Where C, L, and A are the circularity, perimeter, and area of the grids respectively. For 3D printing of hydrogel, the ideal gelation condition should demonstrate square shape of grids (Pr = 1), with higher and lower Pr values representing over-gelation and under-gelation states, respectively. In fig. 1f & 1g, the measured printability coincided with rheology results, in which 8 mmol/L samples showed Pr values closest to 1 (proper gelation). Fe³⁺ ion, at the concentration of 8 mmol/L, exhibited the best results in both rheology and printability, which was, therefore, selected as the optimal concentration for further printing.

Then, structures with different filament distances (1 mm, 1.5 mm, 2 mm, 2.5 mm, and 3 mm) were printed in order to test the forming property of Fe³⁺-crosslinked GO hydrogel under different printing conditions. During the printing process, parameters including the nozzle diameter, extrusion speed, scanning speed, and nozzle height were adjusted to achieve an optimized printed structure. The grid within the printed network became a macroporous structure. Ultimately, under the conditions of all filament distances, an intact structure with three-way pores (along x, y, and z axes in fig. 2b) was obtained (fig. 2c).

Based on the stable and well-controlled structure with the help of Fe³⁺ ion, lyophilization properties could be further studied. There are three steps in lyophilization: freezing, main drying, and final drying. Printed structures were frozen under -80 ℃ and -20 ℃ to explore the relationship between freezing temperature and pore size. Pore structure with a mean diameter of 11.19 ± 5.46 μm (n = 107) was produced by freeze-drying at a uniform temperature field of -80 ℃ (fig. 3a, 3b). For -20 ℃ field, the diameter of pores was 22.41 ± 11.55 μm (n = 93) (fig. 3d, 3e). The distribution of pore size, as shown in fig. 3c & 3f, can be roughly fitted into Gaussian distribution. Additionally, samples were frozen at the temperature gradient field to test their feasibility to form oriented porous structures (fig. 3g, 3h, and 3i). Oriented pores with a mean width of 86.7 ± 19.0 μm were observed therein. During the freeze drying process, ice crystals were formed in the freezing stage that sublimed in the main drying stage. As a result, the space that used to be occupied by ice crystals became pores after lyophilization, shape, and size of the pores being decided by the crystallization process, and ultimately controlled by the freezing parameters. Compared to those obtained by freezing at -80 ℃, we obtained pores of approximately double the diameter (22.41 μm vs. 11.19 μm) at a higher freezing temperature (-20 ℃), which is in line with the mechanism of crystalline growth. The oriented pores in fig. 3g, 3h, and 3i proved the temperature gradient field to successfully guide the growth direction of ice crystal.

Cell attachment experiment was then conducted to test the performance of GO structures as a bio-scaffold using Human hepatoma cell line HepaRG, which is able to be differentiated and applied to build liver tissue model. According to the live/dead assay (fig. 4), we found HepaRG cells to have good viability (of 91.0 ± 7.2%) in the GO scaffold, as evident from the assistance of enhanced oxygen and nutrient transportation through the porous structures. More attached cells were found near the grid area, both on the surface and interior, which proved that the macro pores increased the specific surface area for cell attachment. According to DNA content measurement, the immobilized cells were approximately 3.06 × 10⁶ cells/cm³ in the GO scaffold. However, during z-axis scanning, few cells were detected in the interior, which could be attributed to either the opaqueness of graphene oxide or less cell distribution within. Therefore, the current thickness of 4 mm of the GO scaffold might be too far for the cell to travel through, even with the help of macro and micro pores; further improvement in design is expected in the future.

Appreciable cell viability and attachment behavior proved the biocompatibility of Fe³⁺-crosslinked GO structures and showed that Fe³⁺ at the concentration ≤ 8 mmol/L does not induce cytotoxicity. Consequently, such GO structures can be applied as promising bio-scaffold in tissue engineering. The remaining problem was the immobilized cell density, which was not comparable in magnitude to that seen in vivo (10⁸ to 10⁹ cells/cm³) due to the thick structure along the z-axis. Hence, further geometric optimization would be helpful for higher cell density. Two-dimensional (2D) graphene materials had demonstrated the capability of pre-concentrating the differentiation inducers, thus accelerating the differentiation of human mesenchymal stem cells (MSCs) (Lee et al., 2011). Therefore, we expect the differentiation of HepaRG into hepatocytes and bile duct cells could also be enhanced with the effect of concentrating dimethylsulfoxide (DMSO) in the 3D GO structure (Guillouzo et al., 2007). Moreover, 3D graphene showed anti-inflammatory ability in microglial cells while 2D graphene failed to do so (Song et al., 2014). Our GO scaffold may also be used to culture neural stem cells owing to the same property. All these novel properties are beneficial to future applications of GO scaffold in multiple cell types.

In summary, we improved the rheological behavior and printability of graphene oxide hydrogel by inducing ferric ion as the crosslinker and using micro extrusion-based 3D printing technology to 3D-print the hydrogel, at room temperature, with a wide range of filament distances and diameters. Additionally, controllable porous structure inside the graphene structures was achieved by varying the freezing parameters in lyophilization process. The Fe³⁺-crosslinked GO structure is of low cytotoxicity, on which HepaRG cells showed good viability and attachment behavior. Considering the excellent properties of graphene materials in cell culture and their anti-inflammatory effects, the current stable, well-controlled, porous, and easily fabricated GO structure would be a promising bio-scaffold for developingin-vitro liver tissue models.