This week’s Journal Club update is inspired by yesterday’s colloquium by Helmut Dosch, one of the directors of Max Planck institute. His excellent talk was on solid-liquid interfaces, one of the most puzzling and inaccessible types of interfaces. Solids and liquids have very different properties, and what happens (on atomic scale) when the two of them meet is currently not known. Little of what is known can be attributed to Max Planck/Stuttgart group lead by Dosch – it could be argued his group made more progress than the rest of the world combined. Here’s a sampler of some of the work Helmut described, with majority already featured in our shared liquids topic news selection.
One of the papers that quickly became “classic” is the first observation of five-fold symmetry in a liquid, published in Nature a few years ago. One of the key reasons for why metals can be supercooled, or why glasses, or amorphous solids can exist is that liquids favor five-fold symmetry of forming local icosahedral structures. This is the most efficient local packing, which famously cannot be extended to periodically fill 3D space. Therefore, on the path to crystallization bonds favoring icosahedral geometry must be broken, and atoms re-arranged in some other symmetry (for example cubic). This creates a barrier to nucleation, which explains why crystallization and melting are first-order, rather than second-order transitions. The five-fold local symmetry in liquids cannot be easily observed – because of random orientation of icosahedral elements, the five-fold effect is “smeared out”. However, by placing liquid (Pb) near hard-wall of solid (Si) Harald Reichert and Dosch could align liquid clusters and observe five-fold symmetry using hard x-ray beam that could penetrate through the material and scatter off the interface.
Another key paper is a study of premelting of ice in contact with solid surface (Silica in this case). They unexpectedly find that premelted layer has a higher (rather than lower) density than water – by almost 20% – consistent with high density amorphous ice, a phase that is typically found at high pressures and much lower temperatures.
Helmut also mentioned two recent papers – one is PNAS study of density “gap” that occurs when water meets hydrophobic interface. Previous neutron scattering data placed the thickness of this interface at about 4 nm, but with an error bar that was arguably larger than 4 nm. In other words, they merely placed an upper limit of 10 nm. Dosch’s argument was that condensed matter physicists should learn from high energy folks about proper use of error bars. But with x-rays they could measure to a much higher resolution, getting a very precise measure of density*thickness, which is consistent with about half a monolayer, and a range of thicknesses ranging from 2 to 8 Angstroms – much better than previous studies. Earlier this year Paul Fenter at Argonne had obtained a similar result.
Finally, the most intriguing result is a recent PRL on giant metal compression in a Schottky junction. When liquid Pb meets crystalline Si, structurally you have a liquid and a solid in contact with each other. But electronically you have a metal and a semiconductor, which can lead to uncompensated charge building up at the interface. The result is a high density layer of Pb near the interface. The thickness of this layer is huge – 15 Angstroms, which is one of the key puzzles. If this was a result of contamination one would expect a monolayer coverage, but not 5-6 layers. It is also hard to imagine any “dirt” having higher density than Pb. There is always a problem of “complex conjugate” density profiles where inverted function can give the same reflectivity, due to loss of phase information, but this inversion would result in density lower than Si, which is also somewhat unphysical, and I have trust that the Stuttgart group explored this possibility carefully. Either way, the effect disappears when you look for the same high-density layer of Pb in contact with dielectric, such as alumina. But it can also be observed for other metal-semiconductor interfaces, for example In and Si – so it is not just Pb that behaves this way. Interestingly, 15-20 Angstroms is about the electronic mean free path in Pb and In.
The amount of technical expertise that goes into these measurements is absolutely astonishing – from ability to focus hard x-ray beams and track them by scanning angles with micron-precision, to growing and cleaving ice in special “ice lab” kept at temperatures low enough to prevent sublimation – some of their artificially grown ice is many decades old. Apparently they don’t make ice of as high quality as they used to – I am not kidding, this is true.