Kondo effect


First, we have to understand what leads to electrical resistance in pure metal. Then we will see what happens if you add magnetic impurities to such a metal.

The fact that electrons interact only weakly with each others and that they can essentially be treated as a gas of free fermions, allows to describe the electrical transport in a metal using a Boltzman theory for the transport, where there are scattering effects from various interactions.

Now the electrons will scatter on impurities or imperfections in the lattice so if the temperature is very low, it is the scattering of the impurities in the lattice which limits the electrical transport to a fixed value.

In addition, when the temperature is raised, the ions in the lattice will move from their ideal position in the lattice, which breaks the translation invariance of the lattice. These motions are known as phonons and the breaking of the periodicity of the lattice leads to additional scattering contribution via this interaction of the electrons with the phonons.

This was captured in the Bloch-Grüneisen formula for the electrical transport, which has a characteristic temperature, the Debye temperature θD, which is a measure of the strength of the lattice vibrations, the phonons. At temperature well above the electrical resistivity ρ varies linearly as a function of the temperature. Well below θD it show a T5 temperature dependence, where the mean frequency path is eventually set by the average distance between the impurities leading to constant value ρ0 at zero temperature.

The electrical resistivity thus can be scaled as function of the Debye temperature for a number of metals and they all collapse on a common curve.

Relatively early on, researchers noted that if one adds magnetic impurities to an otherwise non-magnetic metal, that the electrical conductivity is no longer temperature independent. Instead, upon cooling the electrical resistivity goes through a minimum and starts to increase logarithmically with decreasing temperature.

We give LaCoIn5 as an example, which is a good non-magnetic metal. When 10% of the La is replaced with magnetic Ce, upon cooling from high temperatures, the resistivity goes through a minimum at about 40K, and starts to increase logarithmically with decreasing temperature. With the semi-logarithmic temperature scale this leads to a straight line behaviour towards the lowest temperatures.

Here, the conduction electrons are scattering on the magnetic Ce atoms leading to the so-called Kondo minimum in the electrical resistivity, as well as a logarithmic divergence at the lowest temperatures.

This was the first indication of what was to come. So, next up is what happens if instead of having just magnetic impurities, instead you make them part of the lattice.

If you wish to contribute to this text, please feel free to email an improved version to Andrea Bianchi.