Monday, October 26, 2015

What is the origin of noise in this bad metal?

Last week I had a helpful discussion with Arindam Ghosh about this recent paper

Conductivity noise study of the insulator-metal transition and phase coexistence in epitaxial samarium nickelate thin films 
Anindita Sahoo, Sieu D Ha, Shriram Ramanathan, and Arindam Ghosh

The abstract states
The normalized magnitude of noise is found to be extremely large, being nearly eight orders of magnitude higher than thin films of common disordered metallic systems, and indicates electrical conduction via classical percolation in a spatially inhomogeneous medium.  
The higher-order statistics of the fluctuations indicate a strong non-Gaussian component of noise close to the transition, attributing the inhomogeneity to the coexistence of the metallic and insulating phases. 
The figure below shows how the non-Gaussian component [measured by the kurtosis] increases dramatically as the temperature decreases below the metal-insulator transition.

Some of the fundamental and related questions that arise are:

How much is the noise due to extrinsic disorder [i.e. due to impurities] and how much due to intrinsic disorder [inhomogeneities arising from phase separation at a first-order metal-insulator transition]?

What happens in a very clean system?

How does one actually do a microscopic calculation of this 1/f noise [e.g. from a disordered Hubbard model]?

What happens in a clean system when the temperature crosses over from a Fermi liquid at low temperatures to a bad metal with increasing temperature?

In a recent PRL there is also noise data for  an organic charge transfer salt, in which the disorder and proximity to the Mott transition can be tuned with the cooling rate.

1 comment:

  1. Maybe useful (for the experimentalists:)
    Ward et al have shown that making narrow wires out of thin films of phase separated oxides (manganites) allows to amplify the noise - if the wire cross section is on the order of the domain size.
    There the noise was very much more prevalent near the phase separation temperature range, suggesting it is intrinsic, and not extrinsic. (I don't know whether the same approach would be feasible in the nickelates as I don't know the characteristic domain length scale.)

    Ward et al, PRL 92 087201, PRB 83 125125.