Introduction

Growth is ubiquitous but generally very sophisticated mode for generating organismal matter1. Starting from a living seed that can absorb and integrate nutrients to propagate, it involves complex concurrent mass transport, exchange, and multi-step biochemical transformations2. These processes are mainly dictated by an intrinsic code but are also highly sensitive to internal and external stimuli. As a result of the growth, living organisms can reshape to adapt to the environment, regenerate tissue to heal unexpected damage, and even transform3,4. Recently, this growth concept of living organisms has been applied to designing several organic materials that can incorporate externally provided reactive compounds to achieve growth5,6,7. Through the process of growth, the materials display many attractive features, including self-strengthing7, tunable bulk properties8,9,10, post-fabrication modifiable surface textures6,11,12, varying structural colors13, damage repairing14, expanding under force15,16, etc. Despite such progress, these materials cannot self-grow into entirely distinct shapes in template-free conditions as living organisms do. Therefore, developing an innovative method to control the shape in self-growth of cross-linked organic materials remains a highly attractive goal.

From the geometric viewpoint, systems can grow via either a morphologically homogeneous mode, in which the shape of a cross-linked system (denoted as a seed) is preserved, or a morphologically heterogeneous manner, in which it undergoes a substantial shape transformation. Typically, self-growing materials expand their sizes by incorporating reactive compounds into the cross-linked polymers constituting the matrices2. These reactive compounds are introduced into the seeds by swelling, and the incorporation reaction engineered into the growing system occurs homogeneously at the molecular level8. The materials typically maintain their geometric shapes in such a growth process. However, it is worth noting that the cross-linked polymer chains are stretched during swelling due to the occurring size expansion17,18. The resulting extended polymer chain conformations are entropically unfavorable, generating mechanical tension that restricts further swelling, even after the reactive compounds have been converted into the resultant polymer networks. Homogenization mechanisms involving chain exchange reactions are usually built into the system to release such tension and enable continuned growth6,8,13. In contrast to these previously reported designs, we propose here that by controlling the local reactivity of the polymerization and chain exchange reactions, we can realize the macroscopic shape transformation of the growing polymer.

Results

We first probed the possibility of macroscopic transformation by comparing two different growth mechanisms using Finite Element (FE) simulations (Fig. 1a, b), based on a physics-based model in our previous work that couples diffusion, polymerization, chain exchange reaction, and large deformation for dynamic polymers. Details of the theory can be found in our previous paper19, and the key equations are listed in the Supplementary Information. In the simulation, the seed polymer was first swelled with a nutrient solution containing monomers and cross-linkers. For the homogeneous growth, we imposed the polymerization and chain exchange reactions to occur homogeneously throughout the sample, while for the heterogeneous growth, we used a core-shell configuration featuring a rapid reaction rate at the core and zero reaction rate at the shell. Simulation results show that in the swelled states, both samples expanded homogeneously, and tensile stresses were built throughout the samples. For the homogeneous growth, when the polymerization and chain exchange reactions were on, the sample preserved the same shape and size. The stress built in the swelling step was released due to the chain exchange reaction (Fig. 1a). For the heterogeneous growth, the reaction in the core consumed the nutrients, which drew the nutrients from the shell to flow to the core (Fig. 1b). As a result, the shell shrinks and the core expands, leading to overall compression of the core and tension of the shell. Due to the continuous chain exchange reaction in the core, the core material in the long timescale behaves like a liquid, transforming itself to a round shape to minimize the overall elastic energy of the system (Supplementary Section 1 and Figs. S1–S3). It was worth noting that in our system, the shrinkage of the shell and expansion of the core were driven by the polymerization reaction, during which the network structure evolved dynamically over time. This deformation mechanism differed from previous work arising from fixed charge generation within the network20. The charge generation induces osmotic pressure differences to drive solvent diffusion, while the network structure remains static.

Fig. 1: Growth modes.
figure 1

Finite element simulation of the (a) homogeneous and (b) heterogeneous modes of growth, based on a physics-based model that couples diffusion, polymerization reaction, chain exchange reaction, and deformation. The color bar denotes the maximum principal stretch ratio. Schematic illustration of the homogeneous (c) and heterogeneous (d) growth from living seeds. c The active species are distributed homogeneously, while in (d), the active species are concentrated in the core. The initial networks and the newly generated networks are shown in green and red, respectively. d As the monomers and cross-linkers in the core region are consumed from the polymerization and chain exchange reactions, those that are unreacted in the shell region are drawn into the core, resulting in the contraction of the shell and expansion of the core, and the heterogenous growth.

Full size image

While Fig. 1c illustrates the strategy that has been applied to achieve homogeneous growth in previous studies19, Fig. 1d describes the molecular mechanism of the design we report here for achieving heterogeneous growth. Briefly, both systems started with “living” polymer network seeds (fabricated as a piece of elastomer), in which the active species at end sites could continue to react with and incorporate monomer molecules21,22. For growth, the nutrient solution (the mixture of monomer and cross-linker) was first added to the elastomer, causing it to swell to form an organogel and increase in size. These polymerizable units would undergo polymerization to create the second networks, followed by a chain exchange-based homogenization process to yield the grown products. For homogeneous growth, the active species should be distributed homogeneously (and/or able to diffuse freely) throughout the sample for the polymerization and homogenization (i.e., integration of the nutrient) accompanied with the swelling, but at a much slower rate. In contrast, the heterogeneous growth relies on the heterogeneous distribution of the active species to initiate spatially controlled polymerization. Figure 1d shows a typical example in which the active species are concentrated in the core. At first, polymerization initiated in the core consumes the monomer and cross-linker around the active site to form the second polymer chain. It generates a concentration gradient of the monomer and the cross-linker from outside in. As a result of the gradient, the monomer and the cross-linker migrate from the shell region into the core. These diffusing-in polymerizable compounds would accelerate the polymerization in the core and then, in turn, induce more mass transport via diffusion. This self-enhanced growing cycle was expected to expand the core but shrink the shell, making the sample transform during the growth. The critical aspect here is the formation of the heterogeneous distribution of the active sites throughout the sample, which restricts the polymerization reaction to the initially active region (the core).

In the previous study, we had demonstrated the homogeneous growth of the silicone systems made from acid-catalyzed ring-opening polymerization (ROP) of octamethylcyclotetrasiloxane (D4, monomer) and 1,1,3,3-tetra(2-heptamethylcyclotetrasiloxane-yl-ethyl)-1,3-dimethyldisiloxane (triD4, cross-linker, Fig. 2a) with trifluoromethanesulfonic acid as the catalyst19. We found that the acid-induced polymerization and chain exchange proceeded at room temperature (rt), with the nutrient (the mixture of D4 and triD4) incorporated homogeneously into the matrices due to the ability of the catalyst (i.e., proton) to percolate throughout the networks easily and distribute uniformly23. As a result, the acid-catalyzed growth occurred spontaneously when nutrients were provided to the seeds by swelling. In this acid-catalyzed case, the swelling, polymerization, and chain exchange processes all occur together, though their kinetics differ quite noticeably. The polymerization is substantially slower than the swelling process, so the nutrients can reach a nearly homogeneous distribution, leading to the preservation of the geometric shapes, as described in Fig. 1c. In sharp contrast, polymerization of D4 and triD4 and chain exchange reactions catalyzed by a strong base, tetramethylammonium siloxanolate (TMAS), typically proceed fast enough only at high temperatures (Supplementary Fig. S4)23,24,25, and the percolation of the basic active species (siloxanolate/hydroxyl) is not a simple diffusion, but rather a chain transfer process (Supplementary Fig. S5)26. Based on these features, we hypothesized that there existed a possibility of forming a heterogeneous, non-uniform distribution of the basic active species, leading to a macroscopic non-uniform shape transformation, as proposed in Fig. 1d. To test this hypothesis, we prepared basic seeds and subjected them to the growth conditions described below. Briefly, the mixture of D4 (97 wt%), trD4 (2 wt%), and TMAS (1 wt%) was stirred at 80 °C under N2 for 5 min to generate a homogeneous viscous liquid. The liquid was poured into a mold immediately, followed by further annealing at 90 °C for 4 h under a sealed condition to yield the seed (an elastomer). Polydimethylsiloxane (PDMS) networks capable of equilibrating were expected to form in the seed because the anionic propagating species in PDMS are known to retain activity after polymerization (Fig. 2a)23,24,25. These anionic species are not only able to continue reacting with monomers but also to induce ongoing chain exchange reactions, keeping the PDMS networks in a dynamic state27.

Fig. 2: Base-catalyzed self-growing silicone materials.
figure 2

a Chemical structure of D4, triD4, and TMAS, base-catalyzed ring-opening polymerization of D4 and chain exchange reactions of the obtained polymer chains. b Weight change of seeds in a growth cycle with a deactivated seed as a control. Annealing temperature: 90 °C. The samples were immersed in a D4 solution containing 2 wt% triD4. Right: the image of a square seed (i), its swelled square sample (ii), and grown spherical product (iii). The plotted data were obtained from eight independent measurements. Scale bars: 3 mm. c Photographs of a swelled sample at different times when annealed at 90 °C. d Plot of the ratio of axes (r/d) against the annealing time. Data in (b, d) are presented as mean values +/− SEM, n = 3.

Full size image

The seed was immersed in the mixture of D4 and triD4 (nutrient) for swelling, with a deactivated sample as the control (annealed at 135 °C for 24 h under vacuum to degrade the TMAS groups). As one would expect, the basic living seed displayed the room temperature physical swelling kinetics and accompanying it increase in size similar to those of the deactivated control (Fig. 2b). Swelling ratios of up to 3.5 in 20 h were observed. Similarly to the deactivated controls, the living samples maintained their geometric shapes during the room temperature swelling (from i to ii states, Fig. 2b). In such living swelled samples, the nutrients were physically entrapped rather than incorporation via polymerization as that of acid-catalyzed systems. The sample would shrink back to its original size as washed with hexane, even after storage at room temperature for two days. Under the same storage, the acid-catalyzed sample maintained its increased size since the polymerization occurred at room temperature. Base-catalyzed polymerization of the nutrients requires a high temperature. Interestingly, when the living swelled sample was annealed at 90 °C, dramatic shape transformation, i.e., a change from a square sheet into a spherical ball, was observed (from ii to iii in Fig. 2b). Accompanying this shape transformation, most of the absorbed nutrient solution was incorporated into the sample. In contrast, the swelled deactivated control shrunk back to its original weight and shape because of the evaporation of the absorbed liquids (unreacted volatile monomer and cross-linker). It is worth noting that such spontaneous deformation fundamentally differs from those observed in sophisticated shape-memory systems, which are mainly designed to return to the initial shape28. The obtained structures were stable, showing no change after storage for 1 year. We monitored the growth of a square sheet using a camera. Briefly, a swollen square sheet was laid in a sealed glass bottle filled with N2. The bottle was then placed in an oven at 90 °C. The oven was equipped with a glass window for allowing us to monitor and record the transformation in-situ. As shown in Fig. 2c, the transformation is gradual, without a sharp, abrupt transition. For this reason, to quantitatively describe the transformation, we calculated and plotted the dependence of the ratio of axes (r/d) on time (Fig. 2d), which characterizes the transformation kinetics. The growth is slow during the first 20 min and then it proceeds gradually in 3 h, before plateauing. We noted that the growth involved two processes, i.e., mass transport and polymerization (Supplementary Figs. S6 and S7). Two kinds of control experiments were conducted to evaluate the kinetics of these two processes, i.e., the polymerization of D4 under the same condition and swelling of D4 in deactivated seed at 90 °C. The polymerization of D4 reached the plateau state in 140 min while a deactivated sample did not reach its saturated swelling state even after 260 min. It indicated that the swelling process was substantially slower than polymerization (Supplementary Fig. S8). Therefore, we concluded that the growth was controlled mostly by the diffusion process of nutrient molecules. This conclusion was supported by the simulation. We set the diffusivity (D = 1) and varied the polymerization rate constant (kp). By comparing the experimental results with the simulation, we obtained a fitted value of kp = 10, meaning the polymerization was faster than diffusion, resulting in a geometry consistent with experiments. Conversely, if kp = 0.1, where diffusion was faster polymerization, the simulated results deviated from experiments (Supplementary Fig. S9).

We attributed the shape transformation to the formation of a core-shell structure in the swelled sample. As shown in Fig. 3a, a core-shell structure self-formed when the seed was immersed in the nutrient solution for swelling. The swelled samples had transparent shells and translucent cores. We interpret this appearance in the following way: the active basic species, tetramethylammonium siloxanolate or hydroxide, are highly hygroscopic, and by attracting water molecules into the hydrophobic matrix, they make the host silicone samples cloudy29. Based on these observations and reasoning, we assigned the transparent shells and the translucent cores to the regions free of the active basic species and the ones containing them, respectively. The swelling was conducted in air where acid CO2 would diffuse into the sample to deactivate the basic species. The CO2 diffusion was slow such that only the basic species in the shell region were reacted during the swelling. To confirm our hypothesis, we employed Fourier transform infrared (FT-IR) spectroscopy to analyze the composition of the core-shell structure. As shown in Fig. 3b, no obvious peak was observed in the range of 3000–3600 cm−1 of the seed’s spectrum. In comparison, a broad peak of water at ~3200 cm−1 and a weak peak at 3024 cm−1 which was assigned to HCO3− 30, were observed in the translucent core region of the swelled sample, indicating the absorption of water and CO2. The presence of the water also illustrated the hygroscopic nature of this region. On the other hand, the peak at 3024 cm−1 becomes even stronger when the broad peak at ~3200 cm−1 decays in the shell region of the swelled sample. These results suggested that the basic species had been neutralized by CO2, and the region became non-hygroscopic (Fig. 3c). The molecular mechanisms were further proved by in-situ FT-IR measurements monitoring a transparent seed exposed to air where both water and CO2 were present (Supplementary Fig. S10). As expected, the top region of the sample exposed to air became translucent in one day. During this transparent-translucent change, the broad water peak at ~3200 cm−1 and sharp HCO3− peak at 3024 cm−1 appeared. After 10-day exposure, the translucent region became transparent, accompanied by the disappearance of the broad water peak at ~3200 cm−1 and the emergence of the stronger peak at 3024 cm−1 (Supplementary Section 4.3). Based on these results, we could attribute the formation of the core-shell structure to the neutralization of active basic species with CO2 from the air in the shell region.

Fig. 3: Core-shell structure of swelled basic seeds.
figure 3

a Schematic illustration (left) of the formation of core-shell structure and the image (right) of a swelled sample with a translucent core (region I) and a transparent shell (region II). Scale bars: 3 mm. b FT-IR spectra of a seed and the different regions of a swelled sample. c Reactions of propagating species and hygroscopic properties of the species.

Full size image

To analyze and visualize how the nutrients are distributed within growing samples, a dyed cross-linker, i.e., azobenzene-conjugated D4 (Supplementary Fig. S7), was designed. We expected that, due to its cross-linking nature within the polymer networks, the dyed cross-linker would not be able to move significantly during chain exchange. A set of complementary experiments was conducted, i.e., the growth of either a dyed seed in a non-dyed nutrient (Fig. 4a) or a non-dyed seed in a dyed nutrient (Fig. 4b). As the dyed seed was swelled in the non-dyed nutrient at rt, the resulting swelled sample was homogeneously colored. However, after annealing, the product displayed a deeper colored shell and a significantly less colored core, indicating that the polymerization of the absorbed nutrient solution occurred predominantly in the core, but not in the shell region. We used UV spectroscopy to analyze the distribution of the incorporated nutrient quantitatively. It was found that the shell region of the slices obtained from the middle section of the grown sample showed a similar absorption intensity with the seed, while the core region displayed remarkably less absorption intensity, indicating that the non-dyed nutrient was selectively incorporated in the core rather than the shell region. The same conclusion was obtained in the complementary growth experiment of a non-dyed seed in a dyed nutrient. The dyed nutrient was selectively integrated into the core region, making the shell region colorless (Fig. 4b). These phenomena further prove our hypothesis in Fig. 1d. It was worth noting that the photoisomerization of azobenzene-conjugated D4 did not make any contribution to the shape transformation since the experiments obtained in the dark condition show the same results.

Fig. 4: Characterization and control of grown products.
figure 4

a Growth of a dyed seed from a non-dyed nutrient solution (2 wt% of triD4 in D4). The seed was dyed with azobenzene-conjugated D4. The UV spectra on the right show the absorption of the circular slice (thickness: 1 mm) cut from the grown product at different regions with the dyed seed as the control. b Growth of non-dyed seed in dyed incubation solution. The seed and incubation solution are shown on the left, and the grown product is shown on the right. A blue background is used for observation, and dash lines highlight the boundary of the product. The scale bars in (a, b) are 3 mm. c Young’s modulus and growth index of grown products obtained from nutrient solutions with different cross-linker (triD4) concentrations. The growth index is defined as Wgrown/Wseed. d Growth index of grown products under different swelling times or growing cycles. The samples were swelled in nutrient solution for different times and stored under an inert atmosphere for 12 h to allow the liquids absorbed to homogeneously distribute throughout the samples, followed by annealing at 90 °C for 6 h. A growth cycle of the basic system includes full swelling (20 h) and annealing (90 °C for 6 h). e Plot of the ratio of axes (r/d) against the growth index. f Plots of the ratio of axes and growth index against the concentration of initiator in the seeds. The seeds were made from 2 wt% triD4 and 1 wt% initiator in (a, b, d, e). The data were obtained from eight independent measurements. Data are presented as mean values +/− SEM.

Full size image

Several material properties of the grown samples, e.g., the sample’s mechanical properties and growth index (Fig. 4c), could be easily tuned by varying the design of the nutrient solution. Growth in the same nutrient solution (2 wt% triD4), as was used for preparing the seed, led to a similar Young’s modulus: 0.58 MPa (seed) vs 0.56 MPa. On the other hand, increasing cross-linker concentration in the nutrient solution stiffened the grown samples. In contrast to the significant change in Young’s modulus, the growth index of the samples, defined as Wgrown/Wseed, does not vary noticeably with the composition of the nutrient solution, suggesting that the overall kinetics of incorporating the newly formed material from the nutrient solutions remained very similar and fairly independent of the concentration of the cross-linker. The growth index was also controllable. It could be finely tuned in a single growth cycle or significantly varied by increasing the number of growth cycles (Fig. 4d). The swelling time has proven to be a convenient parameter to precisely control the growth index in a single growth cycle, while increasing the number of growth cycles could be applied to induce a significant increase in the weight of the sample.

The shape transformation depended on the degree of growth. As shown in Fig. 4e, when the ellipsoidal products were taken as examples, the ratio of axes (r/d) significantly increased with the growth index. Besides swelling time, various other parameters could be used to control the size and the shape of the materials resulting from base-catalyzed growth: initiator concentration, cross-linking degree, and annealing temperature (Fig. 4e, f and Supplementary Fig. S11). Some parameters even allowed for tuning only the size of the grown products. For example, the growth index could be significantly increased by increasing feed initiator concentration, keeping the r/d ratio of the grown product nearly constant (Fig. 4f).

It is very instructive to compare the base- and acid-catalyzed19 systems. Both systems allowed for the post-modification of bulk compositions (then the resulting mechanical properties). In the acid-catalyzed system, the nutrients are homogenously integrated into the matrices because the fast percolation of the protons leads to their homogeneous distribution, which leads to the samples retaining their geometric shapes19. In contrast, the basic active species were substantially less mobile and much more susceptible to deactivation by adventitious CO2, resulting in selective nutrient incorporation in the core region of the base-catalyzed systems, and macroscopic transformation of the shape of the sample. In our previous study of the acid-catalyzed system, we have demonstrated the ability to modulate and control sizes of micropost arrays, in which the arrays are homogeneously magnified without changing the aspect ratio of the microposts, the ratio of the micropost dimension, and the distances between them (the pitch)19. In the current study, the base-catalyzed silicone growth selectively occurred in the core rather than the shell regions containing the microposts. Therefore, it could be expected that the microposts could retain their size when the substrate was expanded. This selective growth could be utilized to modulate the spacing of microposts. As shown in Fig. 5a, a square sheet seed with micropost arrays on its surface was fabricated by soft lithography as a seed. When this seed was grown into a sphere, the central area was expanded biaxially, resulting in the magnification of the inter-micropost distances in the x-y plane, while the middle region of the semi-sphere was expanded mostly uniaxially, leading to the elongation in the y distance without changing the size of the microposts. To deliver precise information of the microstructure, we combined the results from photo microscopy, confocal microscopy, and SEM. The expansion on the substrate would deform the vertical microposts into “volcanic” structures with tilting side faces. The tilting side faces could be observed in the SEM images. Since the titling surface would reflect light to induce dark pattern in confocal microscopy, we could clearly see the bottom edges of the microposts, which allowed us to provide the precise values of the bottom.

Fig. 5: Application of heterogeneous base-catalyzed growth of silicone.
figure 5

a Images of the micropatterns on the surfaces of the basic seed and its grown products. The images of the grown product were obtained from optical microscopy (left), SEM (middle), and confocal microscopy (right). Insets show the images of the samples on which the observed areas are highlighted. The schematic shows the mechanism of growth. b Restoration of large damage in a sample under base-catalyzed conditions. The sample was swelled in nutrient solution for 20 h and then annealed at 90 °C for 6 h. The damaged region is refilled. c Selective growth of the basic seeds: i, swelled sample was selectively heated on one side; ii, the seed re-bonded from a deactivated shell and a living circular-shaped center; iii, swelled sample with a crack cutting through the transparent shell from one side.

Full size image

Besides manipulating the patterns, the base-catalyzed heterogeneous growth of the silicones presents other appealing capabilities, e.g., large damage restoration. It is well-known that both acid- and base-catalyzed silicones display excellent self-healing ability because of the ongoing chain exchange reactions in the silicone networks23,24,25. However, the self-healing behavior is based on the tight contact between the separated surfaces such that chain exchange reactions can occur, which is difficult to achieve in the case of large damage31. Since the current system displays significant localized volume changes of the samples, it could be applied to restore damage significantly larger than a simple cut. As shown in Fig. 5b, a hole with a diameter of 3 mm was made in a sample with a width of 10 mm. After growth from the typical nutrient (2 wt% triD4), the hole was readily filled.

Compared with spontaneous heterogeneous growth, controllable heterogeneous growth induced by external stimuli could create different shapes without any template (Fig. 5c). For example, growth induced by localized heating one side of a square sheet resulted in a mushroom shape. In this transformation, polymerization occurred preferably in the heated region, which induced nutrient transport from the other side to the heated side (Supplementary Fig. S12). In addition, placing a defect, e.g., a cut, on the shell allowed for selective growth from the location of the defect (Supplementary Fig. S13). As previously discussed, when the chain exchange reaction is activated in the core region, nutrients from the shell are drawn into the core. Consequently, the shell experiences high tension while the core undergoes compression. In the presence of a defect on the shell, the slit will propagate once the tensile stress on the shell reaches a critical value. Due to the core being under compression, the opening will locally release this compressive stress, causing the core to bulge out from the opening. Besides these examples of spatially selective growth, one could capitalize on the self-healing ability of these materials, by placing the living seed in/on a deactivated sample to induce site-specific growth (Supplementary Fig. S14a). For example, we cut a living square seed using a hollow hole puncher, into two parts: a circular piece and a square piece with a circular hole in the center. The square piece was deactivated, and the circular part was placed back into it. An integrated sample was obtained after annealing treatment (self-healing). After growth, as expected, only the circular core expanded while the deactivated region retracted. It indicated that the growth region could be easily programmed by pre-localizing active species. As a complementary example shown in Supplementary Fig. S14b, a seed with a deactivated circular center and a living circular frame was fabricated and subjected to the growth conditions. It resulted in the formation of a torus-bowl-like product. Based on these results, we concluded the features of base-catalyzed heterogeneous growth in the current work compared to previous acid-catalyzed homogeneous growth: (1) geometric structure control, for example, a sheet into a sphere or a mushroom shape, rather than a bigger sheet; (2) polymerization-induced mass transport, a process that was not involved in homogeneous growth; (3) capability to change the distance of microposts in a pattern, rather than the size of the microposts observed in homogeneous growth; (4) significant molecular diffusion, as shown in our designed computer simulation mode.

Discussion

In summary, we have reported a class of self-growing silicone materials that undergo macroscopic transformations when they absorb externally provided nutrients to induce base-catalyzed heterogeneous growth. Such type of growth has its origin in the heterogeneous distribution of the active species due to their non-uniform deactivation at the seed’s shell and core regions. When these non-uniformly distributed active species induce polymerization of the nutrient compounds, significant mass transport occurs within the samples, leading to pronounced deformation. With a number of parameters available for tuning the swelling and polymerization, the size, shape, composition, and mechanical properties of the grown products can be controlled. We have also demonstrated that the heterogeneously growing silicone system can be applied to restoring large damage as well as modulating the shapes of the grown products. This post-fabrication heterogeneous growth strategy may open pioneering directions in fabricating complex structures at micro-to-macro scales.

Methods

Materials

D4 (98%, Sigma-Aldrich), trifluoromethanesulfonic acid (99%, Sigma-Aldrich), heptamethylcyclotetrasiloxane (tech-95, Gelest), trivinylmethylsilane (95%, Gelest), tetramethylammonium siloxanolate (Gelest), platinum(0)-1,3-divinyl-1,1,3,3-tertramethyldisiloxane in xylene (Pt catalyst, 2% of Pt, Sigma-Aldrich), perylene-3,4,9,10-tetracarboxylic dianhydride (97%, Sigma-Aldrich), 4,4’-dihydroxyazobenzene (97%, Synthon Chemicals GmbH) and solvents were used as purchased.

Instruments

Solution 1H spectra were measured in CDCl3 solution at 25 °C using Varian M400 400 MHz or I500C 500 MHz spectrometer. FT-IR spectra were obtained using a Bruker Vertex 70 FT-IR spectrometer equipped with a Pike ATR accessory. UV spectra were measured on Agilent 8453 UV-Vis spectrometer. Fluorescence images were obtained using a fluorescence confocal microscope LSM 710 (ZEISS). The mechanical properties were tested on Instron Model 5566 (Instron).

Preparation of living seeds

The mixture of monomer D4, cross-linker triD4, and tetramethylammonium siloxanolate was stirred at 80 °C under N2 for 5 min, which resulted in a very viscous liquid. The viscous liquid was poured into PS (polystyrene) containers/templates. The samples were sealed and annealed at 90 °C for 4 h to get the basic seeds. The seeds were kept in an inert environment before use. Typically, the seeds contained 2 wt% triD4 and 1 wt% initiator. To prepare dyed basic seeds, azo-conjugated D4 (0.1 wt%) was mixed into the starting mixture.

Deactivation of seeds

As-prepared seeds were annealed at 135 °C for 24 h under vacuum. This method was used to prepare deactivated basic seeds for swelling study. Alternatively, the seeds can also be deactivated by immersing them in a solution of acetic acid in THF (5 vol% of acetic acid) and then washing them with hexane. This method was used to deactivate the seeds used for selective growth, as shown in Fig. S11, because this milder treatment preserves the seed shape best.

Swelling

Typical swelling protocol was used with the nutrient solution (the mixture of D4 and triD4) as the solution (purged with dry N2 for 5 min before use). Eight samples (1 × 1 × 0.4 cm) were put together as a group for collecting data and the swelled samples were typically stored before annealing in sealed bottles for 12 h to allow for homogeneous distribution of absorbed liquids throughout the samples.

In designing the swelling/polymerization experiments, we assumed that the active species would not transfer from the sample into the nutrient solution during the swelling. To check whether this assumption is correct, we annealed the residual nutrient solutions at 90 °C for 10 h. After such annealing, the nutrient solutions maintained their liquid state without any visible increase in viscosity. Moreover, the 1H NMR spectra of the annealed and nonannealed residual nutrient solutions were indistinguishable.

Washing treatment: the swelled samples were taken out from the nutrient solution, followed by storage at room temperature for two days. The swelled samples were then immersed in hexane for one day, followed by tried in air. The immersion-dry process was repeated for three times to ensure full removal of unreacted compound entrapped within the sample.

Polymerization

The swelled samples were taken out of the nutrient solution, sealed in polystyrene containers, and then annealed at 90 °C for different times. The samples were cooled down and weighed to give the total weights (they included the weights of unreacted reagents present in the samples) that were used to plot the kinetic curves in Supplementary Fig. S8. Although the samples were sealed, some of the absorbed liquids (mainly D4 due to its low boiling point) still escaped from the samples and condensed on the bottom of the containers. Therefore, the weight of the samples decreases slightly with annealing time. Since the weight of the samples stabilized after six hours at 90 °C, these conditions were used for typical base-activated polymerization experiments.

For the deactivated controls, the swelled samples were treated under the same condition in a big container such that there was enough space for the liquid to evaporate and then condense. The total weights of the sample containing residual liquid were collected for the measurements.

For the dyed samples, the polymerization was conducted under two conditions: (1) UV light irradiation with a lamp (150 mW/cm2); (2) in dark. The dark condition was provided by annealing the sample in a sealed oven. For UV irradiation, the sample was annealed in the oven with a glass window which allowed UV light to pass through. The obtained samples did not show substantial differences.

Equilibrium polymerization of D4 in the presence of tetramethylammonium hydroxyl (1 wt%): the 100 mL mixture of D4 and tetramethylammonium hydroxyl (1 wt%) was stirred using a mechanical stirrer equipped with a force sensor recording the viscosity. The mixture was annealed at 90 °C and the viscosity was monitored. The maximum viscosity obtained was defined as 1 to normalize the polymerization process.

Growth of dyed basic seed in non-dyed nutrient solution

As shown in Fig. 4a, we swelled a dyed basic seed with a non-dyed nutrient solution. The dyed basic seed contains azo-conjugated D4 (a cross-linker rather than the monomer), which was swelled in a non-dyed incubation solution (2 wt% triD4) at room temperature. The resulting swelled sample was homogeneously colored. After the swelled sample was annealed at 90 °C for 4 h, the grown product was cut into thin slices for observation and UV spectroscopy measurements.

Growth of non-dyed basic seed in dyed nutrient solution

It was a complementary experiment to that described above. The typical basic seed (non-dyed) was swelled in a dyed nutrient solution (with 2 wt% triD4). A homogeneous color product was obtained. After annealing at 90 °C, the product displayed a deeper colored core and a significantly less colored shell, indicating that the polymerization of the absorbed incubation solution occurred predominantly in the core, but not in the shell region. In other words, the initiating species in the shell region had been quenched and/or removed.