The Shortcomings of Our Experimental Probes

Most experimental probes in condensed matter suffer from some inherent limitations. Though these limitations can be understood by reading reviews, I find that the most fruitful approach is to speak to someone working in the area. This is because new experimental probes are always cropping up and it is difficult to read up on a probe when there are not a lot of articles available.

Here are some relatively common probes and their corresponding limitations:

1. Scanning Tunneling Microscopy/Spectroscopy (STM): Surface sensitive; energy gaps can sometimes depend on sample-to-tip distance; the density of states of the tip material can affect measurements and is not often measured
2. Angle-Resolved Photoemission Spectroscopy (ARPES): Surface sensitive; $p_z$ not resolvable as a result; needs a freshly cleaved surface; material must be a reasonable metal so that it doesn’t charge during photoemission process
3. Inelastic Neutron Scattering (INS): Large single crystals needed; can’t measure collective modes like plasmons, as neutrons are not charged particles
4. Inelastic X-ray Scattering (IXS): Particularly useful for smaller samples where INS cannot be used; even though IXS is sensitive to the electron density, most electrons are “core” electrons, meaning that phonons are more prominent in IXS spectra than electronic excitations, especially for materials containing high atomic numbers
5. Infrared/Optics: To obtain most quantities of interest, such as conductivity or loss function, a Kramers-Kronig analysis is needed — this can affect the spectra depending on the energy range of data taken. Ellipsometry obviates the need for a K-K analysis, however; optics also cannot probe excitations away from the $\Gamma$ point; for reflectivity measurements, a reference sample, such as gold, is needed
6. Electron Energy Loss Spectroscopy (EELS): Often suffers from multiple scattering problems so that response is not perfectly linear; energy resolution is not great for transmission EELS (~80 meV), while reflection EELS is usually surface sensitive (probe depth ~5-20 Angstroms).
7. NMR Knight Shift: the orbital magnetization is usually subtracted — how this is done is important.

Transport and thermodynamic measurements suffer from fewer shortcomings than most of the spectroscopic tools that I’ve mentioned above. I’m far from an expert in those measurements, so I’d be interested to hear more in the comments section about the inherent limitations of those experimental probes.

I do feel that it is not only experimentalists that need to have a good grasp of the limitations of experimental probes, but also theorists. A deeper understanding of experimental methods will ultimately result in a more nuanced view of data. This point is especially important when it comes to a topic like high-Tc superconductivity where it seems like there are new probes being used every day.

If anyone wants to add to the list, please feel free to do so in the comments. If there are any glaring ones that I’ve missed, please let me know!

2 responses to “The Shortcomings of Our Experimental Probes”

1. drie koffers

@1: current always depends on tip-sample distance. That is the point of the whole experiment… So I think you were trying to say something else here. A problem is that (dI/dV) spectra depend on the DOS of your tip, and I have the impression a lot of people don’t check that on a known sample. Measured gaps are artificially increased when tip-sample distances are larger (extreme: you can’t tunnel with a bias of 1 mV while you are 3 nm away, i.e. you will have a gap even if your sample is a metal), or when the sample charges at low temperature which stretches the voltage/energy axis.
@2: z-dispersion is measurable by changing the photon energy! (which is the standard way of proving something is a 2D state: change hv and note the absence of dispersion).

@ transport: contacts, contacts, contacts… It is crucial, especially for materials like complex oxides, that you know exactly how your contact behaves as a fct of T, V, and I.

Like

• Anshul Kogar

Thanks for your comment. The gap variation with tip-to-sample distance was what I was referring to — thanks for catching that. I will update the post to reflect this. Also, I will add your comment about the DOS of the tip, that’s an important point as well.

As for the ARPES, a true z-dispersion is still not truly possible without a complicated model. I agree though that one can tell if a surface state is truly confined to the surface.

Like