2.1 Ink Formulation and Determination of Gel Content and Swelling Ratio

The base formulation used for optimizing the ink properties consisted of 5.0 g deionized water, 1.50 g (21.1 mmol) of the monomer AAm (30 mass% relative to water), 0.24 mmol of the cross-linker N,N’-methylenebisamide (MBA, 1 mol% of the monomer amount), and ammonium persulfate (APS), a thermal initiator for the free radical polymerization of the target hydrogel. The magnetic components were iron(II, III) oxide nanoparticles (NP). The NP amount was typically 0.4 mass%, and the laponite (L) concentration was varied from 1.5 to 6.9 mass%. The sample naming indicates the laponite and iron oxide mass percentage. A sample with 3.1 mass% L and 0.4 mass% NP was thus termed L3.1NP0.4.

For optimal shape fidelity and resolution of 3D printed hydrogels, it is important that DIW inks used do not flow after deposition onto the print bed, and that they polymerize near-quantitatively during post-printing treatment. In the presented work, the flow was prevented by laponite addition, and the printed ink structures were transformed into hydrogels by thermally induced free radical polymerization (FRP). As is well known, the reaction rate of FRP strongly depends on the monomer and initiator concentration. In the context of actuators, a suitable amount of cross-linker is needed to obtain a sufficiently stiff and tough material. To confirm that the chosen ink base formulation polymerizes well and cross-links to a sufficient extent (so that the actuators obtained can be handled without breaking) and within a suitable reaction time, its gel content was determined prior to optimizing its rheological properties. For that purpose, rectangular hydrogel sheets of 2 by 2 cm were fabricated by molding using an ink formulation with the above specified amounts of AAm, MBA, and APS, as well as 4.6 mass% L and 1.5 mass% NP (= L4.6NP1.5). Each hydrogel sheet was cross-linked at 50 °C for a determined timespan. The mass of the dry gel before and after extracting the unpolymerized monomer was determined gravimetrically to calculate the gel content ($$Gel\;content\;( \% ) = \frac{{dry\;gel\;after\;extraction}}{{dry\;gel\;before\;extraction}}\; \cdot 100$$). Mechanically stable hydrogels were already obtained after 20 min of thermal treatment. In the plot of the gel content versus the cross-linking time, a plateau was reached after 40 min (Figure 2a). In the light of these results, the standard cross-linking time was set to 40 min, where a gel content of 95% was observed. The swelling ratio was determined by measuring the dry mass and the wet mass of the polymer hydrogels ($$Swelling\;ratio\left( {in\% } \right) = {{\left( {wet\;mass - dry\;mass} \right) \cdot 100} \over {dry\;mass}}$$). As an average over three samples, this gave 397.0 ± 6.7%.

2.2 Effect of the Laponite Concentration on the Iron Oxide Dispersion and Ink Viscosity

Laponite served a dual purpose in this work. First, varying the laponite concentration allowed to tune the rheological properties of the system. Second, the interaction of the NPs with laponite enabled their lasting dispersion within the ink, where they would otherwise precipitate within minutes. To investigate the effect of laponite concentration on the ink viscosity, it was varied from 1.5 to 6.9 mass% (relative to the overall mass of the base formulation consisting of water, AAm, and MBA). Up to 3.1 mass% laponite, the inks had a low viscosity and did not stabilize the NPs sufficiently. Lines printed using L3.1NP0.4 ink immediately flowed (Figure 2b). At a laponite concentration of 3.8 mass%, the ink viscosity visibly increased, and the nanoparticles did not sediment within 24 h (Figure 2b). At a laponite concentration of 4.6 mass%, the viscosity was sufficiently high to prevent flow when the sample vial was turned upside down, and lines printed using an L4.6NP0.4 ink retained their shape (Figure 2b). At 5.4 mass%, the ink viscosity was too high for a steady, continuous flow during printing, so the maximal laponite concentration for printing should be kept below that value.

The rheological behavior of inks with different laponite concentrations was quantified by rheology using a steady-state flow shear rate sweep test. The shear rate was increased step-wise from 1 to 10³ s⁻¹ (waiting for 30 s to equilibrate the system after each measurement) and the resulting viscosity at constant rotation was recorded (Figure 2c). In a log-log plot of viscosity versus shear rate, all samples with 3.8 to 6.9 mass% laponite had a linear reduction of viscosity with shear rate, thus confirming shear-thinning behavior. For example, when the data for L4.6NP0.4 was fitted using the Ostwald de Waele relationship, $$\eta = k \cdot {\dot \gamma ^n}$$, with $$k$$ and $$n$$ as fitting parameters, which describe the behavior of a fluid, k  =  59.26  ±  2.29 and a negative flow index n  =   − 0.92  ±  0.01 was obtained, which in this power-law confirms shear thinning behavior. A shear stress versus shear rate plot of L4.6NP04 also confirmed typical shear thinning behavior (Figure S1, Supporting Information). The curve could be fitted with the Herschel–Bulkley model, $$\tau = {\tau _0} + k \cdot {\dot \gamma ^n}$$, where $$\tau$$ is the shear stress, τ₀ the initial shear stress, and k and n are fitting parameters. The flow index n  =  0.54  ±  0.01, which is smaller than one, indicates shear thinning. The linear fit of the Bingham model $$\tau = {\tau _0} + {n_B} \cdot \dot \gamma $$, where n_B is the fitting parameter, does not follow the data.

Within the measured shear rate range, the viscosity of the L3.8NP0.4 ink was almost two orders of magnitude lower than that of the L6.9NP0.4 ink (at rest: 7.8 Pa⋅s vs 334.8 Pa⋅s). The viscosity at rest of L3.8NP0.4, 7.8 Pa⋅s, can be considered as the approximate minimum viscosity for shape fidelity when printing this system. The viscosity at rest of L5.5NP0.4 was 131.2 Pa⋅s, which should be considered as the approximate maximum viscosity value to avoid printing problems such as unsteady flow. Note that these viscosity limits are only applicable to the DIW system used in this work and need to be evaluated for different setups separately. To determine the relative viscosity decrease of the inks under shear, a shear thinning index was calculated by taking the ratio between the viscosity at 1 s⁻¹ and the viscosity at 10³ s⁻¹ ($$shear\;thinning\;index = \frac{{\eta ( {1 {s^{ - 1}}} )}}{{\eta ( {{{10}^3} {s^{ - 1}}} )}}\;$$). Figure 2d shows a plot of the shear thinning index versus the laponite concentration. That index increased with increasing laponite concentration and leveled off into a plateau at 5.7% laponite, i.e., this amount was sufficient to build a percolating superstructure through the ink volume. A laponite amount of 4.6% qualitatively gave the best shape fidelity and printing performance, indicating that there is no need to reach the maximum possible shear thinning index for the formulation to obtain a printable ink.

2.3 Temperature, Shear Rate, and Time Dependency of the Mechanical Properties of the Inks

While the viscosity of the inks had a clear shear-rate dependency at constant rotation, frequency sweeps from 1 to 10 Hz under oscillatory conditions had only slight effects on the magnitude of the shear moduli G′ and G′′ (measured at 0.1% strain, Figure 3a). Only for the highest laponite concentration (5.7 mass%), an increase of G′′ in the measurement range (from 1⋅10⁻¹ to 2.5⋅10¹ Hz) was observed. For 4.2 and 4.6 mass%, the value of G′′ remained almost constant (at ≈8 Pa) over the entire measurement range. G′ showed an increase for all concentrations, which leveled off into a plateau afterward. Importantly, in the entire measurement range, G′ was larger than G′′, indicating that the ink at rest behaved more like a viscoelastic solid than like a liquid. Temperature-dependent oscillatory measurements from 22 to 60 °C at a shear rate of 1 Hz using L4.6NP0.4 also showed that G′ was larger than G′′ over the entire shear rate range (Figure 3b). This is substantially different from the temperature dependency under dynamic conditions of the previously studied F127-based inks.^[14] Starting at a temperature of 27 °C and up to 50 °C, G′of these ink increased by almost two orders of magnitude compared to room temperature. Further, these inks had a cross-over from G′ <  G′′ to G′ >  G′′ at 27 °C due to the well-known aggregation of the F127 micelles. In these systems, the cross-over behavior was crucial to prevent ink flow, and the shapes printed from these inks only showed shape fidelity at higher temperatures.^[14] For the here investigated system, G′ was larger than G′′ already at room temperature. Also, the ratio $${{G''} \over {G'}} = \tan \left( \delta \right)$$ did not change in the temperature range from 27 to 60 °C, so that the temperature of the print bed can be kept at room temperature during printing. Another advantage of the presented system is that already at a laponite concentration of 4.2 mass%, the ink viscosity was as high as that of the F127 system at 30 mass% F127, i.e., a much smaller mass and volume fraction of rheology modifier was needed.

The time dependency of the mechanical response, i.e., the thixotropic properties of these laponite-based inks was evaluated for L3.8NP0.4 and L6.7NP0.4 by ramping up the shear rate from 0.1 to 100 s⁻¹ and back while measuring the shear stress τ (Figure 3c,d). Further data for other laponite concentrations can be found in the supporting information, in Figure S2 (Supporting Information). The shear rate $$\dot \gamma $$ versus shear stress τ curves (Figure 3c,d) show a hysteresis, which is typical for thixotropic behavior. By calculating the difference in the area between the up-sweep and the down-sweep from the τ versus $$\dot \gamma $$ plot, the so-called thixotropy index was determined.^[28] For the here investigated samples, the thixotropy index ranged between 301 and 3724 Pa⋅s⁻¹ (Table S1, Supporting Information). Thus, the thixotropy index increased with increasing laponite concentration, indicating that the samples with more laponite suffered a larger degradation of the superstructure by shear. Overall, the structure recovery rate of these samples (quantified by the closeness of the up-sweep and the down-sweep curve) was sufficiently fast to prevent ink flow and ensure shape fidelity.

Both the area of these hysteresis curves and the value of the thixotropy index strongly depend on the experimental parameters and therefore are not used as quantitative materials properties.^[14-16] However, these parameters were useful to compare the thixotropic properties within this series of inks and can be used as a benchmark for future work using the same experimental conditions.

2.4 Prototype Printing by DIW, Post-Printing Treatment, and Prototype Actuation

As the above-described experiments indicated, inks with a laponite content of ≈4.6 mass% had optimal printing performance and shape fidelity. Using this ink formulation, hydrogel precursors consisting of five layers were obtained by DIW. After printing, the print bed was heated to 50 °C to induce thermal cross-linking. The thus cured hydrogel prototypes could be removed from the print bed without damage. There was no need for further post-printing treatment. This is a significant advantage over the previously reported magnetic inks containing F127, which were mechanically sensitive and required further treatment with sulfate ions to cross-link the comprised chitosan, a third polymer component of that system besides F127 and polyacrylamide.^[14] The hydrogels printed from precursor ink L4.6NP2.3 are shown in Figure S3 (Supporting Information).

Simple magnetic actuation was shown by placing a rectangular hydrogel strip (size: 12.5 mm x 40 mm x 1 mm) into the magnetic field of a NdFeB (B = 1.29 T) permanent magnet (Figure 4a). The strip bent until it touched the magnet. The reversibility of the effect was shown using an electromagnet (120 N, 7 W, Figure 4b). Another prototype had a propeller-like shape (length: 60 mm, width: 10 mm, thickness: 2 mm, Figure 4c). When the permanent magnet approached the propeller, a gripping behavior was obtained. Flower-like shapes (diameter: 26 mm, thickness: 2 mm) floating in water followed a moving permanent magnet (Figure 4d; Video S1, Supporting Information). The type of magnetic actuation shown in Figures 4a,b can be useful as a valve or dynamic fluid doser in soft fluidic systems. Gripping actuation as shown in Figure 4c could find applications in soft robotics.

2.5 Morphological and Mechanical Characterization of the Hydrogels

A cross-section of a hydrogel actuator printed using L4.6NP0.8 was prepared by freeze drying and freeze-fracture, sputtered with gold, and imaged by scanning electron microscopy (SEM) at 5 kV using the secondary electron detector. At 500 x magnification, a porous structure with a relatively uniform pore distribution was found (Figure 5a). The pore size was larger than in the previous study where 30 mass% F127 was used as a rheology modifier,^[14] i.e., this laponite-based ink yielded less dense hydrogels. In a further image taken at 10 000 x magnification, clusters of the magnetic nanoparticles were observed (Figure 5b). The particles had a diameter of roughly 100 nm and formed clusters that were ≈1 µm in diameter. These clusters have the same size as those imaged by optical microscopy in the DIW inks of the F127 system.^[14] While in that hydrogel material, the clusters were embedded within the denser hydrogel structure, in the laponite system the clusters could be clearly observed. They were also identified by electron dispersive x-ray spectroscopy (Figure S4, Supporting Information). Further, SEM studies indicated that an increase in the iron oxide nanoparticle concentration increased the number of clusters, but not the cluster size, i.e., the clusters form before the ink is mixed.

For mechanical testing, dog-bone shape hydrogel samples (Figure S5a, Supporting Information) were DIW-printed from L4.6NP2.3, thermally cross-linked, and subject to a uniaxial tensile test. This gave a median elastic modulus of 22 ± 5 kPa, with a maximum tensile stress of 0.12 N mm⁻² and a yield strain of 5.5% (Figure S5b, Supporting Information). In comparison, a hydrogel obtained from the base formulation without laponite and iron oxide nanoparticles obtained by mold casting had an elastic modulus of 22 ± 7 kPa, a maximum tensile stress of 0.10 N mm⁻² and a yield strain of 4.0%. Thus, the mechanical properties of the hydrogels were hardly affected by the presence of the inorganic fillers at the mass percentages used, particularly in light of the large experimental errors in these mechanical tests, which reflect the difficulty of mechanically clamping hydrogel samples for tensile tests. Note that the reference material without laponite and nanoparticles had to be mold-cast as the viscosity of its precursor solution was too low for DIW. Still, the measured mechanical properties are remarkably similar, indicating that neither the fabrication method nor laponite or iron oxide had adverse effects on these values. Literature values of the elastic modulus of polyacrylamide hydrogels range from 10 to 100 kPa,^{[17, 18]} which is in good agreement with the here presented data.