Light-switchable swarming of biohybrid microrobots

Abstract

Controlling collective behavior in the microscale is essential for advancing autonomous robotic systems in complex environments. While biohybrid microrobotic swarms offer considerable promise for targeted therapeutic and remediation applications, their programmable assembly and collective behavior remain challenging. Here, we describe an attractive light-triggered approach for enabling reconfigurable swarming of biohybrid microrobots based on the green microalga Chlamydomonas reinhardtii (CR). Such reversible swarming behavior is realized by combining the wavelength-dependent assembly ability of CR and its inherent phototactic properties with light exposures through a series of different mask openings that define the desired swarm geometry. Changes in the projected light enable dynamic modulation of the swarm shape and size, including real-time swarm splitting and merging behaviors. The concept was explored toward artificial intelligence–assisted wound targeting applications through the creation of microrobot swarms customized to exposed wound areas. Such powerful swarming capabilities offer considerable promise for the collective behavior of biohybrid microrobots toward important practical applications.

Swarms of biohybrid microrobots, with diverse shapes, can be formed and actuated using different light patterns and colors.

INTRODUCTION

The field of microrobotics has been profoundly inspired by biological systems as natural evolution granted living organisms unique capabilities to propel, interact, and collectively perform tasks in a variety of dynamically changing environments (1). Such collective behavior of biological systems allows them to accomplish complex tasks that exceed the capabilities of single animals (2–8). Similarly, the creation of large swarms of multiple synthetic microrobots is a prerequisite for many practical biomedical or environmental applications of microscale robots (9). Different external fields (i.e., magnetic, acoustic, and light) are commonly used to trigger the swarming of artificial microrobots (10). These actuations can lead to dynamic assembly processes and to the formation of stable swarm patterns. Controlled changes in the swarm shape are used to meet the requirements of specific missions and enable a better task completion. Although natural swarming has been widely replicated with artificial microrobots (9, 10), mimicking the advanced level of complexity, adaptability, reversibility, and autonomy observed in natural systems remains largely unattained. For example, there is considerable interest in generating reconfigurable microrobotic swarms that perform complex collective tasks and can adapt rapidly to dynamically changing requirements and conditions. Changes in the swarm morphology are usually achieved by adjusting external inputs. Most of these studies have focused on the swarming of artificial microrobots. Biohybrid microrobots, combining biological components with synthetic functional materials, have recently emerged as a powerful class of microrobots due to numerous distinct advantages such as autonomous movement without the need of external stimuli and long-lasting propulsion across a wide variety of relevant environmental and biomedical media (11–14). Despite the rapid growth and advantages of biohybrid microrobots, little attention has been given to the collective behavior and the creation of defined assemblies of biohybrid microrobots.

In this communication, we demonstrate the light-triggered creation of diverse swarm patterns of biohybrid microrobots and the ability to dynamically and reversibly reconfigure their shape and size. Such unique swarming capabilities are demonstrated in the following sections using the biflagellated microalga Chlamydomonas reinhartii (CR). Biohybrid microrobots, based on functionalizing the CR microengine with a wide variety of materials, have been reported toward a variety of biomedical and environmental applications (15–17). CR displays unique light-dependent properties that make it attractive for diverse robotic applications. For example, CR shows positive and negative phototaxis toward self-regulating its photocatalytic activity, mainly through light-gated channelrhodopsins (ChR1, and ChR2) (18–20). This phototactic behavior can be used to guide the swimming direction of biohybrid microrobots (21). In addition, CR has wavelength-dependent receptors, like the blue-light phototropin cryptochromes as blue and red-light receptors (22–24). A blue/red light balance can thus modulate key functions of CR microorganisms, including their circadian clock, cell cycle, environmental acclimation, or biochemical profile (25–29). Blue and red irradiations can also trigger the reversible transition between free-swimming and surface association states of CR in a timescale of seconds (30). Such light switchable surface adhesion involves binding of the flagella to the substrate with an average adhesion force around 2 nN (31). However, the systematic formation of well-defined swarms of CR-based biohybrid microrobotic platforms has not been explored.

RESULTS

The formation of well-defined shape-changing microrobot swarms relies on the combination of wavelength-dependent assembly/dispersion of CR and distinct changes in the projected light for creating a variety of geometrical swarm patterns (Fig. 1). A schematic illustration of the experimental setup used for triggering the light-induced microalgae swarming through customized illumination is illustrated in Fig. 1A. By using different mask patterns (i.e., switching the projected light) and tuning the irradiation wavelength (between blue and red lights through different optical filters), it is thus possible to induce the light-switchable formation of well-defined swarms of CR-based microrobotic platforms. In this regard, a modified fluorescence microscope was used for simultaneous actuation and visualization of the microrobotic swarms. The microscope was modified to allow the introduction of custom laser-cut masks in the excitation light path to provide structured irradiation. Also, two optical filters (470 and 650 nm) were used to induce blue and red irradiation, respectively. Last, white light was applied in the opposite direction to enable real-time visualization with a camera. Notably, under blue irradiation, two effects are noted experimentally. First, because of the native phototactic activity of CR, negative phototaxis is triggered, leading to a collective migration to the upper plane (upon irradiating the sample from the bottom), confining the irradiated CR in the liquid-air interface. In addition, the bound state of algae is induced under this irradiation. Notably, under blue irradiation, microalgae may attach either to a solid surface or to other microalgae, forming microalgal clusters. In experimental conditions, because of the lack of a solid surface in the liquid-air interface, the microalgae are limited to forming algal clusters in the surface-bound state. High-magnification images of the light-induced clustering effect are displayed in fig. S1. Under adequate irradiation and confluency conditions, these two effects coalesce into the well-defined microrobot swarming behavior illustrated in Fig. 1B. Alternatively, red light irradiation triggers the planktonic (freely swimming) state of CR. This effect can be used to efficiently disperse the formed swarms by recovering the motility of individual microalgae, which ultimately leads to the dispersion of the formed swarms. Notably, this method allows the formation of swarms of algal microrobotic platforms with unprecedented resolution. To illustrate this highly efficient light-induced swarming, intricate light patterns of different Earth regions (Americas and Afro-Eurasia) were applied as shown in Fig. 1C. Here, a mask with the shape of the Americas region was applied first using blue irradiation (approximately 3 min), showing the capability of the method to translate complex features in the microscale into swarms of living algal biohybrid microrobots. After application of red light (approximately 5 min) and irradiating the sample with blue light using a mask depicting a different (Afro-Eurasia) Earth region, the next complex swarm was rapidly formed (within approximately 3 min). Last, after dispersing the Afro-Eurasia–shaped swarm and irradiating the microalgae using the first mask, it was possible to regenerate the first (Americas) swarm, demonstrating clearly the reversibility of the method.

Fig. 1. Light triggered swarming with CR microrobots.

(A) Schematic of the irradiation elements including: ultraviolet-visible (UV-vis) light source, blue/red filter, customized mask, dichroic mirror, magnification objective, algae sample, and projected image to charge-coupled device camera. (B) Simulation results of the swarm formation process (top). Schematic illustration of the processes involved in algal swarming (bottom) and illustration of the observed collective behavior under different irradiation modes: (i) initial step with random swimming, (ii) initial blue light–induced clustering (blue arrows) and swarm formation on the surface with combined blue light–induced clustering and negative phototaxis (orange arrows), and (iii) swarm dispersion (red arrows) with the red light. (C) Example of creating complex CR swarms in the microscale using masks patterned with different Earth regions (Americas and Afro-Eurasia microscope images from movie S1).

The versatility of the CR-based microrobotic swarming approach in creating and dynamically changing different swarm shapes upon switching the light pattern and wavelength is illustrated in Fig. 2 (A to F). This strategy offers reversible transformation of the swarm pattern in response to dynamically changing conditions toward diverse microrobotic applications. As illustrated in Fig. 2A, such shape reconfiguration relies on the use of diverse mask patterns, corresponding to different optical projections. For example, Fig. 2B illustrates the formation of different swarms by intercalating blue- and red-light actuation. In this case, different masks were used with intermittent blue (approximately 3 min) and red (approximately 5 min) irradiation. These images illustrate capability of CR-based biohybrids to form swarms with different shapes at command toward specific case of use. Similarly, Fig. 2C shows the intermittent formation of star-shaped swarms, illustrating such efficient swarm formation and dispersion in three consecutive cycles. These swarm assembly-dispersion capabilities can be repeated across multiple operational cycles. The swarms display good shape fidelity over three consecutive cycles with average Dice scores over 0.95, with a moderately intricate pattern (star shape), demonstrating the high precision of the reported mechanism (fig. S3). The biohybrid swarms can be dynamically assembled and dispersed to meet specific operational needs, allowing the algal biohybrid swarms to engage in their collective action for a predetermined time period and to disassemble completely upon completing their operation. Such reversible swarming capabilities were observed within different relevant environmental and biomedical media (fig. S4). The swarms can be dynamically assembled and dispersed according to the operational needs, allowing the algal biohybrids to engage in collective tasks for predetermined time periods and to completely disassemble the swarms at will. Figure 2D illustrates the main steps in the algae swarming process. In this case, a Nano Bioelectronics logo was designed using a computer-aided design (CAD) software. This design was then transferred to a black cardboard substrate using a laser cutter. Next, the mask was placed in a microscope setup, as previously described, and the built-in irradiation system was used to project custom light patterns according to the desired shape and wavelength. Notably, by using the optical elements in the microscope, different amplifications can be induced in the projected irradiation, allowing millimeter-sized masks to project features in the microscale. This allows the use of low-definition tools to generate the masks required for smart actuation in the microscale.

Fig. 2. Demonstration of light-switchable microrobotic swarm formation.

(A) Schematic representation of the swarm shape-switching process by changing the applied light projection using a series of different mask patterns. Sequential swarm formation experiments: (B) Formation of different swarms under intermittent blue and red irradiation. (C) Sequential formations of a star-shaped swarm under intermittent blue and red irradiations while keeping the same irradiation mask. (D) Experimental steps for swarm formation using the Nano Bioelectronics (NBE) logo as a mask. (E) Sequential formation and disassembly of swarms in the millimeter scale and their intensity profiles. (F) Formation of square-shaped swarms with different sizes with stacked intensity profiles. Swarming efficiency of algal swarms (approximately 5 mm in diameter) under different conditions (n = 3, reference data in fig. S2): (G) different microalgae concentrations, (H) different blue light intensities (as a percentage of the maximum light irradiation; 2 mW cm⁻²), and (I) swarm disassembly evolution after irradiation under different irradiation conditions (ambient light, darkness, and red light). Microscope images from movie S2. a.u., arbitrary units.

Figure 2E illustrates the unique assembly and dispersion capabilities of algal swarms in the millimeter scale. This was demonstrated by applying the masks directly on the sample holder without using the built-in optical magnification systems. This approach showed similar performance to that recorded with microscale swarms, demonstrating the versatility of the proposed mechanism. Alternatively, Fig. 2F demonstrates the application of square-sized masks ranging from 1 to 5 mm. Because of the limited number of algae in the given time (5 min) and irradiation (0.02 mW cm⁻²) settings, the microalgae aggregate preferentially on the edge of the irradiated area. This becomes apparent when comparing the stacked intensity profiles. This is in line with the light-induced clustering mechanism proposed here. In summary, this approach allows the formation of algal swarms from the microscale to several millimeters. This is especially important in practical applications as the required area for collective biohybrid actuation may expand up to several millimeters or more.

To estimate the evolution of the observed behavior under different conditions, the swarming efficiency was estimated as follows

$${\mathit{Eff}}_{\mathrm{S}}=I_{\mathrm{S}}·\raisebox{1ex}{$I_{\mathrm{S}}$}\!\left/ \!\raisebox{-1ex}{$I_{\text{BG}}$}\right.$$

where Eff_S is the calculated swarming efficiency, I_S is the measured intensity of the blue channel in the irradiated area after completing the swarming process, and I_BG is the intensity of the background corresponding to the unirradiated area around the swarm. Additional experiments were conducted to further understand the influence of key parameters involved in the formation and dispersion of algal swarms. First, the effect of algal density and, therefore, degree of confluence, was explored in Fig. 2G. Notably, a moderate algae concentration of 1.5·10⁶ algae ml⁻¹ leads to the best swarming efficiency in connection to 5-min irradiation time. Also, lower irradiation power leads to well-defined swarm formation, whereas higher irradiation power results in less efficient swarming due to an increased algal accumulation in the nonirradiated areas. In this line, higher algae concentration (2.0·10⁶ algae ml⁻¹) leads to overpopulated swarms that extend to the nonirradiated areas. In contrast, a lower algae concentration (1.0·10⁶ algae ml⁻¹) translates into less available algae for engaging in the swarm formation process. Notably, a higher irradiation power increases the swarm formation efficiency, which is attributed to the presence of more algae in the swarming plane or to increased clustering efficiency.

Next, the swarm formation evolution time was evaluated using different irradiation powers and a fixed algae density (1.5·10⁶ algae ml⁻¹). As shown in Fig. 2H, under low-power irradiation, the swarm formation follows a sigmoidal trend. Notably, in these irradiation conditions, the initial swarm formation is low until enough algae are in the irradiated area to start the clustering process followed by a transient swarm-formation stage. Last, algae enter a steady state, with the algae engaging the swarm formation and leaving the swarm forming a dynamic equilibrium. This is in line with the light-induced clustering mechanism that will be further expanded in the corresponding section of this manuscript. Real-time transient algae density in the liquid-air interphase during the swarm formation under blue light followed a similar trend with sigmoidal increase of the algae density followed by a dynamic equilibrium in the steady state (fig. S5). An important aspect regarding irradiation intensity is to evaluate the possible photodamage induced to the algae. Specifically, photosystem II is particularly labile to photodamage but benefits from specific photoreceptor-induced protection mechanisms (32). In particular, photoprotection in CR is mediated through a blue-light photoreceptor-dependent signaling pathway that induces a high energy–quenching effector to dissipate excess light energy and prevent photodamage (33). In this line, CR demonstrated negligible phototoxicity under high-intensity irradiation above the required threshold to induce swarming for up to 4 hours, which is in line with our experimental observations (34). However, possible phototoxicity should be evaluated for long time and high-energy applications. The swarm disassembly process under different conditions was also explored using a higher irradiation power (Fig. 2I). In this case, swarms were induced, and, after stopping the irradiation, the evolution of swarming efficiency was evaluated in different conditions (i.e., under ambient light, with applied red light, and without any external irradiation). The swarm evolution results illustrate that the generated swarms are mostly stable with a small decrease in the swarming metrics (~85% remaining swarming efficiency after 5 min under ambient light). Depriving the algae of external stimulation leads to a rapidly decreasing swarming efficiency, reflecting that algae are entering the light-harvesting stage for continuing the photosynthetic process. Last, as expected, red-light irradiation accelerates the swarm dissociation process, leading to around 50% disassembly within 1 min and a complete swarm disassembly after 2 min.

To this point, the intermittent swarming and dissociation of CR-based microrobot platforms have been explored. Certain applications require transient control on the formation and operation of microrobots. In this regard, the behavior of algal swarms under continuous operation with blue light is explored in Fig. 3. Such ability to dynamically alter the position, size, and shape of a microrobotic swarm is of paramount importance to a wide range of potential practical applications. The ability to reposition the swarm was tested initially by changing the spatial position of an arrow-shaped projected light while keeping the blue-light actuation (Fig. 3A). These images illustrate swarm translations by tens of millimeters, with the swarms maintaining their shape and density. The displaced swarm retained around 70% of the original swarm intensity after two sequential 10-mm displacements with continuous blue-light operation. Notably, the efficiency of this repositioning mechanism can be potentially boosted with a more advanced illumination system, with tailored step sizes, according to the desired total travel length and time of operation.

Fig. 3. Dynamic microrobotic swarm assembly.

(A) Spatial evolution of microrobotic swarms with different irradiation source positions. Evolution of a diamond shape swarm under (B) diverging and (C) converging structured illumination, leading to splitting and merging of the biohybrid swarms, respectively. (D) Irradiation of a microscale swarm with structured irradiation of different sizes. (E) Demonstration of irradiation with moving and diverging light sources in the microscale. (F) Morphing experiments by applying different shapes while keeping the blue irradiation and subsequent disassembly of the generated swarm. Microscope images from movie S3.

In addition, Fig. 3 (B and C) shows the evolution of a diamond-shaped swarm under diverging and converging structured illumination, respectively. As noted in the corresponding intensity profiles, the diamond swarm structures are able to split into triangle ones, while the triangle structures rapidly merge into the diamond-shaped swarm within 5 min, without overlapping or noticeable residual algae in the nonirradiated areas, leading to relocation efficiencies of over 95% in milliscale swarms (Fig. 3B) and close to 70% in microscale swarms (Fig. 3C). These data demonstrate the dynamic nature of millimeter-size swarms with an ability to adapt “on-the-fly” to specific operational requirements in a timely and efficient manner. Subsequently, the dynamic capabilities of algae-based swarms were explored in the microscale. First, the size of an assembly was tailored dynamically—from 100 to 200 μm—by adapting the size of the corresponding mask without inducing swarm dispersion (Fig. 3D). Next, swarm repositioning capabilities were explored in the micrometer scale in two experiments, as displayed in Fig. 3E. In this case, a 100-μm swarm was initially repositioned by displacing the position of the irradiated area by 100 μm. Subsequently, the 100-μm swarm was split into four quarters of around 50 μm by changing the shape of the applied mask. In both cases, the swarms display good reconfigurability with minimal memory effect. Last, Fig. 3F demonstrates the morphing ability of the generated swarms in the microscale. In this case, the generated swarms were subjected to continuous blue-light irradiation while using different masks (and 5-min intervals), without triggering the dispersion of the previous swarm. Such algae swarms display good dynamic conversion capabilities, leading to four different well-defined swarm shapes.

In this regard, a probabilistic model was developed, and the results were compared and confirmed experimentally. Briefly, algae have been simulated as objects with stochastic motion, and the probability of joining or leaving a cluster is defined by their current clustering and irradiation state. Details on this simulation are given in the Supplementary Materials. The results from these studies are presented using a colormap, indicating the probability of finding a particle in a grid, as shown in Fig. 4.

Fig. 4. Detailed of the swarm formation and disassembly dynamics with related microscopy images (top) and probability maps from the corresponding simulations (bottom).

(A) Formation of a square-shaped swarm, (B) disassembly of the swarm under red light irradiation, and (C) formation of the outline of different shapes (e.g., square, triangle and circle).

Notably, the explored probabilistic mechanism fits the experimental evidence recorded by inducing the formation of a square-shaped swarm focused below the swarming plane (Fig. 4A). This highlights the relevance of the clustering mechanism in the swarm formation process followed by a reconfiguration of the clustered algae on the upper layer due to the collective phototactic effect. The interplay between the wavelength-triggered clustering effect and the intensity of the negative phototactic effect is critical for obtaining well-defined microscale features. Negative phototaxis is key to spatially confine a well-defined algae swarm in a thin layer in the liquid-air interface. This is supported by the fact that, by increasing the intensity of the blue-light irradiation, the algae clusters can migrate to the nonirradiated region while keeping their clustering capabilities (see related simulation in fig. S6). This mechanism allows to generate forbidden areas in the microscale, on command. In this line, the application of low-intensity and high-intensity lights or of spatially differentiated wavelengths can further enhance the capabilities of algal swarms (fig. S7). The experimental observations support the proposed probabilistic state transition mechanism in connection to the disassembly of algal swarms. As illustrated in Fig. 4B, both in the microscopic images and theoretical simulations, the swarm density decreases upon triggering in the disassembly process, followed by the dispersion of algae around the original swarm. Similar swarm forming dynamics—with different shapes—is illustrated in Fig. 4C, demonstrating the versatility of the proposed light-induced swarming mechanism toward diverse features and geometries.

The proposed mechanism holds great potential to tailor the size, position, and shape of microrobotic swarms to perform complex tasks. One such potential operation is enhanced wound healing processes in connection to precise swarm localization. In this regard, the collective behavior of microalgae was combined with artificial intelligence (AI) to grant the individual algae with complex contextual information. Figure 5A displays such AI-guided generation of the wound-tailored algal swarms (i) and the preparation of poly(lactic-co-glycolic acid) (PLGA)–algae microrobots by leveraging electrostatic interactions between the negatively charged algal surface and positively charged polylysine (PLL)–modified PLGA nanoparticles (ii). The size and zeta potential of PLGA and PLL-PLGA particles were confirmed before binding to the algae microengines (fig. S8). Also, the presence of PLL-PLGA nanoparticles was confirmed with scanning electron microscopy imaging (fig. S9). PLGA biohybrids have been successfully used earlier for advanced antibiotic delivery (35) and are used here as a proof-of-concept drug delivery platform. First, we demonstrated the successful reversible swarming capabilities of unmodified algae in simulated wound fluid (SWF) [Fig. 5B, (i)] and of the PLGA-functionalized biohybrids [Fig. 5B, (ii)]. While negligible differences in the swarming kinetics were observed using SWF media (compared to experiments in DI water), the disassembly process was significantly (threefold) longer in the case of the algae-PLGA microrobots, reflecting the electrostatic forces between the PLL-PLGA nanoparticles (~20 mV zeta potential) and algae surface (−15 to −20 zeta potential) (36). Electrostatic interactions can lead to increased aggregation through electrostatic bridging between positively charged nanoparticles and negatively charged microalgae (37). This effect may impose an additional hurdle toward cluster disassembly in the single to few-algae level. The PLGA-algae microrobots were successfully loaded on a medical tape substrate [Fig. 5B, (iii)], demonstrating the ability of the wavelength-dependent surface bonding to transfer drug-loaded biohybrids to solid substrates. The loading of PLGA-algae microrobots on the tape was further confirmed by fluorescent imaging of DiO dye loaded onto the nanoparticles (fig. S10). Next, to create suitable microrobot swarms for the wound dressing applications, an infection-suspicious wound was photographed and analyzed with an open-source AI segmentation algorithm (Segment Anything Model). Results of the segmentation process are displayed in Fig. 5C (i), with acceptable stability scores ranging from 0.89 to 0.96 for the wound tissues. As illustrated in Fig. 5C (ii), this model can efficiently segment the image into different regions that can be processed as CAD sketches to generate the corresponding masks. Once the images are segmented, the highlighted regions can be classified according to their clinical assessment. Once the image is segmented and the target region is defined by the AI algorithms, a customized mask can be prepared. The resulting mask can thus be used to induce swarming, with its shape and size defined by the actual wound area. Figure 5D shows the light-directed transfer process of algae swarms. First, algae are directly printed on a piece of medical tape by irradiating the sample with blue light for 5 to 10 min using the custom-made mask for the corresponding target [Fig. 5E, (i)]. An alginate phantom skin containing a simulated wound was used as a model for the biohybrid delivery experiments [Fig. 5E, (ii)]. Once the desired algae-based features are printed onto the medical tape, the tape can be directly applied to the desired treatment area [Fig. 5E, (iii)]. After placing the biohybrid-loaded medical tape on the simulated wound with SWF, the biohybrid swarm can be released from the tape to the wound location by using red-light irradiation through the corresponding customized mask. Under red irradiation, about 90% of the biohybrids are released to the wound cavity in less than 2 min, demonstrating the high efficiency of the light-based algae delivery mechanism (fig. S11). However, the loading and release kinetics heavily depend on substrate properties, such as charge, hydrophobicity, roughness, or porosity, among others. In this sense, loading and release conditions should be optimized for different substrates in real application scenarios. Last, a blue light is applied to assemble the algae onto the wound surface; a schematic representation of the algae release process is presented in fig. S12. After finalizing the biohybrid transfer process, it is possible to confirm visually the absence of biohybrids on the medical tape surface and the presence of biohybrid microrobots on the wound surface [Fig. 5E, (iv)]. In addition, experiments with nonirradiated biohybrids, biohybrids irradiated with blue light to trigger the assembly, and irradiated with red light to trigger the release, demonstrate that the swarming and deployment of biohybrids from/to the substrates do not influence their loading (fig. S13). This delivery approach is expected to be a highly versatile for diverse biohybrid microrobot carriers in relevant medical applications. Notably, the most relevant limitation of this method is light penetration, which currently restricts its applicability to topical use. Nonetheless, through a mechanistic understanding of switchable adhesiveness, unexplored biomedical applications could be envisaged toward fully exploiting the unique capabilities of microalgae in further advanced therapies.

Fig. 5. Demonstration of algae swarming for tailored wound dressing.

(A) (i) Schematic of the smart algae-based bandage concept and (ii) preparation of PLGA-algae microrobots for tailored wavelength-induced delivery. (B) Reversible swarming of (i) unmodified algae in SWF, (ii) PLGA-algae biohybrids, and (iii) PLGA-algae biohybrids on a medical tape substrate. (C) AI-assisted generation of custom masks for wound dressing: (i) AI-based image segmentation with segment number and stability scores and (ii) characterization of tissue types. (D) Schematic of the preparation of the smart algae-based bandages by swarming biohybrids on medical tape with a mask corresponding to the shape and size of the target wounded tissue. (E) Demonstration of the light-enabled algae transfer mechanism: (i) Prepared swarm-based smart bandage based on a medical tape platform, (ii) simulated wound on a skin phantom, (iii) algae-loaded medical tape on a simulated wound, and (iv) transferred algae on the surface of the simulated wound with SWF. Microscope images from movie S4.

DISCUSSION

In summary, we presented an innovative light-switchable swarming strategy that allows swarms of algae-based microrobots to change their shape in response to changing requirements and conditions. These previously unexplored reversible swarming capabilities of biohybrid microrobots have been realized by combining light-switchable localized clustering with the phototaxis-enabled spatial confinement of the CR microalgae. Such coupling enables precise control of the swarm pattern formation by dynamically alternating the projected blue and red lights in connection to customized light masks according to the operational needs. As a result, the biohybrid microrobot swarms can dynamically change their morphology, size, and position. The reversible nature of the generated swarms and their remarkable versatility and reconfigurability hold considerable promise for a myriad of possible microrobotic applications. Furthermore, image analysis can be used to identify target areas and create tailored swarms of biohybrids. This grants the formed swarms with contextual information, expanding the applicability of this swarming mechanism. Numerous applications, ranging from biomedical to environmental, will greatly benefit from the developed light-switchable collective control of biohybrid microrobots. Realizing the full potential and promise of such unique “on-off” switchable swarming capabilities would benefit from deep understanding of the processes involved in the swarming processes of these and other biohybrid microrobots.

MATERIALS AND METHODS

Microalgae culture

Wild-type C. reinhardtii acquired from Chlamydomonas Resource Center (reference CC-409) were cultured in sterilized tris-acetate phosphate (TAP) growth media (Gibco, ref. A13798-01). The cultures were kept at 25°C under natural day/night light cycles. Algae density was estimated using the optical density at 660 nm.

Light-induced swarming

Microscale swarms were generated using a solid-state white-light source (Lumencor) properly filtered with a Cy5-404A-NTE-ZERO (Semrock) and a B-2A (Nikon) filter blocks mounted on an Eclipse Ti (Nikon) microscope. Images were recorded with a complementary metal-oxide semiconductor camera (Hamamatsu). Milliscale swarms were induced with the light-emitting diode system in an Evos FL fluorescence microscope (Thermo Fisher Scientific), and images were taken in a light box (Finnhomy) using a mobile phone camera (iPhone 16 Pro). In all cases masks were prepared using an F1 laser engraver (xTool).

Image analysis

Image analysis was carried out using the ImageJ software. Intensity maps were processed and plotted using OriginPro.

Numerical simulations

Custom Python scripts were developed to simulate the interactions between particles depending on their irradiation conditions. The simulation results were stored in CSV files. For illustration purposes, another custom Python script was developed to average the results of all the simulation files and represent the probability map with a predefined colormap.

Some assumptions were made to design the theoretical cell-interaction model leading to wavelength-dependent swarm formation and dissociation. First, the area of work was limited to 10 superimposed independent two-dimensional spaces, greatly simplifying the computing power and number of entities required to run the simulations. In this case, the results of the simulation correspond to the projection of the algae solution. Also, individual algae were simulated as free particles with a noisy ballistic trajectory defined by $x_{\mathrm{i}}(t+1)=x_{\mathrm{i}}\left(t\right)+v_{\mathrm{i}}\text{cos}\left[{\mathrm{\theta }}_{\mathrm{i}}\left(t\right)\right]$, and $y_{\mathrm{i}}(t+1)=y_{\mathrm{i}}\left(t\right)+v_{\mathrm{i}}\text{sin}\left[{\mathrm{\theta }}_{\mathrm{i}}\left(t\right)\right]$, where x_i and y_i correspond to the x and y position of the algae i at a given time and the angle (θ_i) is defined as ${\mathrm{\theta }}_{\mathrm{i}}\left(t\right)={\mathrm{\theta }}_{\mathrm{i}}(t-1)+{\mathrm{ϵ}}_{\mathrm{i}}\left(t\right)$, where ϵ_i corresponds to a random angular perturbation drawn sampled from a uniform distribution in the range of the angular variation amplitude. Interactions between algae are tracked by evaluating the Euclidean distance between particles, and overlaps are handled by applying a displacement to each particle equal to half the overlap distance along the line connecting their centers. When two objects (single algae and/or clusters) interact, the state of each particle is described by a two-state Markov chain with a transmission matrix

$$P=\left[\begin{array}{cc}1-p_{\text{join}}& p_{\text{join}}\\ p_{\text{leave}}& 1-p_{\text{leave}}\end{array}\right]$$

where p_join and p_leave are defined by the environmental conditions (e.g., irradiated with blue light, irradiated with red light, and nonirradiated). Cluster joining and leaving probabilities are estimated iteratively by comparing the swarm probability evolution in our system and may not be directly translatable to the actual processes in the single cell level. Last, a stochastic Brownian-like drift is applied to each cluster in addition to the interactions with free particles to consider the mechanical interactions between algae that make the clusters move and interact as cohesive bodies.

PLGA NP creation and DiO tagging

Twenty milligrams of PLGA was suspended in 1 ml of acetone in a 7-ml glass scintillation vial with a Teflon cap and sealed with parafilm. A second 7-ml vial containing 1 ml of 10 mM tris buffer was placed in an ice bath. After sonication for 1.25 min at 14°C, 2 μl of DiO dye was incorporated into the PLGA solution. The PLGA solution was then rapidly pipetted into the tris buffer solution. The resulting solution was then placed under vacuum for 1 hour. After this, the solution was centrifuged (3000 rpm, 3 min) and the supernatant was tested with a zetasizer. Particles were then checked under fluorescence microscopy with green fluorescent protein filter to confirm DiO dye presence.

PLL surface modification

The PLGA solution was centrifuged (12700 rpm, 3 min) to form a nanoparticle pellet. After supernatant removal, the nanoparticles were resuspended in 1 ml of ultrapure water (Gibco). PLL solution (1%) was obtained and diluted to 0.01% with TAP buffer. In a 20-ml glass vial, 1 ml of PLL 0.01% was added along with a micro stirrer bar. With constant stirring at 700 rpm, 1 ml of PLGA nanoparticles was then incorporated dropwise. The solution was then left to stir for 1 hour before washing via centrifugation (three times washing, 12700 rpm, 3 min, resuspension with ultrapure water between washes). In the final wash, the nanoparticle pellet was resuspended in 500 μl of ultrapure water.

Algae binding

PLGA DiO PLL nanoparticle solution (350 μl) was incubated with 1.4 ml of algae (6 million/ml) for 1 hour. The resulting solution was then exposed to red light for 30 min.

Dice coefficient evaluation for generated swarms

Swarms were segmented using the SAM library to determine the swarm shape with high fidelity (~99%), the obtained segments are repositioned (scale, rotation, and position) using OpenCV to account with misaligned with the theoretical applied pattern, and, finally, the Intersection over Union (IoU) was calculated using NumPy. The Dice coefficient for each swarm is calculated as Dice = 2 · IoU / (1 + IoU).

Supplementary Materials

The PDF file includes:

Figs. S1 to S13

Legends for movies S1 to S6

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S6