Island Sliding Barriers: A first-principles metric for determining remote epitaxy viability

Abstract

Remote epitaxy, where a 2D van der Waals material (usually graphene) is inserted on top of the substrate before film epitaxy, has emerged as a promising path for growing electronics with lower defect rates and less stringent lattice matching requirements. The exact mechanism behind remote epitaxy has not been definitively shown, however, and it is not obvious when examining a new substrate-film pair whether they would be compatible with the remote epitaxy process. In this paper, we use first principles calculations to test several different mechanisms for determining whether a given substrate-film pair will successfully be grown with remote epitaxy. We find that previously calculated metrics such as electrostatic potential do not hold sufficient explanatory power. We find that the sliding barrier of small islands on the surface when the atomic positions are allowed to optimize provides the most rigorous criteria for whether a given substrate-film pair is remote epitaxy active. This indicates that remote epitaxy is likely a phenomenon related to the kinetics and ease of island migration on the graphene surface.

Figures (7)

• Figure 1: A schematic of a typical remote epitaxial (RE) scheme, where a substrate is covered by a 2D van der Waals (vDW) layer, which is then used to grow high quality, single crystal films that would not previously be available for the substrate. We then present a schematic of which systems have been found to be viable with RE, with green boxes representing that experimental evidence has been shown with single crystalline layers from these setups, red indicating that only multicrystalline, vDW like layers have been grown, and yellow indicating that the evidence is ambiguous. The data for GaAs/graphene/GaAs comes from Ref. kim2017remote, Ge/graphene/GaAs and GaAs/graphene/Ge from Ref. chang2023remote, GaN/graphene/SiC from Ref. journot2019remoteqiao2021graphenechang2023remote, and GaN/graphene/GaN from Ref. kong2018polarity.
• Figure 2: The planar average charge moving toward the surface for (a) SiC, (b) GaAs, and (c) GaN. We then examine (d) how the decay distance of each material changes as a function of the number of graphene layers.
• Figure 3: The electrostatic potentials 3 Å above the surface of RE growth substrates. We show the electrostatic potentials for SiC with no reconstruction, the SiC $6\sqrt{3}\times6\sqrt{3}$ reconstruction, GaAs with no reconstruction, the GaAs (100) 2$\times$4 reconstruction, and GaN. We show the electrostatic potentials for when the substrate is covered with one and two layers of graphene. For the SiC case, the first layer of graphene bonds with the underlying silicon atoms and is often called the buffer layer graphene. We show this in the same column as one graphene layer results for the other materials, and a graphene layer on top of the buffer layer as being equivalent to the two graphene layer cases in other systems.
• Figure 4: (a) The max electrostatic potential, as shown in Fig. \ref{['fig:electrostatic-pot']} for each substrate and (b) the largest difference in electrostatic potential for each of the different substrates.
• Figure 5: The adsorption energy of individual atoms above the surface of SiC, GaAs, and GaN, with one or two layers of graphene included above the surface. The corresponding electrostatic potential is plotted in orange squares for each system.