This document discusses the ground state phase diagram of the Hubbard model at different electron fillings. It argues that the system should be metallic for any electron filling between 0 and 2, as the strong Hubbard U does not inhibit electron hopping to neighboring empty sites. However, near filling of 1, disorder effects may localize carriers and make the system insulating. Therefore, the Mott transition at filling of 1 remains relevant despite the system being metallic at other fillings.
This document discusses the ground state phase diagram of the Hubbard model at different electron fillings. It argues that the system should be metallic for any electron filling between 0 and 2, as the strong Hubbard U does not inhibit electron hopping to neighboring empty sites. However, near filling of 1, disorder effects may localize carriers and make the system insulating. Therefore, the Mott transition at filling of 1 remains relevant despite the system being metallic at other fillings.
This document discusses the ground state phase diagram of the Hubbard model at different electron fillings. It argues that the system should be metallic for any electron filling between 0 and 2, as the strong Hubbard U does not inhibit electron hopping to neighboring empty sites. However, near filling of 1, disorder effects may localize carriers and make the system insulating. Therefore, the Mott transition at filling of 1 remains relevant despite the system being metallic at other fillings.
ing. Previously, we have convinced ourselves that at half-filling ( n = l),
a Mott-transition should set in at some critical value (U/t)cr, so the sys- tem is metallic for ( U / t ) < (U/t)cr, and insulating for ( U / t ) > (U/t)cr. What happens at n # l? Let us make the plausible assumption that if the system is metallic at U / t = 00 then it is metallic at any finite U / t . It is easy to argue that the ground state should be conducting for any 0 < n < 1 and 1 < n < 2. For less electrons than sites N < L (or n < l), a large U forbids double occupation, so some sites are empty (positively charged), while others have one electron (neutral sites). It looks plausible25 that the strong Hubbard U does not inhibit the jumping of an electron to a neighbouring empty site: n (4.60) The process can be looked upon as the hopping of the empty site which is a charge carrier. This costs no energy thus the system is a metal. This finding may raise doubts about the relevance of arguments for the Mott transition as an observable phenomenon. Since n can vary continuously between 0 and 2, it would seem that the point n = 1 is a set of zero measure, utterly without interest. This would be a false conclusion. On the one hand, we have seen (Fig. 4.6) that the filling tends to be pinned near integer values of n, thus n = 1 is not so unexpected. On the other hand, for n sufficiently near to 1, disorder effects (Anderson localization) come into play and make the system a non-conductor by localizing the few (- In - 11) carriers implied by our arguments. As a matter of experimental fact, Mott insulators (or if you like, Mott-Anderson insulators) are robust, and their nature can be largely understood by studying the case n = 1. Next: which kinds of ordered ground states are possible? Expressly avoiding the thorny question whether the positive4 Hubbard model 2'Actually, there is a subtle difficulty here. While it is true that we cannot identify an activation energy of the order U ,the hopping of an empty site on the background of spins may leave a trail of disturbed spin arrangement, so it does not look like the free propagation of a hole, either. The argument works simply if we assume that all spins are t.