Thread by @plasmon

While the dust may have settled and you may be sick of hearing about #lk99, it is worth highlighting the central irony of #lk99: a completely different "super" property masqueraded as a signature of superconductivity: superionicity. So, this is short thread about superionicity

But if you stick to the end, I make it worth your while and link back to the #lk99 puzzle with a clue about oxygen that most have missed. You won't need to read the preprint: arxiv.org/abs/2308.05222

This property, superionicity, while entirely different from superconductivity and not as popularly known is fascinating in its own right. So, what is superionicity? If you guessed "ions moving around super fast", you'd be mostly right.

Ions are much more massive than electrons. Besides in a crystalline solid, you'd expect ions to be vibrating around their mean positions but not moving around. And yet, in some materials such as copper sulfide, copper sulfide, and silver iodide, something unusual happens...

The cations are zipping about as if they were in a liquid. The material becomes a conductor of electricity except the carriers in this case are cations rather than electrons. A superionic conductor!

Copper sulfide, copper selenide, and silver iodide are all examples of ionic solids made of cations and anions arranged in a crystalline lattice, but there is something unusual about them.

To highlight this unusual nature, take the example of table salt, sodium chloride made of sodium cations and chloride anions. Heat up sodium chloride and at about 800 deg C, the solid melts to form a molten salt (a liquid phase).

But now take the example of copper (I) sulfide, the guest star of #lk99. It is an ordered crystal of copper cations and sulfide anions. Heat it up.

Well before you reach its melting temperature of 1130 deg C, you cross a fascinating phase transition known as the superionic phase transition at 104 deg C.

Oddly, at this temperature, only "half of the solid" melts. The copper cation sub-lattice melts while the sulfide anions stay put in their rigid hexagonal crystalline framework.

How can that be? This is still being worked out, but loosely speaking, the cations being significantly smaller than the anions allows the two sub-lattices to "decouple". Being smaller, there are many interstices for the cations to hop to and from.

The effect of this: the cations form a "liquid like" mobile network, constantly hopping from one interstice to another: nature.com/articles/ncomm…

However, the solid still maintains the strength and mechanical properties of a solid due to the rigid crystalline arrangement of the anions.

If it still looks like a solid, how can we tell that the cation sub-lattice is molten? In X-ray diffraction or electron microscopy, the diffraction peaks or lattice fringes associated with the ordered arrangement of the cations disappear at the superionic transition temperature.

Fascinating to watch. We captured this in an-in-situ TEM experiment on copper selenide: nature.com/articles/s4146…

Or if you hook up electrodes to the solid and apply a voltage across it, you can detect the flow of ions and measure ion conductivity (using method called AC impedance spectroscopy that helps you distinguish conductivity due to flow of ions from that due to flow of electrons).

This is a useful property when you desire to move cations around, as in a rechargeable battery. In principle, superionic phases can be ideal electrolytes for batteries...

None of the problems of a flammable liquid electrolyte while enabling fast charging/discharging by allowing cations to zip about at speeds (diffusivities) as in a liquid.

So why aren't these materials used commercially? The useful superionic phase is accessible only at high temperatures (above 104 deg C for copper (I) sulfide) that are not conducive to battery operation. We'd like the superionic phase to be accessible at room temperature.

We want room-temperature superionicity! So, the outstanding challenge is to decrease the phase transition temperature, Tc. You might note that this is the exact opposite of the goal in superconductvity research, where we desire to push the Tc higher.

One of the ways for decreasing Tc we found serendipitiously in copper selenide is strain.

Compressive strain on the lattice makes the different interstices in the lattice more energetically similar and cations can hop from one to another without a large energy barrier: pubs.acs.org/doi/full/10.10…

A 2% strain can reduce the Tc to well below room temperature. At room temperature, the copper ions are zipping about as if in a liquid. Fascinating!

But how do we apply strain? We use nanocrystals, which naturally have an in-built compressive strain. But that creates other problems for practical implementation that needs a whole separate thread.

Now as promised, back to the superionic transition that was central to the #lk99 puzzle.

So as you heat up copper sulfide, right at the superionic phase transition, 104 deg C for copper (I) sulfide, you would see a large, orders-of-magnitude jump in the conductivity.

This is because the low-temperature ordered crystalline phase has barely any ionic conductivity whereas the high-temperature superionic phase has a large ionic conductivity.

In other words, as you raise the temperature, at 104 deg C, you would see a sharp drop in resistivity. (reminder: resistivity is reciprocal of conductivity).

Wait a minute! That is the exact opposite of what was seen for #lk99. There was a sharp jump in resistivity for #lk99 as you heat up and cross 104 deg C and not a drop.

So what gives? This is where the part of the story where oxygen plays a role.

Copper (I) sulfide is by way of a molecular formula, Cu2S, which it is if you make it in completely oxygen-free conditions. In mineralogy, this is called chalcocite. But any bit of oxygen exposure and some of the Cu atoms happily leave the lattice to react with oxygen.

At 2-3% Cu vacancies, you form a stable crystal structure, called djureleite, which is a structural local minimum in the Cu-S composition space.

Each Cu atom that leaves the lattice leaves with its electron. So, a hole (positive charge carrier) gets left behind. So in djurleite, there are holes in the valence band with a density of 10^21 per cubic cm. That is 1 hole for every cubic nm.

This is a large enough carrier density to make the material electronically conductive. It is a pretty good p-type conductor.

So, while chalcocite (Cu2S), is a poor electronic conductor because of a very low concentration of carriers, djurleite (Cu2-xS, where x represents the amount of Cu missing from the lattice) is an exceptional electronic conductor due to all those hole carriers.

Djurleite has a fairly low resistivity at temperatures below 104 deg C. I think it is Cu2-xS (djurleite) and not Cu2S (chalcocite) that is the key to the #lk99 puzzle.

There were reports of oxygen or air exposure for the "successful" samples. For instance, from an English translation of Lee et al.'s Korean paper:

"While studying the superconductivity phenomenon... repeated experiments were conducted in earnest, and cases where the quartz tube was destroyed due to internal pressure during rapid cooling or reaction".

So when you heat up Cu2-xS, you go through the same superionic transition as Cu2S around (although not exactly) a temperature of 104 deg C. The solid goes from an ordered crystal to a superionic solid with a disordered mobile network of Cu.

No different from Cu2S, there is a sharp jump in the ionic conductivity at this transition. But the exact opposite happens to the electronic conductivity caused by the hole carriers. And it is the latter that matters because electronic conductivity dwarfs ionic conductivity.

At around 104 deg C, you transition from a fully ordered phase to a phase with a disordered Cu sub-lattice. This disorder is detrimental for the mobility of hole carriers. The mobility of hole carriers sharply drops by 2-3 orders of magnitude.

In other words, the electrical resistivity jumps sharply by 2-3 orders of magnitude as you heat djurleite and cross the superionic transition of around (although not exactly) 104 deg C.

This is precisely what was seen for #lk99 and mistaken as a signature of superconductivity. So it wasn't superconductivity nor a metal-to-insulator transition, but a ordered-to-superionic phase transition.

Superionicity is itself a worthy topic of study with open questions about the role of collective effects, role of phonons, and the nature of bonding that makes an ionic solid behave superionically.

Many phenomenal researchers working on these questions.

End of long-overdue thread on superionicity
@dangaristo @MichaelSFuhrer @InnaVishik
@RidwanSakidja @Samuel_Gregson
@8teAPi @mbw61567742 @moreisdifferent
@AlexandruBG @bedoya_pinto

@sa58R6rq0H2CgY5 Thx for the paper. LaF3 and MoS2 are distinct components of the device. LaF3 is the (superionic) solid electrolyte used to apply a gate bias on MoS2, the component undergoing said metal-insulator transition.

@sa58R6rq0H2CgY5 Besides the metal-insulator transition in the paper you linked is gate-induced rather than temperature-induced.