24.12.2014 Views

Set of supplementary notes.

Set of supplementary notes.

Set of supplementary notes.

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 10<br />

Fermi liquid theory<br />

10.1 The problem with the Fermi gas<br />

Our modelling <strong>of</strong> electrons in solids so far has been based on a fairly simple-minded approach:<br />

instead <strong>of</strong> looking for the eigenstates <strong>of</strong> the many-particle system we are interested in, we<br />

instead calculate the electronic eigenstates <strong>of</strong> a single-particle Hamiltonian (subject to the<br />

periodic lattice potential). We then fill these eigenstates with electrons according to the Fermi<br />

occupation factor, treating our system as a degenerate Fermi gas. This separation <strong>of</strong> a manyparticle<br />

problem into single-particle states relies on being able to separate the many-body wave<br />

function into an antisymmetrised product <strong>of</strong> single-particle wavefunctions. For a two-electron<br />

state, for instance, we could write Ψ(r 1 , r 2 ) = √ 1<br />

2<br />

(ψ a (r 1 )ψ b (r 2 ) − ψ a (r 2 )ψ b (r 1 )). Such product<br />

states, however, implicitly ignore correlations between the electrons. For example, we would find<br />

that expectation values <strong>of</strong> products such as 〈r 1 r 2 〉 decompose into the products <strong>of</strong> expectation<br />

values 〈r 1 〉 〈r 2 〉.<br />

In the presence <strong>of</strong> strong electronic interactions, it is doubtful that the electrons’ motion<br />

remains uncorrelated. We saw last term that in the Thomas-Fermi approximation, the electrons<br />

in a Fermi gas react to the introduction <strong>of</strong> a charged impurity, in such a way that they screen<br />

the impurity potential at long distances. Taking this idea to the next level, we could say<br />

that for any one electron under consideration, which <strong>of</strong> course carries a negative charge, the<br />

other electrons in the metal execute a correlated screening motion, which would reduce their<br />

density in the vicinity <strong>of</strong> the first electron. This reduces the range <strong>of</strong> the Coulomb potential<br />

due to the electron under consideration, which we might take as justification for ignoring the<br />

Coulomb interaction. However, this also implies that the electrons correlate their motion. Such<br />

correlations are not contained in a single-particle description.<br />

On the other hand, the band structure approach which we have used so far has been remarkably<br />

successful in modelling a wide range <strong>of</strong> materials and phenomena. It explains electronic<br />

transport and thermodynamic properties such as the heat capacity in metals, we have used it<br />

to understand semiconductors and semiconductor devices, and it is consistent with the Fermi<br />

surface probes such as quantum oscillations in high magnetic fields, as well as other probes<br />

<strong>of</strong> electronic structure. These successes suggests that it is not altogether wrong. How can we<br />

reconcile the success <strong>of</strong> the single particle picture with its conceptual difficulties<br />

145

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!