Coaxial nanowires from van der Waals epitaxy

The development of optoelectronic devices such as sensors, emitters, and photovoltaic cells is increasingly dependent on the fabrication of novel semiconductor alloys made from materials that do not readily form the monolithic crystals needed for high-performance operation. Heteroepitaxy provides a route toward such crystals by ‘growing’ the desired alloy on the planar face of a crystalline substrate. Van der Waals epitaxy refers to a very particular variation on this method, where layers are formed on inert 2D substrates such as graphene, an entirely planar material with all atoms and bonds between them in the same plane. The lack of dangling bonds at the 2D substrate's free surface disallows covalent bond formation, resulting in an atomically smooth interface. The interaction between the graphene layer and ‘epilayers’ grown on top therefore consists only of transient van der Waals forces, temporary fluctuations in polarity resulting from orbiting electrons. In principle, this allows fabrication of architectures on substrates where there is a high degree of lattice mismatching, without the commonly observed dislocations that result from lattice distortion.

Originally applied to the growth of transition metal dichalcogenides (metals combined with elements of group VI of the periodic table, such as sulfur and selenium) in the early 1990s, van der Waals epitaxy was shown to enable high-quality crystal growth, even in heterogeneous systems with up to 10% lattice mismatch.¹ More recently, the same concept has been applied to the heterogeneous growth of silicon nanowire (NW) arrays on mica²and indium arsenide NWs on graphene.³ In particular, the latter study demonstrated a large leap forward toward the integration of technologically relevant III-V semiconductor materials on a conductive, transparent, and flexible platform. This work suggested that significant cost reduction may be realized through epitaxial growth of large-area NW arrays, bypassing the use of III-V materials or standard silicon substrates while also adding functionality in the form of mechanical flexibility.

Figure 1. (a) Tilted-view scanning electron microscopy (SEM) image of a field of indium gallium arsenide (InGaAs) nanowires (NWs) grown on a monolayer graphene sheet. (b) False-color scanning transmission electron microscopy image obtained from a single core-shell InAs-InGaAs NW, highlighting the phase-segregated structure.

We recently demonstrated⁴ the growth of dense arrays of vertically oriented ternary indium gallium arsenide (InGaAs) NWs on monolayer graphene sheets, with the area of growth limited only by the size of the sheets used: see Figure 1(a). Our approach relies on direct van der Waals epitaxy in a metalorganic chemical vapor deposition reactor and does not require seed particles or pregrowth substrate patterning. We are able to make our NWs over the compositional range In_xGa_1−xAs (0.2<x<1). The bandgap engineering that this allows makes these NWs suitable for multijunction photovoltaic applications, where control over the absorption spectrum of a material through compositional tuning is desired.

Through extensive structural characterization, we have determined that a spontaneous phase segregation phenomenon occurs when InGaAs NWs are self-assembled on graphene substrates. Although In- and Ga-containing precursors were supplied at constant flows during growth, the resultant NWs were composed of two discrete compositional phases, arranged in a coaxial geometry: see Figure 1(b). An InAs central core is encapsulated by an outer InGaAs shell. This was true for all the NWs irrespective of the In:Ga ratio of the precursor and other growth conditions such as temperature, flow rates, and III-V ratio. This kind of core-shell architecture is required for many types of NW-based devices, but is usually achieved by sequentially growing the core and the shell under different conditions. Here, the core-shell structure formation is spontaneous, monolithic, and abrupt, making it ideal for applications where this coaxial architecture is required.

Figure 2. Schematic illustration of the spontaneous formation of InAs-InGaAs core-shell NW structure on graphene substrates versus homogeneous InGaAs NWs on noncommensurate 2D molybdenum sulfide (MoS₂) substrates. (Courtesy of J. D. Wood.)

We used analytical scanning transmission electron microscopy experiments to explore the structure and composition of these core-shell NWs at various stages of growth. We found that spontaneous segregation was initiated at the onset of crystal assembly. We attribute the initiation of phase segregation in this system to the nature of van der Waals epitaxy and the unique commensurate relationship of the atomic lattice of binary InAs with graphene. The cubic InAs lattice along the 〈110〉direction corresponds to a near-integer multiple of neighboring carbon atom distances within the substrate, so that InAs and graphene represent an almost lattice-matched composite. However, when Ga is introduced, a deviation from ideal epilayer lattice registry with the substrate occurs, and the lack of dangling bonds on graphene prevents strain accommodation. Subsequently, nucleation of InAs occurs on graphene first, while Ga incorporates as adatoms on the sidewalls of the InAs core, forming a ternary InGaAs shell.

To investigate this phenomenon further, we also grew InGaAs NWs on molybdenum sulfide (MoS₂) flakes under identical conditions as on graphene. In this scenario, no lattice registry exists between MoS₂ and the InGaAs crystal or its binary constituents. The result is the self-assembly of a compositionally tunable, single-phase InGaAs system, as expected for conventional van der Waals epitaxy. Further, when we removed this freedom of self-assembly by growing InGaAs NWs on graphene using an alternative vapor-liquid-solid (VLS) method with gold-seeded nucleation, we found that again no phase segregation occurs. Thus, VLS growth of InGaAs NWs on graphene also leads to tunable compositional phase purity.

This is the first time that the importance of lattice matching in van der Waals epitaxy has been experimentally tested and shown to be as important as it is in bulk single-crystal-on- single-crystal epitaxy. Van der Waals epitaxy offers the promise of a new class of hybrid nanomaterials by combining functional 1D nanostructures with 2D sheets and layered 2D superlattice heterostructures. The epitaxial growth of single-crystal membranes on 2D sheets can be achieved while preserving perfect interfaces and high crystal quality. In addition, combinatorial architectures may be realized through the serial addition of NW array-on-graphene layers to conventional photonic and opto- electronic devices. These include supplementary sensors, emitters, and energy conversion components. Our future work will focus on novel photovoltaic solar cell and LED architectures using InGaAs NWs grown on graphene substrates.