System and synthesis
Acrylate-based polymers were selected as the growing matrices because of their widespread usage in various applications. The initial photonic crystal films were made from SiO2 nanospheres (Figures S2 & S3) and poly(ethylene glycol) diacrylate (PEGDA). Briefly, the suspending solutions of SiO2 nanoparticles and PEGDA in ethanol were allowed to evaporate to induce the self-assembly of SiO2 into colloidal crystals, followed by curing under UV irradiation (Scheme S1) in homemade containers. After annealing, vivid films were obtained. The formation mechanism and driving force of SiO2 forming an ordered structure were polymerization-induced colloidal assembly which had been carried out by Ge et al28. Scanning electron microscopy (SEM) study illustrates that the samples consisted of both ordered and disordered domains (Figure S4). In the ordered domain, SiO2 nanoparticles arranged into ordered structures with uniform photonic bandgaps. Due to such a short-range ordered structure, the obtained structural color is angle-independent (Figure S5)29. The wavelength (λ) of the reflected light to this class of structural color films could be estimated by the Bragg’s diffraction Eq. 30:
where \(d\) refers to the interplanar spacing of the photonic crystal lattice, \({n}_{eff}\) indicates the mean refractive index of the composites, and m is the order of reflection (m = 1, 2, ...). In our design, \(d\) could be modified by allowing the polymer matrices to grow.
The mixture of an acrylate monomer, 1,6-hexanediol diacrylate (HDDA, crosslinker), 2-hydroxy-2-methylpropiophenone (photoinitiator), and benzenesulfonic acid (BZSA, transesterification catalyst) was used as the nutrient solution. Here, three kinds of monomers were employed, including 4-hydroxybutyl acrylate (HBA), PEGDA, and 2-hydroxyethyl methacrylate (HEMA). HBA is a commonly used elastomer precursor, while PEGDA is the compound used to prepare the initiated sample and therefore would not vary the composition of the sample matrices during growth; compared to HBA and PEGDA, HEMA could form rigid polymer main chains and thus provided an approach to tune material mechanical properties. The nutrient solution containing HBA, PEGDA, or HEMA used for growth was defined as nutrient solution B, EG, or M, respectively. For growth, as-prepared purple films (typical mass percentage of the polymer matrix: 40 wt%) were selected as the initial samples. They were first immersed in a nutrient solution for swelling. During swelling, the samples increase in size and change their colors (Fig. 2a). The samples could swell the tested nutrient solutions with equilibrium swelling ratios of 9.1 wt% (B), 6.2 wt% (EG), and 7.1 wt% (M), respectively (Figure S6). The relatively low swelling ratios were attributed to the low polymer fraction in the photonic crystal films and their rigid nature. Despite low swelling ratios, obvious color changes were observed. The colors were unstable and would turn back to purple again if the samples were washed to remove the entrapped nutrient solution. Besides, the swelling significantly weakened the Young's modulus of the samples (Figure S7). By contrast, the samples that were subsequently irradiated by UV light (365 nm, 10 mW·cm-2) to trigger photopolymerization preserved their after-swelling weights and colors, as well as became tougher (Fig. 2a and Figure S8). The photopolymerization reaction was exothermic, which could heat the systems to 47.1oC in 50 s (Figure S9). Such thermal effects could trigger transesterification reactions to release any polymerization-induced mechanical tension in such dynamic networks31. The grown samples could swell nutrients again for further growing cycles (Fig. 2b). To illustrate the role of this homogeneous step in growth, a control sample without transesterification catalyst was treated under the same process aforementioned, and as expected, this catalyst-free sample was nearly non-swellable to the nutrient solution after the first growing cycle. We denoted the initial sample as \({EG}_{n}\) where n is the mass percentage of the polymer matrix, while the grown samples as \({EG}_{n}-B/EB/{M}_{m}\), where m is the increased mass of corresponding polymer matrices compared to the entire original sample. For example, specimen \({EG}_{40}-{B}_{9.7}\) was the grown sample of \({EG}_{40}\) obtained from nutrient B with a net increase of 9.7 wt% compared to initiated weight (total polymer fraction: (40 + 9.7)/(100 + 9.7) wt%). All the specimens, including both initiated and grown ones, were self-standing and macroscopically uniform.
Optical property
The homogeneous growth allowed for flexible modulation of sample colors in the full visible light spectrum from purple to red (Fig. 2c). During growth, SiO2 colloidal nanoparticles maintained their ordered arrangement in the polymer matrix but showed increasing interplanar lattices, as evidenced by the gradual change of the double interplanar spacing 2d from 313 to 473 nm collected from the SEM images (Fig. 2d). UV-Visible reflection spectroscopy was employed to characterize the optical properties of different grown samples (Fig. 2e). With the increase in polymer matrices (then lattice distances), the reflection peak (λ) shifts from 481 to 690 nm. These reflection peaks were surprisingly intensive and narrow, implying a highly ordered arrangement of SiO2 colloid crystals. In our self-growing photonic composites, the changes in both double interplanar spacing 2d and reflection peaks (the colors) were predictable. For a photonic composite with a homogeneous ordered structure, the double interplanar spacing 2d was proportional to the cube root of the volume fraction of SiO2 nanoparticles (\(1/\sqrt[3]{1-\text{\varnothing }}\), where \(\varnothing\) is the volume fraction of the polymer matrix). By using the experimental data obtained from the initial photonic composite to establish the linear relationship (See section 9 in supplementary information for calculation detail, Table S1), we found that the theoretic curve compared well with the experimental 2d values (Fig. 2f), suggesting that the 2d could be precisely controlled by growth. Furthermore, in our system, the monomer showed a similar refractive index n with that of SiO2 (SiO2: 1.45; monomer: 1.457 at 20oC), and therefore, the \({n}_{eff}\) could be considered as a constant during growth. It indicated a linear relationship between λ and 2d based on equation #1. When we plotted the experimental λ to 2d, they all well fit with the theoretic curve (obtained from equation #1 with a \({n}_{eff}\) of 1.453, Fig. 2g). All these results indicated that the optical properties of photonic crystal films could be precisely post-modulated by growth.
Mechanical properties
Growth allowed the samples to vary their compositions for modulating their mechanical properties. Two kinds of approaches could be applied, including changing the monomer types or the crosslinker concentration in the nutrient solution. For example, in the case of using crosslinker-free nutrient solutions, the sample would maintain its modulus when grown from nutrient EG but become softer or stiffer from nutrient B or M, respectively (Figs. 3a). On the other hand, increasing crosslinker concentration in the nutrient solution stiffens the samples regardless the monomer types (Figs. 3b).
Besides stiffness modulation, growth also significantly toughened the photonic composite structure. The as-prepared initial samples were brittle and would fracture upon bending (Fig. 3c). Such brittle samples could grow to be fully flexible by adding agents that could soften the matrices in the nutrient. In the current system, the composites were stiffened by both the highly crosslinking structure of polymer matrices and the covalent connection between SiO2 nanoparticles and polymer chains of matrices (Figure S10&S11, FTIR results indicated the existing of the transesterification between the ester polymer and the hydroxyl on the surface of SiO2 nanoparticles during the preparation of the initiated samples). Since both linkages were ester-based bonding, alcohol that could hydrolyze the ester connection was selected as the additive (Fig. 3d). As expected, the grown sample obtained from an alcohol-contained nutrient was very flexible. More than bending, the flexible grown sample can be rolled up, twisted, or folded (Fig. 3e) and shows fair elasticity. The sample became more stretchable without a tradeoff in strength (Fig. 3f). Here, the alcohol could either be used as an additive in normal nutrient solutions or an independent agent (a nutrient consisting of only alcohol) to make the sample flexible (Figure S12). The transesterification-based softening mechanism was confirmed by using a SiO2-free PEGDA film in which the contribution of SiO2 nanoparticles to material mechanical properties was absent. The film was grown in alcohol-free and alcohol-containing EG, respectively. The grown sample made from an alcohol-containing nutrient shows a nearly 50% increase in flexibility compared to the grown sample prepared by alcohol-free EG (Figure S13). To further prove the mechanism, hydroxyl-free agents like tetrahydrofuran were employed as the additive, and none of them could make the sample flexible (Figure S14). Traditional photonic crystal-based structural color materials are normally brittle and improving their mechanical properties thus constitutes one of the most critical aspects of current research32,33. Two strategies, i.e. softening photonic crystal materials by the use of flexibilizers or increasing polymer fractions34 and fabricating full organic-based inverse opal structures35, had been developed. In the former, increasing polymer/flexibililizer fraction would disturb the self-assembly of particles and thus destroy the ordered crystal microstructure. The materials with inverse opal structures show excellent flexibility, but their preparation normally required HF-based etching treatment and the materials display milky appearance and low color saturation due to the higher light scattering in inverse opal36. In contrast, the growth method allowed for post-modulation of the composition without compromising the ordered microstructure, implying a more efficient method to get mechanically robust photonic crystal materials.
Reshape
The alcohol-based transesterification reactions were reversible and the broken ester linkages could reform by removal of the absorbed alcohol (Fig. 3d). To elucidate this idea, the softened PEGDA film made from the alcohol-containing nutrient was annealed at 70oC for 4 h to remove the alcohol moieties. The grown sample became stiffer and showed a modulus similar to that of the grown sample directly obtained from the alcohol-free nutrient (Figure S13). Such reversible softening-stiffening process allowed for multiple-time reshaping of the samples. As shown in Fig. 3g, a rigid flat film was first softened by growth with alcohol and then deformed, followed by annealing treatment (to remove the alcohol moieties) to fix the obtained structures. The reshaping could be repeated to achieve different forms. Interestingly, the deformed sample maintained its new shape when it was turned into an elastic state. Compared to this processability, traditional structural color materials are either rigid (photonic crystal composite, would fracture) or elastic (inverse opal material, turn back to original shape), and do not show such reshaping capability.
Spatial-selective growth and multicolor patterning
Another significant advantage of this photonic composite system is its spatially selective growth capability by using patterned photoirradiation (Fig. 4a). As shown by the surface profile, the irradiated region grew up with a sharp boundary. A height drop of 9.5 µm (yellow-orange change) occurred in ⁓1 µm distance (aspect ratio: 9.5). Consequently, the grown region showed a significantly different structural color from the unirradiated region. Note that the modulation could be achieved without the removal of the unreacted nutrient solution. In a previous study31, we have proved that region-selective conversion of the monomer and crosslinker into polymer matrices by photopolymerization would induce a concentration gradient of the monomer and crosslinker. Such gradient would drive the nutrient to transport from unirradiated region to irradiated one, leading to the formation of convex surface textures at the irradiated region. The same mechanism was expected here. After growth, the residual nutrient entrapped in the composite was still active and could be transported to new irradiated regions for creating other surface textures and colors. The structure of the grown region was further probed by SEM. In the cross-section image of a grown sample, a sharp boundary was also observed. In the grown region, SiO2 nanoparticles distribute orderly as those in the unirradiated region but show a larger interplanar lattice, implying homogeneous expansion of the polymer matrices.
Localized growth of photonic crystal films indicated a novel approach for chromatic patterning. We fabricated a “Sichuan facebook” using a \({EG}_{48}\) substrate with single blue color to demonstrate the flexibility of this approach. As shown in Fig. 4b, the profile of the “Sichuan facebook” was outlined on the substrate swollen with nutrient EG after simple irradiation through a mask. Without any additional treatment, further selective irradiations through the masks could induce the formation of new structural colors and a clear multicolor image was created. This growth-based patterning approach differed from the reported methods based on either complicated instruments37,38 or sophisticated skills. Here the structural color was patterned by user-friendly photo-lithography to directly achieve complicated, high-resolution images. The obtained pictures were not only stable but also always permitted further modification, due to the “living” growing mechanism, implying a more facile and effective direction for batch manufactures.
Self-healing
The self-healing of mechanical damages on structure colored materials was extremely challenging because of their rigid nature39,40. Since growth allowed for selective formation of new matrices in the irradiated region, it opened up a novel approach for our composite photonic system to restore the damage. To demonstrate this self-healing capability, a photonic crystal structural color film was first scratched on its surface. Although the polymer matrix itself is self-healable at high temperatures due to the transesterification-based chain exchange reaction41, the scratch did not effectively repair upon heating treatment (Figure S15). In contrast, when we selectively induced growth on the damaged region by light irradiation, the scratch was completely repaired, and an intact photonic crystal film was achieved (Fig. 4c). The healed film maintained its vivid structural color through the whole body (color in the healed region might slightly shift due to the change in the photonic bandgap, Figure S16). This growth-based self-healing method differed from traditional intrinsic and extrinsic self-healing strategies. In the intrinsic self-healing, no new matrix would form and the polymer chains should be mobile enough to undergo reconfiguration42,43. It is difficult for a rigid substrate to self-heal through this mechanism. Extrinsic self-healing strategy allows for the formation of new matrices but no ordered structure forms in the healed matrices44. Moreover, an interface between original and healed matrices often appears, which is unfavorable for the recovery of the mechanical properties. Here, our method combined the merits of both intrinsic and extrinsic strategies, and the healed sample completely restored its color and mechanical properties. The healed region exhibits comparable mechanical strength as the original sample (the self-healing efficiency: 80.1%, evaluated by the work, Fig. 4d), therefore, when stretched, the sample may even break at the intact region, rather than the healed one. By contrast, the scratched sample without self-healing was easily broken (8.8%).