A 5 nm gold sacrificial film turns a 2D materials clean-up problem into a process module

One-tenth is the number doing the work here, and it runs backwards as well as forwards. If the transistor's contact resistance came out at about 2.58x10^5 ohm-micrometre and that is a tenth of the previous level ^[7], the baseline these devices had been measured against sat near 2.58x10^6 ohm-micrometre ^[13]. On-current rose by roughly the same factor that resistance fell ^[8], which points at the contacts rather than the channel as the thing that had been setting the ceiling.

The interesting part for an underwriter is not the ohms but the order of operations. The team's own simulation work says the damage is chemical: oxygen in the plasma rewrites the photoresist, more than doubling its adsorption energy and leaving defect levels that block charge ^[5]. Once that has happened there is nothing good left to do, because solvents will not lift hardened resist and ultrasound or heat take the film off along with the residue ^[4]. So the countermeasure is inserted ahead of the damage. Five nanometres of gold goes down before the resist, patterning happens on top of it, and the chemical strip that dissolves the gold carries the hardened residue away with it ^[6]. Microscopy and spectroscopy after processing found no damage to the crystal structure ^[9].

That is a process module rather than a materials record, and process modules are the thing 2D logic has been short of. Which makes the portability claim more important than the transistor: first author Kim Da-hyun says the flow also works for finer circuits drawn with an electron beam, and works regardless of how the molybdenum disulfide was synthesised ^[10]. A contamination step that is indifferent to synthesis route can be evaluated on somebody else's material, which is the difference between a result and a recipe.

The gap is equally specific. What UNIST published is a device-level figure with no distribution behind it: no yield, no device-to-device spread ^[12]. Yield is the claim being implied by a tenfold resistance drop, and yield is not the claim being measured. Professor Kim Myung-soo's own framing is that clearing the contamination obstacle can contribute to future highly integrated logic and memory ^[14], which is a research statement and unobjectionable as one. The capital read is narrower. Two-dimensional semiconductors have been stuck at the point where a good film becomes a bad device, because residue that obstructs charge transport arrives with the patterning itself ^[3], and this work attacks that failure mode with a named cause and a measured countermeasure ^[5][6]. Whether it survives contact with a line the inventors do not control is the whole of what follows.