Source of the post
I think it would be more like a red/white substance since the mercury might glow red-hot or even white-hot when it gets that hot. Fascinating that mercury's conductivity is dramatically increased as it heats up to supercritical, makes you wonder.
Yeah, I discount the thermal glow, which would probably be an intense yellow and very bright at 1800K. This is similar to the temperature at which wrought iron melts. So at these temperatures and pressures visual inspection of the substance would not only be fairly difficult, but also perhaps not very revealing.
As for conductivity, this is actually increasing with pressure rather than temperature. In these experiments they were staying at a constant temperature, and increasing the pressure. (If we instead held the pressure constant but increase the temperature, it will destroy the conductivity.) At temperatures well below the critical point the transition from non-metal to metal as pressure increases is immediate and associated with the mercury condensing to liquid. But near the critical point the transition branches off from the condensing line, extending up into the supercritical fluid region and becoming more fuzzy. I've finally managed to track down a paper showing where this transition lies on a phase diagram:
So it's really a separate transition from becoming supercritical. What's happening is that increasing the pressure increases the density, and this brings the atoms and the electrons closer together, until they (in some complicated way) form a conducting "sea" of mobile electrons, just like in a metal.
I had never heard of this transition occurring in supercritical fluids before. It's very interesting. I wonder if it also has similarities to the transition of hydrogen becoming a liquid metal?
Source of the post
I figured Mercury would have the least extreme critical point because of its relatively low boiling point, but is there another metal (maybe cadmium?) that has a lower critical point? Maybe it shows the same properties.
Apparently, nobody has determined the critical point of cadmium! But I would guess it must be higher than mercury, since the melting and boiling points are also higher. Sodium does have a known critical point (2573K and 350atm, based on extrapolation), and potassium
(2198K and 153atm, directly measured).
The idea of experimenting with high temperature and high pressure sodium or potassium terrifies
me by the way.
Source of the post
What prevents us from perfectly modelling what these properties are?
Or, to turn the question upside down, by measuring these properties what new fundamental knowledge of physics do we get?
Another really good question. This falls under the subject of "condensed matter physics", which is one of the most active fields of experimental and theoretical physics today. A fair number of faculty at my university do research in it. And the short story about it is "it's very
The key problems for predicting the properties of condensed matter lie in accurately modelling the inter-atomic forces, lattice structure and modes of energy storage (if a solid), and the behavior of the electrons. The electrons may even be the most important and complicated part of it all. Electrons bound within atomic orbitals determine how the material bonds and behaves chemically, as well as its magnetic properties. Unbound electrons determine the electrical properties. But computing either is an extremely complicated quantum mechanical problem. Not because it uses quantum mechanics (for which we have a very good theory), but because the system is difficult to model and apply it to.
For example, the unbound electrons move about the material, but still feel forces due to each of the atomic nuclei (so we must solve the Schrodinger equation for the electron with the potential it feels from the atomic lattice). We must also account for the electrons being fermions (no two occupy the same state), and that the electrons also interact with one another! In solids this leads to the theory of electron band structure
, and something similar happens in fluids.
So in general, there are nice trends and a range of models we can use to understand or predict the properties of many materials, but accurately predicting the properties of any material ab initio
is quite hard, especially for more complicated ones or in more unusual conditions. It can be common to get significant (10s of percent or more) errors compared to experimental results. It is also possible for nature to do something unexpected (superconductors and superfluids were experimental surprises).
For the metal-nonmetal transition in particular, it turns out there is a deep history of research and understanding of it (and this research is still active). The paper I got the above phase diagram from (A Peculiar Fluctuation in the Metal-nonmetal Transition Observed in the Supercritical Fluid Mercury
) has a very nice review of this in the introduction. The transition was actually predicted
by Landau, who was awarded the Nobel Prize in physics in 1962 for his other work on condensed matter.
More recently (2014) there was a study investigating the same thing in supercritical iron
, from the theoretical side using simulations. The details of this may be important in astronomy with the cores and magnetic fields of planets. They compare it with a transition between a rigid and non-rigid state called the "Frenkel line
", and show a method for computing it and describing how it works through a conceptual model. Here's a pretty good excerpt:
What help does measuring this transition experimentally do for our understanding of the physics? It might help give some insight on how well this conceptual "sphere-packing" model works for describing it, or how to modify or replace it with a model that predicts it better.
For a historical analogy we can look back to the successive iterations of models for the heat capacity of a solid, where multiple physicists (Einstein, Nernst and Lindemann, Debye) worked out how to understand and predict the behavior in different regimes of temperature.