Introduction
As a single-layered carbon material, graphene exhibits high mechanical strength, large surface area, superior chemical stability, and extraordinary electrical and thermal conductivities, thus finding application in energy storage, sensors, etc. (Y. Zhu et al., 2010). Recently, graphene-based materials have shown promise as bio-scaffold owing to their biocompatibility, conductivity, anti-inflammatory properties, and differentiation effects during cell growth in multiple cell types (Sahni et al., 2013). The ultra-low density of graphene leads to easy biodegradation, and its high porosity enhances oxygen diffusion. All these factors collectively help graphene mimic the physiological environment better than many hydrogels and polymers (Loeblein et al., 2016). However, most of the potential applications require a large specific surface area for cells and chemical compounds to attach, along with a macrostructure to operate, connect, and examine. In order to circumvent the limitations of 2D materials and transfer the outstanding properties into three dimensions, there is an emerging need to assemble graphene into a 3D structure with specific dimensions and a well-controlled micro-structure (C. Zhu et al., 2015). Since graphene is hydrophobic, it can be modified with oxygen-containing groups, such as hydroxyl and carboxyl groups, by a reversible oxidation process (Hummers’ method), thereby forming graphene oxide (GO) (Hummers Jr & Offeman, 1958), which has enhanced hydrophilicity, dispersity, and forming property. GO retains most properties of graphene, except conductivity, which can also be easily recovered by reduction reaction with hydroiodic acid or ascorbic acid. Transformation of GO into a 3D structure, based on aqueous solution or gel, can be easily controlled in an inexpensive way, thus making it a preferred method for assembling graphene-based materials toward multi-functional applications.
Conventional forming methods for graphene-based materials include hydrothermal method and chemical vapor deposition (CVD)(Xu, Sheng, Li, & Shi, 2010), among many others. The hydrothermal method, developed by Xu et al, can fabricate self-assembled GO into 3D structure, with good conductivity and mechanical strength; however, the shape is defined by the container and pores are formed randomly (Xu et al., 2010). The graphene foam (GF), developed by Chen et al through CVD, is widely used as a bio-scaffold for hepatocytes, osteoblasts, chondrocytes, etc., and shows good biocompatibility and biodegradability (Loeblein et al., 2016; Xie et al., 2018; Yocham et al., 2018). Besides having similar disadvantages as the product from hydrothermal method, this material turns fragile under low pressure, and the method is not scalable.
Compared to the above-mentioned methods, 3D printing possesses the ability to fabricate complicated and well-controlled 3D shapes through additive manufacture process with the help of computer-assisted design (CAD) (Wang, Jiang, Zhou, Gou, & Hui, 2017). Mainly, two kinds of 3D printing technologies are used in GO printing: freeze-casting 3D printing and extrusion-based 3D-printing technique. Zhang et al had developed the freeze casting 3D printing method, which selectively solidifies the droplets of GO suspension into the cold sink (-25 ℃) while the water and low-viscous GO suspension are printed in a drop-on-demand mode (Zhang et al., 2016). While the low concentration of GO and drop-by-drop printing can achieve high precision, it can also result in low mechanical strength of the structure, low speed of printing, and large deformation in lyophilization process. Extrusion-based 3D printing, on the other hand, directly accumulates the ink into the 3D structure with higher efficiency and strength. Nevertheless, this method requires high viscosity, high yield stress, and a shear-thinning behavior of the ink material (Jiang et al., 2018). Normally, GO aqueous solution is of low viscosity, and requires further treatments. Jakus et al and Yao et al had achieved printable GO ink property via concentration using evaporation and lyophilization, respectively; however, both the methods were highly time-consuming (Jakus et al., 2015; Yao et al., 2016). Centrifugation is a fast and inexpensive way for concentrating, although the highest achievable concentration depends on the speed limit of the centrifuge (Jiang et al., 2018; C. Zhu et al., 2015). Based on centrifugation, Zhu et al had used 10–20 wt% silica as the viscosifier to increase the storage modulus of GO by an order of magnitude and achieve a highly compressible 3D structure (C. Zhu et al., 2015). However, silica is removed by hydrofluoric acid, which can, in turn, induce cytotoxicity. Jiang et al used 15 mmol/L CaCl2 as a crosslinker and achieved high-resolution printing (Jiang et al., 2018). Almost all mentioned 3D printing methods used lyophilization as the final treatment. However, the diameter, distribution, and orientation of pores, which are crucial parameters for cell attachment and proliferation, remain to be explored in detail. Multiple organic compounds (PEO, HPC, etc.), including divalent and trivalent ions (Ca2+, Mg2+, Cu2+, Pb2+, Cr3+, Fe3+), have been proven to induce GO gelation; however, only seldom have been studied for rheological properties, printability, or toxicity (Bai, Li, Wang, & Shi, 2011). Therefore, 3D printing of GO is still in need of further research and optimization, for designing a GO 3D structure with better microstructure and stable mechanical strength, for advanced application in tissue engineering. We herein proposed our work of improving the mechanical strength of GO by adding trivalent ions with optimized concentration as crosslinker. Fe3+ was selected due to its low cytotoxicity and high coordination number, and was used as crosslinker to fabricate the three-dimensional GO structure with both macro and micro porosity.
In order to characterize and optimize GO hydrogel, rheological properties of pure GO, following centrifugation, were tested first. As shown in fig. 1a, oscillatory measurements proved GO to be in gel state under low shear stress, with a storage modulus of 519 Pa and loss modulus of 111 Pa. The sol-gel transformation point was 50.5 Pa. Rotational measure proved the shear thinning behavior of GO hydrogel (fig. 1b), and the shear stress-shear rate relation coincided with the Herschel-Bulkley non-Newtonian fluid model (fig. 1c) (Truby & Lewis, 2016). A yield stress of 56.9 Pa was calculated by curve fitting.