Peeling electronics off indefinitely re-usable wafers

April 25, 2017 // By Julien Happich
Leveraging the so-called wetting transparency of graphene, a team of researchers at the Massachusetts Institute of Technology (MIT) has taken a novel approach to semiconductor epitaxial growth, one that would leave the original substrate wafer intact and re-usable for the repetitive growth of new devices, indefinitely.

To achieve defect-free crystalline semiconductor materials on a substrate, epitaxial growth requires a good lattice matching between the substrate material and the new semiconductor layers being grown (for the elaboration of electronic devices). Any mismatch in the materials will yield dislocations and other defects likely to impair the charge mobility and operation of the devices.

Hence, the need to start with bulk wafers of the same material as the one that will be grown on top. But drawing silicon ingots and slicing them into wafers is a costly operation, and exotic materials such as SiC, GaAs, InP or GaP are difficult to grow in large ingots, making the resulting wafers smaller and their use for electronic devices more expensive.

Today's cost-saving strategy is to slice as many wafers as possible from a given ingot, and then again, most of the substrate material is sacrificed through wafer thinning. Once a device has been grown on top, the reminder of the substrate material merely becomes part of the device's package (an expensive one in the case of exotic materials).

Publishing their results in Nature under the title "Remote epitaxy through graphene enables two-dimensional material-based layer transfer", the researchers have used monolayer graphene as a non-stick interfacing sheet between the substrate wafer and the epitaxially grown material.

Through density functional theory calculations, they found that in most cases, a sufficiently thin and electrically penetrable layer of graphene with a weak van der Waals potential would not completely screen the stronger potential field of the underlying substrate, allowing adatoms to experience remote epitaxial registry with the substrate.

Their theoretical calculations indicated that remote epitaxial interactions would be felt through a separation distance up to nine ångströms, a gap large-enough to accommodate a monolayer of graphene.

Experimentally, the researchers confirmed that a monolayer of graphene (even going to four layers) was electrically transparent-enough to allow the adatoms to interact with the underlying substrate, growing a new layer of identical material through so-called remote homoepitaxy.

They applied their new findings to the homoepitaxial growth of GaAs(001) on GaAs(001) substrates through monolayer graphene, explaining that such an approach was also applicable to InP and GaP.


High-resolution STEM images showing the excellent remote alignment of the GaAs(001) lattices through the graphene. Convergent-beam electron diffraction patterns from the epilayer (top inset) and the substrate (bottom inset) show identical zinc-blende (001) orientations. Right, a low-angle annular dark field STEM image showing no dislocations.