In a typical experiment, the periodical patterned hydrogel films were fabricated by introducing well-aligned microchannels on one side of the films, as shown in Figures 1(a)–1(f). Firstly, these microchannel structures were fabricated by photolithography on the surface of silica wafers (fig. S1A, B). These periodical patterns endowed the whole surface with well-aligned microchannels (width: 800 μm; spacing: 800 μm; height: 100 μm), which facilitated SA (an anionic polysaccharide) pregel solution infiltration. After the pregel solution had filled all the channels by capillary force, it was cured to form patterned hydrogel films by water evaporation (fig. S1C). Then, the patterned hydrogels on the templates were first immersed into a 0.1 M CaCl₂ solution for 10 min, where the SA was pre-cross-linked to form ionic cross-links with Ca²⁺ ions (fig. S2A, B). Afterwards, microchannel-structured hydrogel films were obtained by peeling off from the silica templates (Figures 1(g)-1(h)). Finally, the hydrogel sheets were cut with microchannels in different orientations and further immersed into the 0.1 M CaCl₂ solution for another 24 h to cross-link thoroughly, enabling controllable 3D transformation of hydrogel structures with microchannels facing inward (fig. S2C, D). This shape transformation is caused by the large shrinking of the hydrogel at both the top surface and bottom surface in the previous cross-linking process.

The 3D shape-morphing degree of hydrogels depends on the pattern width (width: spacing = 1:1), pre-cross-linking time, and hydrogel thickness. Specifically, increasing the pattern width from 200 μm to 3200 μm decreases the bending degree of the resulting rolling structure from 2430°±28.6° to 1080°±27.5° correspondingly. The spacing in between the microchannels has an effect on the elastic tensors and, thus, on the deformation [28]. It clarifies that increasing the width of the microchannels decreases the tightness and smoothness of the rolling structures (fig. S3A-E, P). In addition, the rolling tightness can also be tuned by extending the pre-cross-linking time from 10 min to 1440 min, leading to a decrease of the bending degree of the resulting rolling structures from 1980°±19.8° to 1080°±20.5° owing to the variation of mechanical properties (fig. S3F-J, Q). Moreover, by changing the thickness of the hydrogel sheets, the rolling degree can also be tuned due to the alteration of bending resistance, thus varying the rolling structures (fig. S3K-O, R) [19].

The direction of the hydrogel deformation can be programmed by cutting the alignment of microchannels at a specific angle θ (Figures 1(i)–1(n)). With microchannels aligned at θ = 0°, 45°, or 90°, the resultant SA hydrogel sheets were imparted with a tube-curling structure, a helical structure, or a rolling structure correspondingly, and the deformed structures show no significant differences in water (Figures 1(i)–1(k)) and in the 0.1 M CaCl₂ solution (Figures 1(l)–1(n)) due to no differences in osmotic pressure. Figures 1(i)–1(n) show that all these bending directions are parallel to the microchannel orientations. The design of microchannel structures enables the programmed deformations and yet reduces the mechanical properties of the hydrogels accordingly. fig. S4 illustrates engineering force-strain curves of the hydrogels. The results indicate that the well-designed microchannel structures and larger cutting angle lead to decreases in the mechanical properties.

To understand the mechanism of the programmed 3D deformations of the hydrogels, hydrogel sheets solely with a cross-linking density gradient or well-aligned microchannels were fabricated, respectively (fig. S5). It is notable that the hydrogel deforms randomly when there is only cross-linking density gradient across the thickness (fig. S5A). Similarly, the hydrogel sheet with microchannel structures but no cross-linking density gradient at the bottom surface also randomly deforms (fig. S5B), resulting from the lower cross-linking density in the middle layer of the hydrogel (fig. S5C). This implies that a designated 3D deformation cannot be obtained for SA hydrogels solely with the cross-linking density gradient or well-aligned microchannel structures. Based on this evidence, it can be concluded that only a combination of both a cross-linking density gradient across the thickness and well-aligned microchannels could demonstrate controllable and precise deformation in a desirable manner (Figures 1(i)–1(n)) due to the cooperative out-of-plane and in-plane stress.

To further investigate the mechanism of the 3D deformations, the mechanical properties at different parts of hydrogel sheets with both the cross-linking density gradient and microchannel structure design were analyzed. Figure 2(a) illustrates that the amount of Ca²⁺ ions diffused into the SA hydrogel network determines the network density in the pre-cross-linking process, in which the SA films are still attached on the patterned silica wafer [29, 30]. Thus, the SA chains form the tightest networks at the top surface due to their complete exposure to the Ca²⁺ reservoir, resulting in the highest cross-linking density. In comparison, the SA chains form looser networks at the ridges and valleys due to decreasing Ca²⁺ concentration diffusible into the deeper parts of the hydrogel sheet, thus causing lower cross-linking density. After the pre-cross-linking process, the SA films were peeled off from the patterned silica wafer and immersed into the 0.1 M CaCl₂ solution for further cross-linking. However, a further cross-linking time of 24 h only increases the overall Young’s moduli of the hydrogel, and the differences of cross-linking density across the thickness of the hydrogel sheet still exist due to their variation of networks in the pre-cross-linking process (Figures 2(b)–2(d), fig. S6). Figure 2(b) demonstrates that the hydrogel networks become looser from the top surface to the bottom surface, strongly indicating the cross-linking density gradient across the thickness even after a further cross-linking time of 24 h. Figures 2(e)–2(g) further experimentally verify that Young’s modulus decreases across the thickness of the hydrogel sheet, which is 57.55±0.91 MPa for the top surface (Figure 2(e)), 14.94±0.73 MPa for the valleys (Figure 2(f)), and 1.42±0.038 MPa for the ridges (Figure 2(g)), respectively. The gradual decrease in Young’s modulus across the thickness of the hydrogel sheet brings about an out-of-plane stress in the swelling or shrinking process. Meanwhile, Young’s moduli are also different at the bottom surface between the ridges and the valleys, which results in an in-plane stress during the swelling or shrinking process. Therefore, the combination of dual cross-linking density gradients both across the thickness and at the bottom induces a cooperative effect between out-of-plane and in-plane stresses during the swelling or shrinking process, leading to designated shape transformations of the hydrogel sheets (fig. S7A). The bending energy is from the out-of-plane stress, while the bending direction is determined by the in-plane stress (the orientations of the microchannels).

To achieve effective reversibility and for a more sophisticated manipulation of the cooperative deformations, a Na⁺/Ca²⁺ cations system was employed (Figure 3). In water or the CaCl₂ solution, the cross-linked hydrogels appeared as a helical structure with the bottom surface of ridges and valleys facing inward (Figures 1(j), 1(m), and 3(d), fig. S7A, B). After immersing the hydrogel sheet into the 0.1 M NaCl solution, interestingly, a secondary deformation occurred in which the helix turned its helical rotation oppositely with the ridges and valleys facing outward (Figure 3(g), fig. S7A, C). It takes only 30 min to complete the actuation process, and the secondary deformed helical structure can retain its shape in both the 0.1 M NaCl solution and water (Figures 3(g) and 3(h), movie S1). After immersing back into the 0.1 M CaCl₂ solution, the oppositely rotated helical structure can recover its primary deformed shape with the ridges and valleys facing inward again in less than 1 min (Figure 3(i), movie S2). The hydrogels preserve excellent performance even after more than 10 cycles of the reversible shape transformations via immersing into the 0.1 M CaCl₂ and 0.1 M NaCl solution repeatedly back and forth (fig. S7D). This tunable actuation implies that the presence of Na⁺ ions will influence the ionic cross-linkages between COO^- and Ca²⁺, thus changing the swelling or shrinking properties and causing an inside-out transformation of the programmed hydrogels.

As previously reported, alginates are linear polysaccharides consisting of 1→4 linked β-D-mannuronic acid (M) and α-L-guluronic acid (G) which are comprised of sequences of M (M-blocks) and G (G-blocks) residues interspersed with MG sequences (MG-blocks). Gelation is driven by the interactions between G-blocks which associate to form tightly held junctions in the presence of Ca²⁺ ions. In addition to G-blocks, MG-blocks also participate, forming weak junctions (Figures 3(a)–3(c)) [29, 30]. Based on this, our hypothesis is that these junctions between COO^- and Ca²⁺ can be broken by Na⁺ ions, where Na⁺ ions diffuse into the hydrogel and compete with Ca²⁺ ions to weaken the ionic cross-links, thus releasing the Ca²⁺ ions from the hydrogel (Figures 3(e) and 3(f)) [31]. To further verify our hypothesis, the deformed hydrogel sheet with helical structure was immersed into a mixed solution of 0.001 M NaCl and 0.001 M CaCl₂. Figures 3(j) and 3(k) show that when the concentration of NaCl increases from 10 v/v% to 60 v/v%, the helix sheet becomes looser due to the replacement of Ca²⁺ ions by the competing Na⁺ ions which causes a decrease in cross-linking density and an increase in the swelling ratio of the hydrogel. The helical sheet becomes flattened when the concentration of NaCl is 64 v/v%. For a further increase in the concentration of NaCl to 90 v/v%, when more Ca²⁺ ions are replaced by Na⁺ ions and the cross-linking density is lowered, the secondary deformation occurs in which the ridges and valleys turn to be facing outward. Additionally, the bending directions of the helical sheet also change from parallel to perpendicular to the microchannel orientations. With an increase in the initial concentration of the mixed solution for comparison (0.1 M NaCl and 0.1 M CaCl₂), the helical sheet turns to be flat when the concentration of NaCl is as high as 98 v/v% because of a sharp increase in the Ca²⁺ ions concentration of the mixed solution (fig. S8). Figure 3(l) further confirms that the swelling ratio increases with the NaCl concentration from 0.0001 M to 0.1 M. Accordingly, Young’s moduli of the top surfaces decrease sharply from 53.15 MPa to 1.31 MPa with an increase of Na⁺ concentration from 0.0001 M to 0.1 M, indicating that more Ca²⁺ ions are released from the hydrogel (Figure 3(m)). In addition, we further verify our hypothesis with an energy dispersive spectrometer (EDS) to demonstrate the change of Na⁺ and Ca²⁺ ions in various conditions. Figure 3(n) shows that there is a Na peak in the SA without any treatment but no Ca peak. After the hydrogel cross-linked with 0.1 M CaCl₂ for 24 h, the Na peak disappears while the Ca peak arises, showing that the free Na⁺ ions are replaced by Ca²⁺ ions. When the cross-linked hydrogel is further immersed in various NaCl solutions (0.0001 M, 0.001 M, 0.01 M, and 0.1 M) for 24 h, respectively, the Na peak increases again while the Ca peak decreases, indicating that more Ca²⁺ ions are released from the hydrogel. Thus, the hydrogel sheets change their helical rotation oppositely due to a larger swelling ratio at the bottom surface compared to the top surface, where the ionic cross-links decrease with the increasing concentration of Na⁺ ions.

By the combination of differently aligned microchannels that generate simple deformations, we design and create a series of complex 3D architectures based on the art of “kirigami.” Starting from the simplest structures, a combination of microchannels aligned at 0°, 90°, and 0°/60°/120° configurations in one hydrogel sheet gives rise to T-shaped, H-shaped, and triangular hydrogel tubes in good shapes (Figures 4(a)–4(c)). For a more advanced deformation, a perfect double helix, which is comparable to the typical DNA molecular configurations, can be generated by a hollow rectangular hydrogel sheet with microchannels aligned at 45°. In addition, oriented microchannels with 45°/135° configurations in a hydrogel sheet result in a torsional helical structure (Figures 4(d) and 4(e)). All the deformations are parallel to the orientation of microchannels (fig. S9), which show good predictability and stability to be applied in 3D scaffolds for tissue engineering, 3D microfluidics, or microswimmers.

Inspired by various flowers, we further designed artificial six-petal flower structures comprised of oriented microchannels with 0°, 90°, and 0°/90° configurations. Figure 4(f) shows that a bud structure can be fabricated by aligning microchannels at 0° at the lower half of each petal. Thus, each petal can “close” parallelly to the orientation of microchannels, which gives the shape of a bud. Figure 4(g) illustrates an artificial blooming flower, in which the microchannels aligned at 90° at the lower half of each petal. With a combination of these two designs, an interesting structure comparable to a bird-of-paradise flower is vividly demonstrated, in which the microchannels are aligned at 0° at the lower half of each petal and 90° at the upper half of each petal (Figure 4(h)). Furthermore, we also mimicked the flower opening and closing processes by treating the previous six-petal flower structure with 0.1 M Na⁺ ions, where five petals were partially coated with blue dye and one petal was written with the letter “R” on the nonmicrochannel surface. The closed petals with the letter “R” outside, in which the microchannels aligned at 0°, opened slowly after immersing in the 0.1 M Na⁺ solution for 3 min. After 7 min, the flower opened completely into a flower. Further increasing the immersion time to 9 min, the flower closed oppositely into a complete bud with the letter “R” inside (Figure 4(i)). By immersing the six-petal flower in the 0.1 M NaCl solution, the different stages of an opening flower can be vividly demonstrated, and the letter “R” on the petal indicates the inside-out shape transformations of the hydrogels.

Sodium alginate (SA) was purchased from Sigma Aldrich. Sodium chloride and calcium chloride were obtained from Sinopharm Chemical Reagent Co., Ltd. SU-8 3050 and SU-8 developer were purchased from MicroChem Corporation. All reagents were used as received. Ultra-pure water was used in all the experiments.

Various microchannels with 100 μm height, different spacings (spacing of 200 μm, 400 μm, 800 μm, 1600 μm, and 3200 μm, width: spacing = 1:1), and combined orientations (0°, 45°, 90°, 0°/90°, 45°/135°, and 0°/60°/120°) on the silica wafers were prepared, respectively, by photolithography (EVG 610, Austria) according to our previous work [33]. Before use, all the patterned silica wafers were treated with oxygen plasma (Weike PDC-M, China) [34].

A 1.0 mL SA pregel solution (5.0 wt %) was poured into a cell composed of a microchannel-patterned silica wafer and a 1 mm thick silicone rubber spacer and then evaporated at room temperature for 5 h. Afterwards, the SA-casted silica wafer was immersed into the 0.1 M calcium chloride solution for 10 min, forming a SA hydrogel sheet (4.0 cm × 4.0 cm, 200.0 μm in thickness) with well-aligned microchannels (800 μm in width, 800 μm in spacing, and 100 μm in height). Then, the obtained SA hydrogel sheet was first cut into various stripes (3.5 cm × 0.5 cm) with cutting angles of 0°, 45°, and 90°, respectively, along the axis of patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for another 24 h. The SA hydrogel was washed with extensive ultra-pure water to remove the residual Ca²⁺ ions for further use. Unless otherwise specified, the hydrogel sheets with different cutting angles or complex configurations were prepared in the same manner as mentioned above.

A 1.0 mL SA pregel solution (5.0 wt %) was poured into a cell composed of a patterned silica wafer and a 1 mm thick silicone rubber spacer and then evaporated at room temperature for 5 h. Afterwards, the SA-casted silica wafers were immersed into the 0.1 M calcium chloride solution for 10 min, forming a SA hydrogel sheet (4.0 cm × 4.0 cm, 200.0 μm in thickness) with various widths of well-aligned microchannels (200 μm, 400 μm, 800 μm, 1600 μm, and 3200 μm in width; 100 μm in height; width: spacing = 1:1), respectively. Then, the obtained SA hydrogel sheets were first cut into stripes (3.5 cm × 0.5 cm) with a cutting angle of 90°, respectively, along the axis of patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for another 24 h.

A 1.0 mL SA pregel solution (5.0 wt %) was poured into a cell composed of a microchannel-patterned silica wafer and a 1 mm thick silicone rubber spacer, respectively, and then evaporated at room temperature for 5 h. Afterwards, the SA-casted silica wafer was immersed into the 0.1 M calcium chloride solution for various pre-cross-linking times of 10 min, 2 h, 4 h, 8 h, 16 h, and 24 h, respectively, forming a SA hydrogel sheet (4.0 cm × 4.0 cm, 200.0 μm in thickness) with well-aligned microchannels (800 μm in width, 800 μm in spacing, and 100 μm in height). Then, the obtained hydrogel sheet was first cut into a stripe (3.5 cm × 0.5 cm) with a cutting angle of 90° along the axis of patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for another 24 h.

A SA pregel solution (5.0 wt %) with a certain volume (0.5 mL, 1.0 mL, 1.5 mL, 2.5 mL, and 2.0 mL) was poured into a cell composed of a microchannel-patterned silica wafer and a 1 mm thick silicone rubber spacer, respectively, and then evaporated at room temperature for 5 h. Afterwards, the hydrogel sheet (4.0 cm × 4.0 cm, 142 μm, 200 μm, 257 μm, 300 μm, and 343 μm in thickness) was peeled off from the silica wafer. Then, the obtained hydrogel sheet was first cut into a stripe (3.5 cm × 0.5 cm) with a cutting angle of 90° along the axis of patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for 24 h.

A 1.0 mL SA pregel solution (5.0 wt %) was poured into a cell composed of a smooth silica wafer and a 1 mm thick silicone rubber spacer and then evaporated at room temperature for 5 h. Afterwards, the SA-casted silica wafer was immersed into the 0.1 M calcium chloride solution for 10 min, forming a SA hydrogel sheet (4.0 cm × 4.0 cm, 200.0 μm in thickness). Then, the obtained SA hydrogel sheet was first cut into a stripe (3.5 cm × 0.5 cm) with a cutting angle of 90° along the axis of the patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for another 24 h.

A 1.0 mL SA pregel solution (5.0 wt %) was poured into a cell composed of a microchannel-patterned silica wafer and a 1 mm thick silicone rubber spacer and then evaporated at room temperature for 5 h. Afterwards, the hydrogel sheet (4.0 cm × 4.0 cm, 200.0 μm in thickness) was peeled off from the silica wafer. Then, the obtained hydrogel sheet was first cut into a stripe (3.5 cm × 0.5 cm) with a cutting angle of 90° along the axis of patterns and then completely cross-linked in the 0.1 M CaCl₂ solution for 24 h.

To study the shape-morphing differences of the hydrogel sheets with oriented microchannels at 45° (3.5 cm × 0.5 cm, 200.0 μm in thickness), the hydrogels were immersed into various mixed solutions (0.001 M NaCl with 10 v/v%, 20 v/v%, 30 v/v%, 40 v/v%, 50 v/v%, 60 v/v%, 64 v/v%, 70 v/v%, 80 v/v%, and 90 v/v% by volume in 0.001 CaCl₂ and 0.1 M NaCl with 10 v/v%, 20 v/v%, 30 v/v%, 40 v/v%, 50 v/v%, 60 v/v%, 70 v/v%, 80 v/v%, 90 v/v%, and 98 v/v% by volume in 0.1 M CaCl₂) for 24 h. To study the actuation of the hydrogel sheets with oriented microchannels at 45°, the cross-linked hydrogels were immersed into the 0.1 M NaCl aqueous solution for 30 min and then transferred into the 0.1 M CaCl₂ solution for 1 min. The reversibility performance was evaluated by repeating the above steps 10 times. Before the switch of immersing solutions, all the former SA hydrogels were washed with extensive ultra-pure water to remove the residual Ca²⁺ or Na⁺ ions for the following experiments. The hydrogel swelling ratio based on mass was calculated by dividing the gel mass after swelling by the initial gel mass.

The morphologies and sizes of the silica wafer and SA hydrogels were determined by an optical microscope (Nikon Ni-U, Japan) and a field-emission-scanning electron microscopy (FE-SEM, ZEISS SIGMA 300). The mechanical properties of the hydrogels were determined on a testing machine (Instron, Legend 2344). All the shape-morphing behaviors of the hydrogels were recorded by a digital camera (Canon, 7D Mark II). The element compositions of SA hydrogels were detected by an energy dispersive spectrometer analyzer (EDS, X-MaxN). The local elasticities of the SA hydrogel samples were probed with a commercial atomic force microscope (AFM, Dimension Icon, Bruker) in force volume (FV) mechanical imaging mode according to our previous work [19]. An average elasticity was obtained via measuring five different areas of the SA hydrogel samples.