Saturday, August 20, 2011

Gating and "real" metals.

Orientation week has kept me very busy - hence the paucity of posts.  I did see something intriguing on the arxiv recently (several things, actually, but time is limited at the moment), though.

Suppose I want to make a capacitor out of two metal plates separated by empty space.  If I apply a voltage, V, across the capacitor using a battery, the electrons in the two plates shift their positions slightly, producing a bit of excess charge density at the plate surfaces.  One electrode ends up with an excess of electrons at the surface, so that it has a negative surface charge density.  The other electrode ends up with a deficit of electrons at the surface, and the ion cores of the metal atoms lead to a positive surface charge density.  The net charge on one plate is Q, and the capacitance is defined as C = Q/V.

So, how deep into the metal surfaces is the charge density altered from that in the bulk metal?  The relevant distance is called the screening length, and it's set in large part by the density of mobile electrons.  In a normal metal like copper or gold, which has a high density of mobile (conduction) electrons on the order of 1022 per cm3, the screening length is comparable to an atomic diameter!  That's very short, and it tells you that it's extremely hard to alter the electronic properties of a piece of normal metal by capacitively messing about with its surface - you just don't mess with the electronic density in most of the material.  (This is in contrast to the situation in semiconductors or graphene, by the way, when a capacitive "gate" electrode can change the number of mobile electrons by orders of magnitude.)

That's why this paper was surprising.  The authors use ionic liquids (essentially a kind of salt that's molten at room temperature) to modulate the surface charge density of gold films by something like 1015 electrons per cm2.  The surprising thing is that they claim to see large (e.g., 10%) changes in the conductance of quite thick (40 nm) gold films as a result of this.  This is weird.  For example, the total number of electrons per cm2 already in such a film is something like (6 x 1022/cm3) x (4 x 10-5 cm) = 2.4 x 1018 per cm2.  That means that the gating should only be changing the 2d electron density by something like a tenth of a percent.  Moreover, only the top 0.1 nm of the Au should really be affected.  The data are what they are, but boy this is odd.  There's no doubt that these ionic liquids are an amazing enabling tool for pushing the frontiers of high charge densities in CM physics....

5 comments:

Don Monroe said...

Very odd indeed.

Anonymous said...

A number of results on field effects (insulating gate, ferroelectric gate and ionic gate) in correlated oxides (with quite high carrier densities) are odds and conflicting...

DaveC said...

These films are pretty inhomogeneous. I suppose if the current passes through some thinner filamentary bits, perhaps with grain boundaries in them, that could increase the sensitivity to near-surface changes quite a bit. And gold does have some chemistry. It's a little worrying that they mention preliminary 30% changes with copper, which must be pretty electrochemically active. The onus is really on them to prove it's not a mundane surface electrochemical effect.

Anonymous said...

Agreed, odd indeed. The vertical shift of the data on a linear scale (Fig 3) suggests another conductance channel. I've been seeing a lot of similarly puzzling data floating about, unpublished, that are proposed as major advances in experimental metals physics. I rather suspect that we're looking at surface electrochemistry.

Anonymous said...

The paper uses a simple doped polymer electrolyte and NOT an ionic liquid.

BTW, thanks for the link to an interesting paper I would not have otherwise come across.