Preparation and rheological properties of GO inks
To verify the effect of GO molecular structure on the rheological properties of GO hydrosol and then prepare GO ink suitable for high-precision 3D printing, GO was synthesized by modified Hummers’ method22,23. The mixture obtained after the oxidation of sulfuric acid-graphite intercalation compound (step II oxidation24) was quenched into deionized water. The quenched product was then heated at 98 ℃ for 0, 3, 6, 9, 12 and 15 min respectively, and a 3 wt% H2O2 solution was subsequently dripped into the solutions.
From Fig. 1a, the GO solutions get gradually darker with increasing heating time, reflecting the structural transformations in GO. After washing and drying, the obtained GO was analyzed through FTIR, XPS, Raman and XRD spectroscopies. FTIR (Fig. 1c) shows the characteristic peak of epoxy group at 880 cm1 which gradually weakens with increasing the heating time25–27. XPS spectra of C1s (Fig. 1d, e) show that the peak corresponding to O–C = O bond (289.2 eV)28 in carboxyl groups is slightly larger at 15 min heating-time compared to that at 0 min. Therefore, the results of FTIR and XPS indicate that with increasing heating time, the amount of the epoxy groups decreased significantly, and the carboxyl groups increased gradually. Based on the Bragg’s law (2dsinθ = λ), the position of XRD peak (θ) determines the interlayer distance (d), and the full width at half maximum (FWHM) reflects the ordering, which is correlated with the alignment and flatness of the nanosheets29–31. Sharp XRD peaks result in small FWHM values indicating gentle fluctuation and orderly stacking of GO nanosheets. From XRD results (Fig. 1g, h), increasing the heating time from 0 to 15 min results in a decrease in interlayer spacing from 9.3 Å to 7.8 Å, while the FWHM increases from 0.39° to 0.88°. The Raman spectrum (Fig. 1f) shows the typical D peak (1,353 cm− 1), G peak (1,600 cm− 1), 2D peak (2,698 cm− 1) and D + G peak (2,945 cm− 1)32 with the ID/IG intensity ratio increasing from 0.98 to 1.12 with heating time. It can be therefore concluded from XRD and Raman spectra that the interlayer spacing, the smoothness of GO sheet layers and degree of layers stacking order33 decrease with the heating time (Fig. 1b). This is due to the reduction of epoxy groups (as confirmed by FTIR) which act as the main support of the carbon grid and thus the interlayer interaction of neighboring GO sheets improve through e.g. hydrogen bonding.
The solutions (2 wt% concentration) were then subjected to rheological studies. All the GO hydrosols exhibit the same shear-thinning non-Newtonian fluid behavior (Supplementary Fig. 1a), that is, the viscosity decreases with increasing shear rates, which is necessary for a printable ink to flow continuously. The 0 min heating-time GO hydrosol reveals the highest viscosity among the six solutions. With the increase of heating time, the viscosity first decreased and then increased slowly. Owing to the presence of numerous epoxy groups on the surface of the GO hydrosol without heat treatment, the interaction between the layers is weak and the graphene lamellae can be fully extended in the solvent phase (water). This enables a full play to the water-binding ability of GO sheets reducing the fluidity of the solvent and increasing the modulus as well as the viscosity. At the initial heating times the amount of epoxy groups in GO decreases sharply improving layer interaction and agglomeration resulting in enhanced solvent fluidity and decrease in viscosity. Meanwhile, prolonging the heating time promotes the formation of carboxyl groups, which enhances the hydrophilicity and water-binding capacity. This leads to a decrease in hydrosol fluidity and thus increase in viscosity. However, even in the absence of heating, the aqueous GO hydrosols cannot realize optimal properties to achieve self-standing precursor inks. Therefore, we added traces of Ca2+ ions as gelator to realize GO hydrosol thickening without increasing the sol concentration which could be detrimental to printing accuracy11,14.
The precursor ink for DIW should exhibit a shear thinning behavior to be extruded continuously and be able to hold a 3D printed shape. For 0 and 3 min heating-time, the viscosity of GO hydrosol after Ca2+ addition is higher than that of neat GO hydrosol at the shear rate of 0.01 ~ 100 (Supplementary Fig. 1b, c). For 6, 9, 12 and 15min heating-time GO hydrosols, adding Ca2+ concentration as low as 5 mM leads to phase separation (Supplementary Fig. 2b) and GO lamellae agglomeration due to the formation of stable Ca2+-carboxyl complexes. Meanwhile, 0 min and 3 min heating-time GO hydrosols with 7.5 mM Ca2+ concentration (labeled as 0-7.5 mM, 3-7.5 mM, respectively) exhibited the highest viscosity of around 100 Pa·s (Fig. 2a) without phase separation (Supplementary Fig. 2a), which guarantees the printability of complex 3D structures with better accuracy.
As shown in Supplementary Fig. 1b, c, when the Ca2+ concentration is below 10 mM, the viscosities of 0 and 3 min heating-time GO hydrosols increase with Ca2+ concentration. This results from the cross-linking ability of Ca2+ to form a viscous hydrosol through coordination interactions with GO functional groups. For Ca2+ concentration higher than 10mM, the sol viscosity decreases sharply at high shear rates due to the strong bridging effect of Ca2+ that leads to GO lamellae agglomeration and phase separation. The higher the concentration of Ca2+, the higher the phase separation and water precipitation. This issue will damage the homogeneity of the sol, subsequently blocking the nozzle during extrusion, and hindering 3D printing.
We performed further rheology tests on the hydrosol with optimal Ca2+ concentration of 7.5 Mm at 3min-heating time (3-7.5 Mm) exhibiting high viscosity without phase separation. Figure 3b compares the storage modulus (G′) (elastic response) and loss modulus (G″) (viscous response) of 3min-heated pure GO hydrosol (3 − 0 mM) with those of 3-7.5 Mm. The 3 − 0 mM dispersion (2 wt%) exhibits a platform G′ around 1000 Pa and a yield stress (τ) around 20 Pa. Adding 7.5 mM of Ca2+ ions increases significantly both G′ and τ to near 40000 Pa and 130 Pa, respectively. Moreover, the value of G′ is six times that of G″. In the plateau region, a large storage modulus indicates that the inks are relatively stiff exhibiting a solid-like response. After the crossover point between the storage and loss moduli, the loss modulus becomes higher than the storage modulus with an increase in shear stress of > 80 Pa. This indicates that viscous deformation begins to dominate the viscoelastic ink properties.
Figure 2c shows that the G′ and G″ values are nearly unchanged in the oscillatory time sweep test (1% oscillation amplitude, 1 Hz, 45 min), which indicates the rheological stability of the sol. Figure 2d shows the rheological behavior under cyclic loading before and after the yield point. The relationship between the storage modulus and the loss modulus of the ink varies according to the applied shear force, showing reversible yield and excellent mechanical stability. These features confer the inks with smooth extrusion and shape retention on the print receiving platform. Significantly, 0-7.5 mM GO hydrosol shows similar rheological properties to 3-7.5 mM hydrosol, however, the graphene lamellae in the heat-free GO hydrosol can be fully extended in water which makes its concentration process a difficult task (Supplementary Information Table.1). Therefore, we chose 3-7.5 mM GO hydrosol as the ink for 3D printing of high-precision graphene aerogels due to its optimal rheological properties, homogeneity and easily adjustable concentration which overcomes the high concentration requirement of GO dispersions.
3D printing of GO aerogel microlattices
For the printing process, the GO hydrosol with excellent rheological properties (3-7.5 mM) was loaded into a movable syringe controlled by a programmable robotic depositing arm and then screw extruded through a nozzle to deposit 3D structures (Fig. 3a). As shown in Fig. 3a, the GO ink exhibits a shiny gold color and is extruded continuously through a 75 µm needle suggesting the high quality and low concentration of the GO sol. Next, the microlattices were freeze dried to prevent the shrinking of the 3D architecture and to remove water and get solid porous aerogels34. After freeze-drying, thanks to the high quality and low concentration of GO hydrosol, the graphene oxide aerogel lattice (GOAL) structure is brownish yellow with well-defined frameworks in both top and side views indicating high printing accuracy (Fig. 3b-e). Particularly, the side view (Fig. 3e) shows that the assembled multi-layer structure is regular without collapsing and thus the printed structure was strictly consistent with the initial design. Scanning electron microscopy (SEM) of GOAL (Fig. 3f-h) confirms that the printing accuracy (in terms of monofilament diameter) of the obtained GOAL is 70 µm which is significantly more precise than previously reported values (200–1000 µm)12,14,15. The grids are clearly visible, and the aerogel presents a porous foam internal structure.
Chemical reduction treatment through hydroiodic (HI) acid can endow GO aerogel lattice with conductivity in a facile way. During the reduction treatment, the color of aerogels turns from brownish yellow to black with a minimum shrinkage of 10% (Fig. 3i). Meanwhile the structural integrity of the frameworks remains stable. From the SEM images (Fig. 3k) the pore structure in the microlattice surface of RGOAL almost vanishes, showing a denser layer stacking and smaller gridline diameter compared with GOAL (Fig. 3j). These features result from the reduction of functional groups of GO and stronger interactions between GO lamellae. Moreover, the one-step chemical reduction with HI conferred the RGOAL an electrical conductivity of 747 S/m, which is similar to the former reports14,34.
EMI shielding and sensing tape based on GO aerogel microlattices
High precision 3D printing can exert a high level of control over porosity and morphology of GO (RGO) scaffolds and microlattices by tuning the processing parameters which could be very promising for EMI shielding35,36. Through 3D printing and chemical reduction, we constructed single-layer and multi-layer RGOAL with square apertures of 0.75, 1.00, 1.25, 1.50, 1.75 and 2.00 mm (labeled as RGOAL-0.75…RGOAL-2.00), and then tested their electromagnetic shielding performance. According to the transmission line theory37, EMI SE (total shielding effectiveness) of the material can be calculated from the parameters of PI (received power), PR, (reflected power) and PT (transmitted power) via the following formulas38.
$$\text{R}={\left|{\text{S}}_{11}\right|}^{2}=\frac{{P}_{R}}{{P}_{I}}$$
$$\text{T}={\left|{\text{S}}_{21}\right|}^{2}=\frac{{P}_{T}}{{P}_{I}}$$
$$\text{A}=1-\text{R}-\text{T}$$
$${SE}_{\text{R}}=-10\text{log}\left(1-\text{R}\right)=10\text{log}\frac{1}{1-\text{R}}$$
$${SE}_{\text{A}}=-10\text{log}\left(\frac{\text{T}}{1-\text{R}}\right)=10\text{log}\frac{1-\text{R}}{\text{T}}$$
$$SE={SE}_{\text{A}}+{SE}_{\text{R}}$$
The S-parameters, S11 (reflection) and S21 (transmission) were measured by Rohde & Schwarz ZNB 40 Vector Network Analyzer (VNA) with a WR-90 waveguide in TE10 mode in the X-band (8.2 to 12.4 GHz). Prior to the measurements, the VNA was calibrated by using the TRL (thru-reflect-line) method. Moreover, according to the corrected electromagnetic shielding theory37, the contribution of reflection and absorption to the shielding effect of 3D printed GO aerogels is evaluated by comparing the power coefficients, R (Reflectivity) and A (Absorptivity).
Decreasing the RGOAL square aperture, the grid becomes denser and SE increases gradually (Fig. 4a). However, decreasing the square aperture below 1.5 mm and thus increasing the grid compactness does not have any significant influence on SE. RGOAL-1.50 with an extremely low weight of 2.1 mg/cm2, reaches a maximum shielding of approximately 12 dB, which corresponds to an attenuation of 93.7% as well as a high absolute shielding effectiveness (SSE/t) of 5714.3 dB cm2 g− 1. In terms of power coefficients shown in Fig. 4b, the reflectivity R of RGOAL-1.50 is slightly higher than absorptivity A at 8.2 ~ 9.6 GHz, while between 9.6 ~ 12.4 GHz, they are basically consistent. In general, the contributions of R and A to SE are roughly equal. As in the case of SE, the power coefficients R and A at 10 GHz remain almost constant as the square aperture is decreased below 1.50 mm while R/ (R + A) is kept at about 50% (Fig. 4c). When the square aperture is larger than 1.50 mm, reflectivity decreases, while absorptivity remains almost unchanged while R/ (R + A) decreases to less than 50%.
Subsequently, we printed RGOALs incorporating two to five layers. Obviously, SE is promoted with the increase in the number of layers reaching nearly 35 dB for the five layers RGOAL-1.50 (Fig. 4d). In terms of power coefficients (Fig. 4e), the shielding effectiveness of five layers RGOAL-1.50 is dominated by reflectivity. Increasing the number of layers promotes the reflection of the incident electromagnetic wave, thereby reducing the wave entering the material and hence absorptivity (Fig. 4f).
We then move forward to transfer the RGO microlattices onto adhesive tape (Fig. 5a). The use of a flexible and functional substrate as tape constitutes a simple and large-scale strategy to replicate RGOAL properties into stretchable devices without undergoing significant degradation. Moreover, these replicas are not limited to flat surfaces and could be cut into different sizes and shapes and thus applied anywhere required. The tape based on RGOAL is highly durable, withstanding repeated bending, folding and winding and therefore retaining the basic properties of the adhesive tape (Fig. 5b). Supplementary Fig. 3a and b show the RGO foam structure flattened in the tape. Under extreme folding (Supplementary Fig. 3c) the internal foam structure at the crease remains without exhibiting cracks and fracture (Supplementary Fig. 3d).
Figure 5d shows the EMI shielding properties of RGOAL tape with 1.50 mm square aperture. The SE of single-layer RGOAL tape preserves the values of RGOAL reaching approximately 12 dB with RGO weight of only 2.1 mg/cm2 and high SSE/t of 5714.3 dB cm2/g. Meanwhile, the SE of five-layer RGOAL tape is more than 20dB in the whole X-band (8.2 ~ 12.4 GHz) reaching a maximum value of 28dB (> 99% shielding) with only 700 µm in thickness (SSE/t = 2687.7 dB cm2/g) and RGO weight of 10.5 mg/cm2. As pointed out early, as the number of layers increases, reflection becomes the dominant shielding mechanism (Supplementary Fig. 4).
The single-layer RGOAL tape was then subjected to bending resistance test where d represents the distance between two ends of the tape during the bending process (Fig. 5b). The RGOAL transferred onto the tape still maintains the piezoelectric properties of RGO-based aerogels15,39 and thus as the tape is bent the resistance R changes accordingly. Figure 5e shows the variation of |ΔR/R0| under cyclic bending (10 cycles) at different bending degree defined as |Δd|/L0 (where L0 represents the length of the tape when it is not bent). The change in resistance enlarges with the increase of bending degree |Δd|/L0, while it also shows excellent stability for each cycle. Even when d = 0 (Δd|/L0 = 100%), i.e. when both ends of the tape touched, the change in resistance remains stable. The resistance can also be restored after unloading, indicating excellent bending toughness and stability of RGOAL structure. Figure 5f shows the sensitivity of the tape which is evaluated by the gauge factor (k) calculated from the relative resistance variation |ΔR/R0| versus |Δd|/L0. In the range of |Δd|/L0 = 0-0.4, the tape exhibits a k1 value of 90 while declining to k2 = 38 in the range of 0.4-1. When |Δd|/L0 reaches 100%, the relative resistance variation decreases by more than 70% (as the RGO foam is compacted during bending the conductive path increases decreasing the resistance), showing high sensitivity to bending strain. Due to the improved printing resolution, more functional units exist per unit area which can sensitively perceive tiny resistance variations during bending deformation. Therefore, RGOALs replicated on an adhesive tape could be also used as a rapid response, conformable, lightweight, and low-cost resistive bending strain sensor. Our approach opens up possibilities for unleashing the attributes of 3D printed GO aerogels into high-performance functional devices that can be conveniently modified to suit a wide range of applications.