Fabrication of Three-Dimensional Graphene-Based Polyhedrons via Origami-Like Self-Folding

For the cubes fabricated with CVD graphene, because each face of a given cube is designed with an outer frame surrounding a ~160 × 160 µm² area of free-standing graphene, a single sheet of monolayer graphene does not have the necessary strength to permit parallel processing of the cubes. For this reason, graphene membranes consisting of three layers of CVD graphene monolayer sheets are produced via three separate graphene transfers using multiple PMMA coating/removal steps. On the other hand, for GO membrane preparation, we use individual GO sheets in water, obtained via a modified Hummer's method²⁷.To pattern the GO membranes, traditional photolithography followed by a lift-off process via flood exposure is used. The process uses a flood exposure after traditional photolithography but before GO membrane deposition. After GO spin-coating, a lift-off process is then performed in developer to remove the unwanted GO areas. Some developers contain sodium hydroxide (NaOH) aqueous alkaline solution which etches aluminum and Al₂O₃. Therefore, a NaOH free developer should be used. For this work, the specific developer used to fulfill this requirement is developer-2 solution.

The frames of the 3D cubes supporting the graphene-based membranes are made of PR-3 due to its high mechanical and thermal stability as well as high optical transparency²⁹. It is known that the thermal and mechanical stability of PR-3 depends on the cross-linking process³⁰. The maximum cross-linking of PR-3 occurs when it is hard-baked over ~200 °C. After the hard-baking, the dynamic modulus of PR-3 improves, indicating that the structures have more mechanical strength during dynamic motion and thus are more mechanically stable. In fact, when the heat is applied to the cubes (or samples) for self-folding, the PR-3 frames maintain their original shape. Another source of potential damage is the deposition of Ti patterns as they might produce compressive stress on the graphene membranes; however, the demonstration of undamaged graphene membranes after self-folding indirectly indicates the mechanical stability of the PR-3 could contribute to the conservation of the graphene membranes (Figure 3, Figure 4). Furthermore, the photo-definable property of PR-3 allows easy control of the sizes and shapes of the 3D cubes, along with easy control of the folding angle of 3D structures for the realization of diverse 3D structures including semi-3D structures.

In the origami-like self-folding principle, a surface tension torque is produced to fold a 2D net structure via the reflow of the hinge materials (e.g., thin metal films or thermal-sensitive polymers)^26,31. The surface tension of the polymer PR-2 hinges is lower (~0.03 N/m) than that of metal hinges (e.g., solder ~0.5 N/m)^26,28. The lower surface tension produces less rotational torque when the 2D nets are folded compared to 2D nets with metal hinges^26,31. The lower torque could reduce the stress on the tri-layer graphene-based membranes during the self-folding process. The 3D graphene-based cubes are folded at ~80 °C (Figure 3), in which the hinges reflow at their melting point (for metal solder hinges, the melting point is ~230 °C)²⁶. Remarkably, under an optimized process, this approach shows a highest yield of ~90%.

During the lithography process and self-folding, the spatial stress on graphene membranes induces delamination, buckling, cracking, and/or ripping. For example, (i) when the 2D nets with graphene membranes are released from the sacrificial layer, strong Van der Waals forces between the graphene and sacrificial layer (including Cu or even many other substrates) can be generated, resulting in broken graphene membranes; and (ii) during the self-folding in liquid, surface tension force, fluidic force, and gravitational force cause cracking and buckling of graphene membranes. A Cu layer is used for a sacrificial layer and an additional patterned Al₂O₃/Cr layer is used as a protection layer to shield the graphene-based membranes. Initially, a thin (10 nm thick) Cr layer has been used as a protection layer. However, the thin Cr layer shows buckling structures since the mechanical properties of the Cr layer are not strong enough to hold the graphene-based membranes when the structure is released from the Cu sacrificial layer. Later, to resolve this issue, 100 nm thick Al₂O₃/10 nm thick Cr layers are added on top of the 10 nm thick Cr/300 nm thick Cu sacrificial layer as described above. As a result, the protection layer allowed retention of the graphene membranes throughout the fabrication process and self- folding. The protection layers on the 3D cube can be removed after self-folding by an appropriate etchant without damage to the graphene membranes.

The 3D CVD graphene-based cube image presents a highly transparent, free-standing, enclosed architecture (Figure 4a) with no noticeable cracks, ripples, holes, or other damage on the membranes (from the zoomed-in image, Figure 4b). As described above, using the same approach used to produce the 3D CVD graphene-based cubes, we also successfully demonstrate the fabrication of cubes with membranes comprised of ~10 layers (~10 nm thick) of GO sheets (Figure 4e, 4f). In addition, Ti surface patterned 3D cubes are very stable (Figure 4c, 4d for graphene and Figure 4g, 4h for GO), and the demonstration of varied surface modifications with dissimilar designs on the different faces suggests a versatile strategy for the construction of 3D multifunctional devices with heterogeneous integration of different combinations of materials. As a result, the 3D graphene-based cubes show (i) free-standing CVD graphene and GO window membranes comprised of layered structures (no composite formation); (ii) enclosed but hollow structures that do not require an additional support or substrate; and (iii) surface modifications via metal patterning on the graphene or GO surfaces with any desired patterns, because our approach is compatible with conventional lithographic processes.

Raman spectroscopy is well-established as an effective, noninvasive method to characterize graphene and related materials, and it can supply a wide variety of details about graphene-based samples such as thickness, doping, disorder, edge and grain boundaries, thermal conductivity, and strain. Furthermore, this characterization method is flexible as it can be applied to a sample in various environmental conditions^32,33,34,35. Therefore, if there are significant changes in the structure of graphene, we should be able to see changes in Raman peaks positions or intensities after self-folding. As shown in Figure 5a-5b, no significant changes in the peak positions and intensities can be seen after the self-folding, since the Al₂O₃/Cr protection layers help to shield the graphene-based membranes (both CVD graphene and GO) during fabrication. However, as shown in Figure 5c, when protection layers are not used, graphene membranes are damaged during the self-folding, resulting in a higher D band (~1340 cm^-1) and a lower 2D band (~2690 cm^-1). The quantitative information on the graphene defects can be analyzed by the peak intensity ratio of the D band and G band (I_D/I_G): low value I_D/I_G means low-defect graphene. From Figure 5a we calculate the I_D/I_G values of the 3D graphene to be ~0.65 which is comparable to other CVD graphene multilayer sheets³⁶. Therefore, these observations indicate the self-folding process did not create significant changes in the CVD graphene and GO membranes (the materials retain their intrinsic properties, and no chemical intercalation between layers occurs), demonstrating the robustness of the reported method.

In addition to producing hollow, free-standing, polyhedral cubes, the self-folding method employed here allows for surface patterning, consisting of metal, insulator, and semiconductor materials on the 2D graphene membranes, to be applied to the cubes while maintaining the intrinsic properties of the graphene. This allows for the development of electronic and optical devices, including sensors and electric circuits, utilizing the numerous advantages of 3D configurations. Furthermore, since the processes used are not limited to only graphene-based materials, this method may be applied to other 2D materials such as transition metal dichalcogenides and black phosphorus, thereby allowing our fabrication approach to be harnessed in developing next-generation 3D reincarnations of 2D materials.

The high-temperature (~80 °C) required by the folding mechanism could be problematic in biomedical applications unless the process can be further optimized to reduce the folding temperature. Additionally, the PR-2 hinge material is not a biocompatible material. Future studies will focus on developing biocompatible hinge materials that respond at low-temperature (or low-energy) such as polyesters and synthetic hydrogels. We have recently been able to manufacture similar structures via a remote-controlled self-folding mechanism that could be useful in this respect³⁷.