Nanoscale patterning of epitaxial single crystals

In nanostructured energy conversion and storage devices—such as electrodes in lithium batteries or dye-sensitized solar cells—efficient charge transport is critical for performance. Charge trapping at grain boundaries limits the efficiency of such devices, but it can be avoided with the use of single-crystal nanostructures. Additionally, epitaxy (i.e., the orientation of crystal growth) influences many physical properties of different materials. Thus, controlling the direction of crystal growth is of critical importance. Here, we describe the use of soft matter self-assembly to direct the growth of single-crystal nanostructures with epitaxial relationships to silicon substrates.

Self-assembly directed by soft matter—such as block copolymers—combines the field of polymer science with inorganic materials. The natural phase separation of the copolymer drives self-assembly, ordering inorganic sol nanoparticles into nanostructures.² In turn, the inorganic phase adds robustness and functionality that typical polymers do not display. Additionally, a bottom-up approach to self-assembly provides a low-cost approach to large-area nanostructure fabrication and, in contrast to traditional lithographic techniques, inherently allows 3D structure formation.

We combined block copolymer self-assembly-directed nanoporous template formation with pulsed excimer laser irradiation-induced transient melting.¹ While block copolymer-directed inorganic structure formation, laser-assisted direct imprinting, and epitaxial nanostructure formation have all been individually demonstrated, the combination of these techniques in a single approach is novel.^2–4 In doing so, we can fabricate materials with adjustable pore sizes (5–50nm) and access unique structural and compositional parameters. Such flexibility may lead to new, highly efficient energy conversion and storage devices, such as improved solar cells.⁵

We began our synthesis with an inverse, hexagonally arranged, block copolymer structure-directed aluminosilicate monolayer template (see Figure 1). We next etched away the native silicon oxide layer of the silicon substrate, prior to depositing an amorphous silicon over-layer. We found that a clean interface between the silicon substrate and backfilled amorphous material was key to obtaining single-crystal epitaxial nanostructures. Both the deposited amorphous silicon and nanoporous template were then irradiated using a 40ns xenon chloride excimer laser (308nm wavelength) to induce melting and subsequent crystallization. While the aluminosilicate template was transparent at this wavelength, the absorbed laser energy was sufficient to melt both the amorphous silicon over-layer and single-crystal silicon substrate. The entire duration of the laser-induced melt and solidification was ∼20–100ns, during which the molten silicon recrystallized epitaxially from the single-crystal silicon substrate. Finally, we dissolved the aluminosilicate template in concentrated hydrofluoric acid to leave single-crystal epitaxial silicon ‘nanopillars’ behind.

Figure 1. (A-E) Schematic illustration of single-crystal epitaxial nanostructure fabrication and (F) laser irradiation of template covered silicon substrate.¹ Reprinted with permission from the American Association for the Advancement of Science.

We verified high pattern transfer (∼90%) from the template—see Figure 2(A)—to silicon—see Figure 2(B)—using atomic force microscopy and Voronoi analysis. We confirmed the single-crystal homoepitaxial nature of the silicon nanopillars relative to the substrate using high-resolution transmission electron microscopy: see Figure 3(A) and (B).

Figure 2. Atomic force microscopy (AFM) height profiles of the (A) block copolymer directed aluminosilicate template and (B) resulting single-crystal homoepitaxial silicon nanopillars. The vertical height scale bar ranges between ± 15nm and scan area is 1 ×1μm². Voronoi tessellation construction of the AFM images (A) and (B) are shown in (C) and (D), respectively. The scan area for Voronoi analysis is 2×2μm². The color bar indicates the number of sides of the calculated polygon.¹Reprinted with permission from the American Association for the Advancement of Science.

Figure 3. (A, B) High-resolution transmission electron micrographs (HRTEMs) show the congruence of lattice fringes in the single-crystal silicon nanopillar and substrate, affirming the homoepitaxy relationship. (C) Plan view and cross-sectional (inset) scanning electron micrograph of the 3D interconnected pore network silicon nanostructure. (D) Corresponding cross-sectional HRTEM of the single-crystal epitaxial silicon nanostructure in (C).¹Reprinted with permission from the American Association for the Advancement of Science.

To the best of our knowledge, this is the first time that such nanostructured, porous single-crystal structures have been directed by block copolymer self-assembly. As a result, an enormous opportunity for innovative research has been opened. Our approach is highly flexible and scalable, both with respect to material selection and dimension. For example, we demonstrated that our process also works with metallic crystals. We made use of the porous nature of the template to laterally control and confine moderately lattice-mismatched metallic nickel silicide on silicon to generate single-crystal heteroepitaxial nanostructures. By reducing the contact area between the silicon substrate and nickel silicide to several tens of nanometers squared (for example, by forming nanopillars), we can prevent accumulation of stresses in the film that usually lead to de-lamination of lattice mismatched materials of this type. This innovation paves the way for gallium-arsenide single crystals on silicon materials, which could enable faster electronic devices. We also employed a block copolymer-directed niobium oxide template to grow a ∼100nm thick, single-crystal homoepitaxial silicon nanoporous film with a 3D interconnected pore network structure: see Figures 3(C) and (D). Such materials may be useful to generate battery electrodes.

With respect to the flexibility of our methodology to fabricate diverse structures, we recently reported the first successful application of an ordered bi-continuous anatase-titania double gyroid network. This polycrystalline structure was directed by a diblock copolymer and resulted in a solid-state dye-sensitized solar cell.⁶ The gyroid structure provides connectivity in 3D, enabling efficient charge transport as well as easy access to pores for backfilling purposes. If fabricated as a single-crystal 3D bi-continuous network, charge transport and mechanical properties would benefit from a lack of grain boundaries and the controllable, flexible ability to epitaxially grow along specific crystal axes (relative to the substrate). Such gyroid structures may lead to significant advances—for example, in photovoltaics—because of their optimal charge transport characteristics.

In summary, we have coupled block copolymer templates with laser irradiation to form porous, single-crystal homo- and heteroepitaxial nanostructures on a silicon substrate. We can leverage the innate properties of soft matter self-assembly to fabricate ordered nanostructures with 3D periodicity in the bulk, which is not easily achievable using 2D lithographic processes without significant processing complexity and cost. Importantly, our method is compatible and adaptable with conventional semiconducting processing technologies and materials. In future work, we will adapt this approach to other inorganic materials, such as germanium, and will test the resulting structures in advanced energy generation and storage applications.