Hot pick-up

Going deeper

Why the interface decides the device

A van der Waals stack is only as good as what is between its layers. Trapped hydrocarbons, water and air do not dissolve into the crystals; they coalesce into bubbles and pockets that locally separate the layers, scatter carriers, dope unevenly and make anything measured an average over clean and dirty regions. Encapsulating graphene in boron nitride raises its mobility by orders of magnitude, but only where the interface is clean.

There is a self-cleansing effect working in the right direction: because the layers attract each other more strongly than they attract contaminants, adsorbates are squeezed into isolated bubbles, leaving atomically clean areas between them. The purpose of a good assembly technique is to help that process rather than fight it – to get the contaminants to leave the interface entirely instead of gathering into bubbles in the middle of the device.

What the heat does

Working at elevated temperature makes the polymer stamp conform smoothly and lets the contact line advance in a controlled way from a single touch-down point. Instead of the layers slapping together and sealing in whatever was there, the front sweeps outward and pushes the contamination ahead of it to the edge of the flake.

The practical result is throughput and yield rather than a single record device. The technique was demonstrated by building 22 mono-, bi- and trilayer graphene stacks encapsulated in boron nitride at close to 100 % yield, with semiclassical mean free paths up to 0.9 µm in the monolayer devices – high enough that the narrowest samples were limited by scattering from their own edges rather than by anything inside. Because layers are picked up one after another, a whole stack can be assembled on the stamp and set down once, and pre-patterned layers can be used, which opens the door to more complicated architectures.

What it is not suitable for

Anything that degrades when warm is excluded, which is a long list among the interesting materials: many magnetic halides, black phosphorus and several tellurides oxidise or decompose well below the useful stamp temperatures, so they are handled cold and in a glovebox instead.

The second exclusion is more subtle. Heat is exactly what lets a twisted stack relax toward aligned stacking, so the same warmth that cleans an interface can destroy the twist angle that the device was built around. Magic-angle work therefore uses assembly at lower temperatures and accepts a dirtier interface, or accepts angle relaxation and measures what it ends up with. The two goals – clean interfaces and preserved twist – pull in opposite directions, and which one wins is a decision made per experiment.