Graphene has been used as a 'transparent' layer that allows single crystals of a material to be grown on a substrate, and then lifted off — in much the same way that baking paper lets cakes be removed easily from tins. See Letter p.340
About ten years ago, atomically thin layers of carbon atoms known as graphene sheets were shown to be highly transparent to visible light, making graphene a promising material for electronic display applications1,2. Then, in 2012, the phenomenon of liquid wetting on solid surfaces was, in some instances, found to be unaffected by an interposed sheet of graphene — the ultra-thin sheet is 'transparent' to wetting, so that the arrangement of water molecules in a droplet is driven by interactions with the substrate that lies beneath the graphene3. On page 340, Kim et al.4 ask whether crystal growth can take place through a sheet of graphene, and report that the answer is yes: atoms of gallium and arsenic 'see through' a graphene layer to an underlying gallium arsenide (GaAs) crystal, and thus adopt an arrangement that seamlessly continues the lattice structure of the substrate.
The process in which materials are grown over the top of a substrate is known as epitaxy. Epitaxial growth of GaAs and related compounds lies at the heart of a wide range of modern technologies, including light-emitting diodes for energy-efficient solid-state lighting5, lasers for optical telecommunication6, and high-speed circuits for mobile phones and other wireless communication7. In ordinary processes for growing GaAs using epitaxy, great pains are taken to ensure that the substrate's surface is free from contamination by other materials or by GaAs oxide. This is because the gallium and arsenic atoms supplied to the surface need to be able to 'feel' the atomic arrangement of the GaAs substrate. Placing a layer of graphene on top of such a carefully prepared surface might therefore be expected to interfere with the ability of the substrate to guide the process of epitaxial growth.
This is where the atomic thinness of graphene comes into play. Using first-principles computations, Kim et al. showed that a graphene interlayer should not prevent newly arriving gallium and arsenic atoms from interacting with the GaAs substrate beneath, provided that the interlayer is less than 9 ångströms thick. The authors then used metal–organic vapour-phase epitaxy, a commonly used industrial technique, to show that smooth, single crystals of GaAs at the square-millimetre scale can grow epitaxially on a graphene-coated GaAs substrate — a process they call remote epitaxy (Fig. 1).
Kim et al. went on to use a wide range of diffraction and electron-microscopy techniques, from the atomic to the millimetre scale, to prove that the observed growth was indeed directed by the GaAs substrate. To underscore the point, the authors repeated the experiment using GaAs substrates covered with two and four layers of graphene, and observed only rough, disordered polycrystalline growth. The substrate can no longer guide the arrangement of gallium and arsenic atoms through such thick barriers — the multi-layer graphene is 'opaque' to epitaxy.
So does this fundamental investigation of crystal growth have technological applications? To address this question, Kim et al. showed that a single layer of graphene enables GaAs films to bond strongly enough with the substrate to allow epitaxy, and yet weakly enough to allow film removal (exfoliation) under mechanical stress. This enabled the authors to grow visible-light-emitting diodes (roughly 0.1- to 1.0-micrometres thick) on graphene–GaAs and then exfoliate them. The devices retained their functionality after being transferred to a silicon carrier. Such thin devices can be highly bendable when separated from their brittle substrates, creating opportunities for flexible electronics8.
The exfoliation enabled by the graphene interlayer differs from typical methods9,10, which require the chemical etching of a sacrificial layer and/or mechanical polishing of the substrate to afford a smooth surface for subsequent epitaxial growth. It is also worth noting that the graphene-coated GaAs substrates can be reused for epitaxy. Given that substrates of GaAs and other semiconductors can be expensive, the ability to reuse substrates could create substantial cost savings in applications that involve large areas, such as high-efficiency solar cells.
Some hurdles remain before Kim and colleagues' epitaxial process is ready for commercial application. First, although the researchers show that the surfaces of the epitaxially grown GaAs films are smooth on the micrometre scale, they are not uniformly smooth over large areas (square millimetres). Second, any defects present in the graphene interlayer cause damage to the substrate and the film during exfoliation. Moreover, the authors prepared epitaxial templates by transferring previously prepared graphene onto GaAs substrates. A process for directly forming single layers of graphene on GaAs could greatly simplify things, and might allow larger, more uniform GaAs films to be made. Work will also be needed to show that devices with state-of-the-art performance can be made, such as low-threshold lasers or high-efficiency solar cells.
In addition to positioning remote epitaxy as a new mode of crystal growth, Kim and colleagues' work might open up manufacturing processes for low-cost flexible optoelectronics. The results also raise important fundamental questions, such as how dislocations, stacking faults and other crystal defects in 3D materials might interact with 2D materials. And would other 2D materials, such as boron nitride, be similarly transparent to epitaxy? Finally, the work might inspire research into devices that marry the distinctive characteristics of 2D materials — such as their electronic, optical, and thermal properties — with the ultra-high performance, exquisite control and structural perfection that can be achieved using 3D epitaxial semiconductors. Footnote 1
Bonaccorso, F., Sun, Z., Hasan, T. & Ferrari, A. C. Nature Photon. 4, 611–622 (2010).
Roddaro, S., Pingue, P., Piazza, V., Pellegrini, V. & Beltram, F. Nano Lett. 7, 2707–2710 (2007).
Rafiee, J. et al. Nature Mater. 11, 217–222 (2012).
Kim, J. et al. Nature 544, 340–343 (2017).
Krames, M. R. et al. J. Disp. Technol. 3, 160–175 (2007).
Tatum, J. A. et al. J. Lightwave Technol. 33, 727–732 (2015).
Raab, F. H. et al. IEEE Trans. Microw. Theory Tech. 50, 814–826 (2002).
Rogers, J. A., Someya, T. & Huang, Y. G. Science 327, 1603–1607 (2010).
Bedell, S. W. et al. IEEE J. Photovolt. 2, 141–147 (2012).
Lee, K., Zimmerman, J. D., Xiao, X., Sun, K. & Forrest, S. R. J. Appl. Phys. 111, 033527 (2012).
Related links in Nature Research
About this article
Chemical Society Reviews (2018)