00 ground state is
predicted to be fully polarized for all N < L.
However, AEHFshould not be equated with the true energy differ-
ence
A E = Es=N/s-I - ENag (8.14)
between the S = N / 2 and S = ( N - 2)/2 ground states because, in
reality, electron correlations play an important role, enabling the -1 -
electron to get out of the way of the f--electrons, and thus reducing
20There may be several degenerate band minima and, in any case, the choice of
the Fermi wavevector kF is not unambiguous. Here we proceed with the argument
assuming that it is irrelevant which of the possible kF and q is chosen.

You might also like