CS hydrogel films with gradient cross-linking density across the thickness are fabricated by top-down diffusion of glutaraldehyde (GA) molecules within a CS pregel solution. In a typical experiment (Figure 1(a)), CS powders were firstly dissolved in an acetic acid solution (2.0 wt%) to prepare CS pregel solution (2.2 wt%), which was then poured into a home-made device, and GA solution was subsequently dripped onto the top of the pregel solution. Thus, CS hydrogel was cross-linked via Schiff-base reaction between the amino groups of CS and the aldehyde groups of GA [37, 38]. After being placed at room temperature for 24 hours, the dried CS hydrogel film was peeled off and then cut into specific sizes, which could quickly change their shapes from planar to 3D structures after immersing into water in seconds (Figures 1(b)–1(d) and Figure S1). The deformed shape and swelling ratio can be kept in deionized water stably after complete removal of the residual acetic acid by a 0.01 M sodium bicarbonate solution (Figure S2). The shape morphing of CS hydrogel film is induced by the synergistic effect of the gradient cross-linking density across the thickness and geometry effect.

We first investigate the effect of the gradient cross-linking density across the thickness on the swelling property of CS hydrogel film. As shown in Figure 2(a), the amount of GA molecules diffused into CS hydrogel films determines the network density of CS chains. Consequently, CS chains form the tightest networks at the top surface owing to their complete exposure to the GA solution, leading to the highest cross-linking density. On the contrary, CS chains form looser networks at the bottom surface due to the decreasing GA molecules diffusible into the deeper sites of the hydrogel film, thus resulting in a lower cross-linking density and looser structure at the bottom surface than at the top surface (Figure 2(b)). Therefore, the gradient cross-linking density results from the GA concentration gradient across the thickness of the CS hydrogel film, which is further verified by the results of Young’s modulus mapping. As shown in Figures 2(c)–2(e), Young’s modulus at the top surface of the CS hydrogel film is always higher than that at the bottom surface, regardless of the changes in the thickness of the film (Table S1). Notably, Young’s modulus of both the top and bottom surfaces decreases with increase of the thickness of the CS hydrogel film due to the increase of CS molecules involved in the cross-linking (Figure 2(e)). Specifically, for a thin hydrogel film, the GA molecules can reach the bottom surface of the hydrogel rapidly due to the short diffusion distance and thus have enough time to react with the CS hydrogel film on both sides. Consequently, the thin CS hydrogel film is saturated with the GA solution, resulting in high Young’s modulus at both sides (for a 20  μm thick hydrogel film, Young’s modulus of the top surface: 6.41\pm 0.49\text{MPa}, and Young’s modulus of the bottom surface: 1.06\pm 0.13\text{MPa}). In comparison, for a thick hydrogel film, the GA molecules need a long time to reach the bottom surface due to the long diffusion distance and thus have short time to react with the increased CS molecules across the thickness, resulting in a low Young’s modulus at both sides (for an 80  μm thick hydrogel film, Young’s modulus of the top surface: 0.94\pm 0.05\text{MPa}, and Young’s modulus of the bottom surface: 0.08\pm 0.01\text{MPa}). These distinct differences in stiffness between the top and bottom surfaces of the hydrogel films further confirm their different cross-linking densities across the thickness [39]. Moreover, these cross-linking density differences induce different swelling/shrinking ratios of CS hydrogel films (Figure 2(e) and Figure S2), which further leads to the mismatch strain/stress (out-of-plane strain/stress).

As reported in our previous work [26, 28], solely out-of-plane strain/stress can lead to nonuniform bending of a planar hydrogel film upon swelling. It means that programming the bending directions of the hydrogel films into well-defined 3D shapes still requires precise tuning of the mismatch strain/stress at specific directions, which is highly relevant to not only the gradient cross-linking density across the thickness of the hydrogel films but also the original geometries (e.g., length-to-width ratio and thickness) of the hydrogel films [32]. To further understand the synergistic effect of the gradient cross-linking density and geometry effect on the shape-morphing behaviors, we analyze the bending moments induced by the internal stresses of the swelling or shrinking hydrogel films. The basic mechanics of the hydrogel film with gradient cross-linking density across the thickness and specific geometry (e.g., length-to-width ratio and thickness) can be simply described by the bimetallic strip model (Equation (1)) [40].where \text{EI} is the bending rigidity of a hydrogel layer, and ρ is the radius of the bending curvature. The \mathbf{M} is positively related to the length difference before and after swelling of one side (\Delta l) and square to the thickness of the hydrogel film (h) (Equation (1)), and k is a Young’s modulus-related parameter. Unlike in a simple beam-like bilayer, the swelling of a hydrogel film with gradient cross-linking density across the thickness triggers a biaxial expansion inside the hydrogel, which creates mismatch strain/stress [32]. The resultant mismatch strain/stress further induces shape transformation of a hydrogel film, which is directed by the resultant bending moment ({\mathbf{M}}_{xy}) that includes {\mathbf{M}}_x ({\mathbf{M}}_{x1}, {\mathbf{M}}_{x2}) along the x-axis and {\mathbf{M}}_y ({\mathbf{M}}_{y1}, {\mathbf{M}}_{y2}) along the y-axis of the four sides of the hydrogel film (Figures 3(a) and 3(b)). By changing the geometry of a hydrogel film, e.g., initial length-to-width ratio (initial length/initial width: L_0/W_0) or thickness (h), the bending moments along the x-axis and/or y-axis will change due to the changing \Delta l and h or the changing Young’s modulus of the top and bottom surfaces of the hydrogel film, leading to the change of the resultant bending moment ({\mathbf{M}}_{xy}). Notably, the shape morphing of a hydrogel film is determined by the changed resultant bending moment, while the shape deformations always meet the request of lowest energy principle [34, 40–43].

To reveal the geometry effect on the shape morphing of CS hydrogel films, we first investigate the influences of the length-to-width ratios (L_0/W_0) on the shape-morphing behaviors of CS hydrogel films with a certain thickness (80  μm). As shown in Figure S3, three types of shape-morphing behaviors (diagonal rolling, helix, and short-side rolling) are observed. When L_0/W_0 is 1 (L_0: 12 mm, 24 mm, and 36 mm), the hydrogel films prefer to demonstrate diagonal rolling (Movie S1). It results from the resultant bending moment ({\mathbf{M}}_{xy}) being along the diagonal direction, where the moment along the x-axis ({\mathbf{M}}_x) is close to the moment along the y-axis ({\mathbf{M}}_y). When the L_0/W_0 ratio increases from 2 to 12 via decreasing the initial width (W_0) of CS hydrogel films, the hydrogel films mainly demonstrate helix structures, which have the lowest energy. Specifically, as the L_0/W_0 ratio increases slightly (<6), the diffusion of water inside the hydrogel layer is similar along both the length (L_0) and width (W_0) directions, resulting in bending simultaneously at these two sides. Afterwards, the bending along the length direction (L_0) eventually dominated due to the slightly larger moment along this direction ({\mathbf{M}}_y), in agreement with energetic considerations, leading to a final helix shape. Comparatively, as the L_0/W_0 ratio increases largely (≥6), the diffusion of water inside the hydrogel layer is faster along the width direction (W_0) than the length direction (L_0), resulting in bending along the width direction (W_0) first and then along the length direction. The bending ({\mathbf{M}}_x) along the width direction (W_0) firstly occurs due to the fast swelling along this direction, and the bending ({\mathbf{M}}_y) along the length direction (L_0) starts later due to the edge effects [34]. Thus, the collaborative bending ({\mathbf{M}}_{xy}) along these two directions leads to helix structures (Movie S2 and Movie S3). Notably, the edge effects become stronger as the L_0/W_0 increases largely, which would change the shape-morphing path. As shown in Figure S3, the hydrogel films (L_0: 12 mm, L_0/W_0\ge 8) show short-side rolling as the W_0 decreases to ~1 mm, which results from the much faster diffusion of water inside the hydrogel layer along the width direction (W_0) than the length direction (L_0) due to the strong edge effects, leading to initial bending along the width direction (Movie S4). Moreover, the initial bending of the hydrogel film determines the shape-morphing path. Even though a helix shape can be obtained in a lower energy state, the bending along the length direction is hindered due to the difficulty of overcoming the high energy barrier to reverse the short-side rolling. In addition, diagonal rolling, helix, and short-side rolling are still observed as the L_0/W_0 ratio increases from 1 to 12, where the initial length (L_0) of CS hydrogel films increases from 1 mm to 60 mm. The edge effects are further confirmed experimentally by using CS hydrogel films with both L_0 and W_0 close to millimeter size as shown in Figure S4, which increases the probability of short-side rolling. Therefore, various shape-morphing behaviors can be facilely obtained via changing the length-to-width ratios (L_0/W_0) of CS hydrogel films.

To confirm the effect of length-to-width ratios (L_0/W_0) on shape-morphing behaviors, we further explore the relationships between the length-to-width ratios of the swollen CS hydrogel films (L_d/W_d; L_d: the length of the swollen hydrogel film, and W_d: the width of the swollen hydrogel film) and various shapes (Figure 3). According to our experimental results (Figure 3), for both diagonal rolling and short-side rolling, the length-to-width ratios (L_d/W_d) of hydrogel films are close to 1. For the helix, the length-to-width ratios (L_d/W_d) of hydrogel films are greater than 1. Moreover, the shape evolution of hydrogel films can be indicated via investigating the relationship between the L_d/W_d and the L_0/W_0. As shown in Figure 3(c) and Figure S3, it is clear that the L_d/W_d ratios (≥1) increase with further increases of L_0/W_0 ratios, indicating that the planar hydrogel films transform into diagonal rolling firstly and then helix mainly. Specially for the hydrogel films with L_0=12\text{mm}, the L_d/W_d ratios increase with increasing L_0/W_0 ratios firstly and then decrease (Figure 3(c)). Correspondingly, the planar hydrogel film evolves from diagonal rolling, helix, to final short-side rolling. Further decreasing the W_0 (from 5 mm to 1 mm), the shape deformations show differences in the same range of L_0/W_0 ratios from 1 to 12 (Movie S4 and Movie S5). As shown in Figure 3(d) and Figure S4, the L_d/W_d (≥1) increases with further increasing L_0/W_0, showing that the planar hydrogel films transform into diagonal rolling firstly, then short-side rolling (increasing probability) and helix mainly. It is worth noting that strong edge effects are particularly significant for the diagonal rolling and short-side rolling (L_d/W_d: ~1, very small L_0 or/and W_0) [34]. Although we can obtain various shape deformations including diagonal rolling, helix, and short-side rolling via solely changing the length-to-width ratios (Movie S1, Movie S6, and Movie S7), the shape-morphing rules still need to be expounded via studying the geometry effect systematically.

In order to enhance the shape-morphing editing capability, we further investigate the influences of hydrogel film thickness on shape-morphing behaviors, which is of equal importance for shape-morphing regulation [40, 41]. As discussed previously, for CS hydrogel films with certain initial length and width (L_0: 5 mm, W_0: 5 mm), Young’s modulus of both the top and bottom surfaces of the CS hydrogel film decreases with increasing thickness from 20  μm, 40  μm, 60  μm, to 80  μm. Conversely, the decreasing Young’s modulus (decreasing cross-linking density) will increase the swelling ratio of the hydrogel film on the one hand, resulting in a larger swelling difference between the top and bottom surfaces of CS hydrogel films. On the other hand, the low Young’s modulus of the hydrogel film indicates high deformability according to our previous work [26]. Therefore, increasing the thickness (h) of CS hydrogel films would increase the swelling difference (thus increasing \Delta l) between the top and bottom surfaces of CS hydrogel films, resulting in a large bending moment ({\mathbf{M}}_{xy}) according to Equation (1) that determines the shape morphing of CS hydrogel films (Figure 4(a)). This effect is further experimentally confirmed by employing CS hydrogel films (certain initial length and width: L_0: 9 mm, W_0: 3 mm, L_0/W_0=3:1) with various thicknesses from 10  μm to 80  μm as shown in Figure S5, which show no bending (10  μm thickness), long-side bending (20  μm thickness), helix (30  μm and 40  μm thicknesses), and short-side bending (50  μm to 80  μm thickness) (Figure 4(b), Movie S8–S11). The shape evolution also agrees with the change of L_d/W_d as shown in Figure S5. To reveal the geometry effect (including both L_0/W_0 and h) on the shape-morphing behaviors systematically, the effects of the both length-to-width ratio and thickness on the deformation behaviors of CS hydrogel films are described in a phase diagram (Figure 4(b) and Figure S6). Guided by the phase diagram, we can precisely program diverse deformations (i.e., no bending, diagonal rolling, long-side rolling, helix, and short-side rolling) of CS hydrogel films by tuning the thicknesses and/or the length-to-width ratios (Figures 4(c) and 4(d)). Moreover, this robust strategy is not only suitable for CS hydrogel films but also appropriate for sodium alginate (SA) hydrogel films. All shape-morphing behaviors (diagonal rolling, short-side rolling, helix, and long-side rolling) can be facilely achieved by tuning the gradient cross-linking density and geometries of SA hydrogel films (Figure S7), which provides a universal and efficient strategy to precisely control the shape-morphing behaviors of various hydrogels.

Combining both length-to-width- and thickness-determining effects on programmable shape morphing of CS hydrogel films, we further prepare some complex actuators from planar CS hydrogel films assisted by the art of “kirigami” (Figures 5(a)–5(d)). For cross CS hydrogel films with certain thicknesses (80  μm) but different length-to-width ratios, they quickly change their planar shapes into an artificial claw (Figure 5(a), L_0=6\text{mm}, W_0=1\text{mm}) and an artificial windmill (Figure 5(b), L_0=12\text{mm}, W_0=5\text{mm}) after immersing in water for 3 min. Furthermore, an artificial horn can also be obtained upon swelling of a U-shaped hydrogel film (Figure 5(c)), where the feet and bridge of the U-shaped hydrogel film are deformed into helix and short-side rolling, respectively, due to the different length-to-width ratios. In addition, through controlling the thicknesses and the length-to-width ratios of CS hydrogel films precisely, an artificial flower with various parts undergoing distinct deformation modes (i.e., long-side rolling, short-side rolling, and helix) can be facilely achieved (Figure 5(d)).

Besides, owing to the existence of massive cations in CS, the hydrogel films also exhibit pH responsiveness [44]. The deformation of CS hydrogel films can therefore be adjusted by different pH values. Compared with shape morphing in deionized water (pH 7), the swelling degrees of CS hydrogel films are enhanced in an acid condition (pH 3) and reduced in an alkaline condition (pH 11) due to the protonation (increased swelling) and deprotonation (shrinking) of CS chains, respectively (Movie S12 and Movie S13). As shown in Figure S8(a), CS hydrogel films show only two types of shape morphing (diagonal rolling and helix) due to the sharply increased swelling of CS hydrogel films in an aqueous solution at pH 3. On the contrary, the shape morphing of CS hydrogel films changes little in an aqueous solution at pH 11 solution, because CS hydrogel films show slight shrinking (Figure S8(b)). Based on this, we realize dynamic transformations of the underwater artificial claw (Figure 5(e), h=80\mu \text{m} and L_0/W_0=6) by tuning the pH values of environments. The opening/closing of the artificial claw can be switched by dynamically controlling the bending angles of the pH-responsive CS hydrogel fingers. And the bending degrees of CS hydrogel films are quantitively described by the curve of bending angles with respect to pH values (Figure 5(f)). These pH-responsive programmable deformation behaviors of CS hydrogel films make them promising candidates for soft actuators and drug delivery.

Chitosan with low average molecular weight, glutaraldehyde aqueous solution (GA, 25.0 wt%), and sodium alginate (SA) were purchased from Sigma-Aldrich. Acetic acid, sodium bicarbonate, and calcium chloride were obtained from Sinopharm Chemical Reagent Co., Ltd. Ammonia solution (NH₃·H₂O, 25.0~28.0 wt%) was purchased from Dongjiang Chemical Reagent Co., Ltd. Ultrapure water (≥18 MΩ) was used in the experiments. Home-made devices for preparing hydrogel films with certain volumes were made of glass slides containing Teflon spacers with specific sizes (40\text{mm}\times 40\text{mm}\times 1.5\text{mm}; 80\text{mm}\times 80\text{mm}\times 2.0\text{mm}).

CS powders were dissolved in an AA solution (2.0 wt%) and stirred overnight to prepare a CS pregel solution (2.2 wt%). Congo red (0.1 wt%) was added in the solution for staining purpose.

A 6.2 mL CS pregel solution (2.2 wt%) was poured into a home-made device (40\text{mm}\times 40\text{mm}\times 1.5\text{mm}). Then, the GA solution was dropped onto the top surface of the CS pregel solution, which was placed at room temperature for 24 h to form a dried CS hydrogel film. To tune the cross-linking degrees of CS hydrogel films, various concentrations of GA (GA/CS (mass ratio%): 9.28%, 4.64%, 2.32%, 1.16%, 0.58%, and 0.29%) were used for chemical cross-linking of CS. To prepare CS hydrogel films with different thicknesses (10  μm, 20  μm, 30  μm, 40  μm, 50  μm, 60  μm, and 80  μm), various volumes of CS pregel solutions (0.78 mL, 1.55 mL, 2.33 mL, 3.10 mL, 3.88 mL, 4.65 mL, and 6.20 mL) were employed and further cross-linked with GA solutions (0.29%). The obtained dried hydrogel films were cut into strips with designated length-to-width ratios. All dried CS films were firstly swollen in deionized water for 10 min. Subsequently, the deformed hydrogel films were immersed into a 0.01 M NaHCO₃ solution for 30 min to remove the residual acetic acid. Finally, the hydrogel films were soaked in deionized water for 24 h to reach swelling equilibrium for further use. CS films with large sizes were prepared as mentioned above via a large home-made device (Teflon spacer size: 80\text{mm}\times 80\text{mm}\times 2.0\text{mm}).

1.0 mL and 2.5 mL SA solutions (4.8 wt%) were poured into a home-made device (40\text{mm}\times 40\text{mm}\times 1.5\text{mm}) and then evaporated at room temperature for 24 h to obtain dried films with certain thicknesses of 20  μm and 50  μm, respectively. Afterwards, the SA hydrogel films were cross-linked by immersing in a 0.05 M calcium chloride (CaCl₂) solution for 1 min and then dried for 6 h [26]. The dried SA hydrogel films were cut into strips with certain length-width ratios, which were finally immersed into water for further use.

To study the shape-morphing differences of hydrogel sheets in acid or alkaline solutions, the dried CS hydrogels were cut into various strips with the same initial width (W_0: 3 mm), different lengths (L_0: 3 mm, 6 mm, 9 mm, 12 mm, 15 mm, 18 mm, 24 mm, 30 mm, and 36 mm), and different thicknesses (h: 10  μm, 20  μm, 30  μm, 40  μm, 50  μm, 60  μm, and 80  μm). After removal of the residual acetic acid, the hydrogels were immersed in deionized water for 24 hours to reach the swelling equilibrium. Subsequently, the swollen hydrogels were transferred into the acid solution with pH 3 or the alkaline solution with pH 11 for 24 h to reach the swelling equilibrium. The acid solution with pH 3 was obtained by adding 3.30 mL acetic acid into 1000 mL deionized water, and the alkaline solution with pH 11 was obtained by adding 7.74 mL ammonia hydroxide solution into 1000 mL deionized water.

The thicknesses of the dried CS and SA hydrogels were measured under an optical microscope (Nikon Ni-U, Japan). The morphologies of the cross-sectional CS hydrogel films were characterized by a field-emission-scanning electron microscope (FE-SEM, ZEISS SIGMA 300). The shape-morphing behaviors of CS and SA hydrogel films were recorded by a digital camera (Canon, 7D Mark II). The local elasticities of CS hydrogel films were probed with a commercial atomic force microscope (AFM, Dimension Icon, Bruker) in the force volume (FV) mechanical imaging mode [26, 28, 51]. Average elasticity was obtained by measuring five different areas of CS hydrogel films.