RESEARCH ARTICLE www.advmat.de On-Command Disassembly of Microrobotic Superstructures for Transport and Delivery of Magnetic Micromachines Fabian C. Landers, Valentin Gantenbein, Lukas Hertle, Andrea Veciana, Joaquin Llacer-Wintle, Xiang-Zhong Chen, Hao Ye,* Carlos Franco, Josep Puigmartí-Luis, Minsoo Kim,* Bradley J. Nelson, and Salvador Pané* 1. Introduction Magnetic microrobots have been developed for navigating microscale Over the last two decades, significant mile- environments by means of remote magnetic fields. However, limited stones have been accomplished in the propulsion speeds at small scales remain an issue in the maneuverability of field of biomedical magnetic small-scale these devices as magnetic force and torque are proportional to their magnetic robots.[1–3] Researchers initially focused volume. Here, a microrobotic superstructure is proposed, which, as on exploring basic magnetic micro- and analogous to a supramolecular system, consists of two or more microrobotic nanoarchitectures, including structures like units that are interconnected and organized through a physical (transient) nanowires, helices, and microparticles.[4] Their primary goal was to understand component (a polymeric frame or a thread). The superstructures consist of the fundamental principles governing mo- microfabricated magnetic helical micromachines interlocked by a magnetic tion at the micro- and nanoscale and gelatin nanocomposite containing iron oxide nanoparticles (IONPs). While to engineer strategies for achieving con- the microhelices enable the motion of the superstructure, the IONPs serve as trolled movement using external magnetic heating transducers for dissolving the gelatin chassis via magnetic fields. Building on these initial investi- gations and leveraging advancements in hyperthermia. In a practical demonstration, the superstructure’s motion with magnetic navigation, material science, and a gradient magnetic field in a large channel, the disassembly of the manufacturing, small-scale roboticists are superstructure and release of the helical micromachines by a high-frequency now pioneering the development of sophis- alternating magnetic field, and the corkscrew locomotion of the released ticated small-scale magnetic machinery. helices through a small channel via a rotating magnetic field, is showcased. These systems incorporate integrated func- This adaptable microrobotic superstructure reacts to different magnetic tional building blocks,[5] such as containers for carrying therapeutic agents or cells,[6–11] inputs, which can be used to perform complex delivery procedures within stimuli-responsive materials for tissue ab- intricate regions of the human body. lation or cell stimulation,[12,13] contrast agents,[14] and/or sensing elements.[15–17] F. C. Landers, V. Gantenbein, L. Hertle, A. Veciana, J. Llacer-Wintle, H. Ye, X.-Z. Chen C. Franco, M. Kim, B. J. Nelson, S. Pané Yiwu Research Institute of Fudan University Multi-Scale Robotics Lab Yiwu, Zhejiang 322000, P. R. China Institute of Robotics and Intelligent Systems J. Puigmartí-Luis ETH Zurich Departament de Ciència dels Materials i Química Física Tannenstrasse 3, Zurich 8092, Switzerland Institut de Química Teòrica i Computacional E-mail: haoyeh@ethz.ch; minkim@ethz.ch; vidalp@ethz.ch University of Barcelona X.-Z. Chen Martí i Franquès 1, Barcelona 08028, Spain Institute of Optoelectronics J. Puigmartí-Luis Shanghai Frontiers Science Research Base of Intelligent Optoelectronics Institució Catalana de Recerca i Estudis Avançats (ICREA) and Perception Pg. Lluís Companys 23, Barcelona 08010, Spain Fudan University Shanghai 200438, P. R. China The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.202310084 © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. DOI: 10.1002/adma.202310084 Adv. Mater. 2023, 2310084 2310084 (1 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Despite the undeniable progress in the field, substantial bar- 2. Results and Discussion riers lie ahead in the journey to transitioning these small- scale robots from laboratory settings to the operating the- 2.1. Design and Fabrication of Microrobotic Superstructures atre. Challenges include the development of magnetic naviga- tion systems compatible with healthcare facilities and imaging Figure 1 schematically illustrates the concept of targeted deliv- equipment,[18,19] as well as issues concerning the implantation ery using a microrobotic superstructure prototype, navigating procedure,[20] tracking and monitoring of these devices,[21] the through a branched vessel system using external magnetic fields therapeutic dosing for a specific condition or disease,[22] or bio- and magnetic gradients. The superstructure comprises a poly- compatibility and biodegradability features.[23] A primary issue meric chassis made of an IONP/gelatin composite, which serves concerns the maneuverability of these devices once these are im- as structural scaffold. This chassis carries a collective of ferro- planted in the vascular system. As magnetic forces and torques magnetic iron helical micromachines, which enable successful are directly proportional to the magnetic moment of the de- steering of the superstructure through a large diameter vessel vice, small-scale robots (<100 μm) suffer from limited propulsion (≈400 μm) with magnetic field gradients at relatively high speeds. speeds, particularly in ultralow Reynolds number regimes, mak- At a bifurcation at which the vessel is branching into smaller scale ing them unsuitable for practical biomedical applications.[24–26] conduits (≈80 μm), the superstructure cannot further advance Conversely, larger designs (100–1000 μm) can travel faster but due to its size. Then, the chassis can be dissolved via magnetic cannot navigate through confined biological conduits. To address hyperthermia of IONPs, releasing its interlocked microhelices. these issues, some solutions have been proposed. For example, These helices are capable of propelling through smaller branches Huang and co-workers have proposed a magnetic microtrans- by corkscrew locomotion, using rotating magnetic fields.[33] In porter comprising an Archimedes screw pumping mechanism the future, we envision the application of these superstructures capable of transporting and delivering magnetic micro particles for drug delivery and minimally invasive interventions in the and microhelices.[27] However, this system relies on a nondegrad- brain’s vasculature (e.g., in the arterioles), where they could reach able large structure, thus requiring retrieval after the magnetic locations currently inaccessible to commercial endovascular de- agents have been delivered. In another approach, biocompatible vices. and biodegradable microrobots were fabricated by Kroupa and The manufacturing of the interlocked microrobotic super- co-workers, culminating in a “mothership” capable of carrying structures is schematically depicted in Figure 2A,B.[32] First, a cargo. While this approach is conceptually enticing, the fabri- conductive copper-coated glass substrate was covered with a pho- cated robots still require the integration of components to be ac- toresist layer. Direct laser writing, also known as two-photon tively steered as they are propelled by hydrogen peroxide in the lithography, was used to create a 3D template within the pho- environment.[28] A recent strategy proposes the use of swarms toresist layer with two types of microchannels that can be in- of magnetic micro- or nanoparticles, which potentially can swim terweaved (Figure 2A). Microchannels type A, which are used cooperatively in fluids thanks to their magnetic dipole–dipole for iron filling (i.e., helical micromachines), penetrate through interactions.[29,30] Interestingly, the morphology and locomotion the entire photoresist layer to establish electrical connectivity be- mechanism of the swarms can be adjusted from ribbons to clus- tween the iron source (electrolyte) and the conductive substrate. ters as a function the applied magnetic inputs. A main drawback The microchannels type B, made for gelatin casting (i.e., the chas- from these assemblies is that forces in the fluid vasculature of sis), are designed to avoid any contact with the conductive sub- large animals or humans will likely disrupt and disaggregate the strate. We conducted a parameter study (laser power and scan assembly, which could be disadvantageous in the context of med- speed) to identify the ideal exposure parameters for writing the ical procedures. 3D channels, as detailed in Figure S1 and Table S1 (Support- Here, we propose microrobotic hierarchical superstructures ing Information). Underexposure resulted in inadequate electri- consisting of magnetic helical micromachines interlocked with a cal connections and inconsistent filling during subsequent elec- thermally responsive transient magnetopolymer chassis. By com- troplating and casting steps. Overexposure, conversely, lead to bining iron oxide nanoparticles (IONPs), possessing either poly- the coalescence of the channels and subsequent uncontrolled acrylic acid (PAA) or polyethylene glycol (PEG) as ligands, we electroplating into adjacent channels. Both the microrobot de- enabled the disassembly and release of the helical microcompo- sign and printing parameters were collectively optimized to es- nents on-command by applying high-frequency magnetic stimu- tablish the ideal design and fabrication conditions, as shown lation, which triggers the dissolution of the chassis by hyperther- in Figures S2 and S3 (Supporting Information). Because of the mia. The helical devices are made of metallic iron having supe- tendency of the photoresist to crack under the stress induced rior magnetic properties, as this metal exhibits the highest satura- by subsequent heating and cooling cycles, the spacing between tion magnetization among magnetic materials with biocompat- the microrobots was increased and the temperature was gradu- ibility features.[31] To fabricate these architectures, we capitalize ally ramped (Figure S4, Supporting Information). By fine tun- on our previous assembly-free manufacturing method that com- ing the microchannel design, the electroplating process allowed bines metal electrodeposition and polymer casting in 3D tem- to only fill the microchannels A with iron. During this pro- plates obtained by 3D direct laser writing.[32] This method allows cess, we monitored the plating current to identify the growth of for building 3D complex mechanically interlocked metal–organic the plated microstructures, as evidenced in Figure 2C. As the microrobots. The proposed smart superstructures could be used current increases proportionally to the area parallel to the sub- to navigate and deliver small magnetic helical micromachines to strate, we can recognize the cut-off point of the plating process, intricate small vessels and capillaries, thereby allowing them to which prevents plating over the template. To fabricate the chas- access difficult-to-reach anatomical sites within the human body. sis, a composite consisting of gelatin with dispersed magnetic Adv. Mater. 2023, 2310084 2310084 (2 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 1. Schematic diagram of targeted delivery using microrobotic superstructures. A) The prototype of the microrobotic superstructure is composed of an IONP/gelatin composite chassis, which carries pure iron helical micromachines. B) In a larger channel, a gradient magnetic field propels the assembled chassis. Upon reaching a small channel, the helical micromachines are disassembled on-command by dissolving the chassis through the ap- plication of magnetic hyperthermia. An alternating high-frequency magnetic field (HF field) induces magnetic hyperthermia by the IONPs. Subsequently, the disassembled helical micromachines continue their movement within the smaller channel, exhibiting a corkscrew motion induced by a rotating magnetic field. nanoparticles was molded into the remaining microchannels B. coils are less favored and turn into assembled 𝛽-sheet structures, After curing the gelatin overnight, the entire structure was re- which stabilize the hydrogel network, allowing it to preserve its leased by dissolving the photoresist layer in acetone. Optical mi- solid-like features even at elevated temperatures.[35] However, croscope images (Figure 2B) verified the successful fabrication when gelatin was mixed with magnetic nanoparticles, gelatin of the microrobot, which comprises the magnetic gelatin com- structures could dissolve upon heating (see further information posite chassis interlocked with six electroplated iron helical mi- in Section 2.3). cromachines. Scanning electron microscope (SEM) and energy- dispersive X-ray spectroscopy (EDX) analyses further verified the composition of the superstructure (Figure 2D). 2.3. Synthesis of Optimized Magnetic Nanoparticles for Gelatin Composite Dissolution 2.2. Pure Gelatin Chassis Stability and Solubility To develop a chassis capable of solubilization through magnetic hyperthermia and releasing its mechanically interlocked iron he- The concentration of gelatin in the chassis significantly influ- lical micromachines, we incorporated magnetic nanoparticles enced its mechanical stability. Lower concentrations of gelatin within the gelatin matrix to serve as remotely actuated heat trans- (≤5%) yielded microstructures with poor mechanical stiffness, ducers. The magnetic IONPs were synthesized through the ther- which tended to collapse likely due to electrostatic interactions mal decomposition of tris(acetylacetonato)iron(III) using a pro- between the helical plated parts (Figure 2E and and Figure S5, tocol developed by Sun et al.[36] (see details of the synthetic proce- Supporting Information). Conversely, structures with improved dure in the Experimental Section). Figure 3A displays a transmis- mechanical integrity were attained by increasing the gelatin con- sion electron microscopy (TEM) image of the synthesized parti- centration in the casting solution. Based on these experiments, cles, showing a highly uniform cubic morphology characterized we concluded that casting formulations with a gelatin content of by an average edge length of 17.2 ± 1.1 nm and a height of 23.3 ± 10% are adequate to manufacture microstructures with the re- 1.3 nm, respectively (Figure S9a, Supporting Information). With quired mechanical stability. Interestingly, pure gelatin cast struc- a highly crystalline inverse spinel structure (XRD analysis in tures exhibited a high resistance to solubilization in water even at Figure S9b, Supporting Information), these magnetic nanopar- elevated temperatures of more than 70 °C (Figures S6–S8, Sup- ticles exhibited soft ferromagnetic behavior, with a coercivity of porting Information). As reported earlier, gelatin grown in con- 6.72 Oe, as shown in the magnetic hysteresis loop (Figure 3B). finement can display distinct mechanical properties and solubil- The synthesized IONPs were initially coated with oleic acid, ity characteristics in comparison to its bulk counterparts.[34] In making them hydrophobic. To ensure efficient dispersion of bulk gelatin, the microscopic secondary structure consists of a these hydrophobic nanoparticles in the gelatin matrix and im- combination of triple helices and random coils of collagen. How- prove their interaction with gelatin, we performed a subsequent ever, when the gelation occurs in microsized spaces, random ligand exchange process, by replacing the existing oleic acid with Adv. Mater. 2023, 2310084 2310084 (3 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 2. Fabrication of interlocked microrobotic superstructures. A) Schematic of template-based microscale 3D architecture manufacturing process using heterogeneous materials (metal through channel A and polymer through channel B). B) Process overview with optical microscope images at each step. The scale bars indicate 75 μm. C) Change of plating current over time. Significant fluctuations are observed in the plating current as the plating domain changes (yellow, red, and blue in sequence). It is used for monitoring the plating process. D) Scanning electron microscope (SEM) and energy-dispersive X-ray spectroscopy (EDX) images of the interlocked microrobotic superstructure. The scale bars indicate 75 μm. E) Comparison of pure gelatin chassis made of 5% (left) and 10% (right) of gelatin concentration. The scale bar indicates 100 μm. either PAA or PEG (FTIR and zeta potential analysis in Figure and subjected to heating on a hotplate set at 50 °C. We observed S9c,d, Supporting Information). Direct light scattering (DLS) that these structures remained insoluble in water (Figure S11, analysis confirmed that PEG-IONPs and PAA-IONPs had hydro- Supporting Information). Conversely, PAA-IONP/gelatin com- dynamic diameters of 32 and 124 nm, respectively, thereby en- posites dissolved upon heating (Figure 3E and and Figures S12 suring effective nanoparticle dispersion in de-ionized (DI) water and S13, Supporting Information). We assume that the 𝛽-sheet (Figure 3C). Finally, we evaluated the magnetic hyperthermia ca- structures become disrupted by physico-chemical changes in the pabilities of these nanoparticles under an alternating magnetic gelatin solution (i.e., pH)[37] or the physical interactions between field at 510 kHz and 20 mT (Figure 3D). While PAA-IONPs ex- the nanoparticles’ ligands and the collagen fibers.[38,39] These fac- hibit 0.99 °C s−1 maximum temperature increase speed, PEG- tors would influence the cross-linking between polymer chains, IONPs display 3.5 times better performance (3.49 °C s−1 ) at a affect the protein chain mobility, and alter the secondary pro- concentration of 73 mg mL−1 . We also conducted a cell toxicity tein structure conformation. We also noted that the compos- test to evaluate the potential of these nanoparticles for biomedi- ites with PAA-IONPs displayed a significant number of agglom- cal applications. PEG-IONPs displayed high cell viability at a wide erates. In contrast, PEG-IONPs distributed homogeneously in range of concentration, while PAA-IONPs showed decreased cell gelatin (Figure 3F and and Figure S14, Supporting Information), viability as particle concentration increased (Figure S10, Support- which would marginally impact the 𝛽-sheet arrangements of the ing Information). collagen fibers,[35] thus preventing the solubility of the compos- As previously reported, pure gelatin structures exhibited re- ite, akin to pure gelatin. markable resistance to solubilization at elevated temperatures. Based on these results, we proceeded to evaluate the com- However, we wanted to investigate if this behavior was also ob- posite solubility through magnetic hyperthermia. Interestingly, served when gelatin was casted together with magnetic nanopar- PAA-IONP/gelatin composites were not dissolved when apply- ticles. To this end, we casted PEG-IONPs/gelatin composites, and ing high-frequency magnetic fields (Figure S15, Supporting In- the interlocked structures were immersed in a water container formation). As shown in Figure 3D, PAA-IONPs do not exhibit Adv. Mater. 2023, 2310084 2310084 (4 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 3. Characterization of the magnetic hyperthermia nanoparticles (A–D) and dissolution performance of their composites in gelatin matrix (E–G). A) Transmission electron microscopy (TEM) images of the synthesized IONPs. The scale bar is 200 nm. B) Magnetization hysteresis loop of IONPs. C) Graph showing the direct light scattering (DLS) spectra of PAA-IONPs and PEG-IONPs. D) Temperature change over time by applying high-frequency magnetic fields to PAA-IONPs and PEG-IONPs in water. E) Micrographs illustrating the dissolution of PAA-IONP/gelatin composite chassis in 50 °C water. The scale bar is 50 μm. F) Interlocked structures with chassis containing different types of particles. From left to right: pure gelatin chassis, gelatin with PAA-IONPs, gelatin with PEG-IONPs, and gelatin with a mixture of PAA- and PEG-IONP. Note that PEG-IONPs distribute very homogenously in gelatin, and display similar finish with that of pure gelatin. Conversely, PAA-IONPs form significant agglomerates in gelatin. The scale bar is 50 μm. G) Microscope images showing the dissolution of a chassis made of PEG-IONP/PAA-IONP/gelatin composite via magnetic hyperthermia (PAA-IONPs and PEG-IONPs were prepared in a 2:1 weight ratio.). optimal magnetic hyperthermia features unlike PEG-IONPs. To tion). Within the assessed magnetic field gradients and magnetic solve this issue, we prepared a gelatin composite containing field strengths, the superstructure achieved up to 486 μm s−1 of a mixture of 73 mg mL−1 PAA-coated and 36 mg mL−1 PEG- coated particles (PAA-IONPs:PEG-IONPs = 2:1), which enabled the chassis to solubilize upon magnetic hyperthermia within 1 Table 1. Effect of adding IONPs on dissolution of the gelatin chassis (O: h (Figure 3G). Thus, we concluded that the chassis should com- dissolved, X: not dissolved). prise PAA-IONP/PEG-IONP/gelatin composite. Table 1 summa- rizes all these results for reader guidance. Chassis compound 50 °C Magnetic water hyperthermia Pure gelatin X X 2.4. Navigation and Actuation of the Microrobotic PAA-IONP/gelatin composite O X Superstructures PEG-IONP/gelatin composite X –a) PAA-IONP/PEG-IONP/gelatin O O The locomotion of the interlocked microrobotic superstructures composite was investigated within a microfluidic chip having a channel with two sections of different widths, 400 and 85 μm. Application of (PAA-IONPs:PEG-IONPs = 2:1) a magnetic field gradient actuates the superstructures along the a) As dissolution was not observed in 50 °C water, the magnetic hyperthermia test 400 μm section (Figure 4A; and Movie S1, Supporting Informa- was not conducted. Adv. Mater. 2023, 2310084 2310084 (5 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 4. Demonstration of targeted delivery using a microrobotic superstructure through the depicted microfluidic tapered channel. A) Optical micro- scope images of the microfluidic channel (width: 400 μm) employed for the navigation experiments of six helical micromachines and gelatin chassis powered by gradient magnetic field. B) Graphs illustrating the velocity change of the robot (six helical micromachines and gelatin chassis) as a function of gradient field. Error bars represent the standard error. C) Micrographs illustrating the on-command gelatin chassis dissolution by IONPs magnetic hyperthermia. D) Optical microscope images of the small channel (width: 85 μm) used for the navigation of six helical micromachines by rotating mag- netic field. E) Graph showing the velocity change of a helical micromachine as a function of actuation frequency. Error bars represent the standard error. Scale bars indicate 100 μm. velocity at 400 mT m−1 in a constant field of 15 mT (Figure 4B). the corkscrew motion is a more efficient strategy when the size of Also, it was observed that the microrobot’s velocity was directly the micromachine becomes smaller or the medium is denser.[40] proportional to the applied magnetic field gradient, which can be Interestingly, the helical micromachines which were assembled attributed to the dependency of the magnetization with the ap- exhibited actuation and forward propulsion (Movie S3, Support- plied magnetic field. Within the larger section of the microchan- ing Information). This assembly can be avoided by further treat- nel, the magnetic gradients served to navigate the superstructure ing the surface of the helical structures with additional coatings into a target circular pocket, as shown in Figure 4C. The robot’s to promote repulsion. Nevertheless, it can result in an interest- progression was impeded at this point due to the decreased width ing feature that would be engineered with different shapes and of the subsequent channel section. At this point, the superstruc- magnetic volumes to demonstrate programmable reassembly. ture was subjected to high-frequency magnetic fields of 20 mT The disparity in propulsion speeds is further accentuated at 510 kHz for approximately an hour to dissolve the chassis when evaluated in terms of body lengths per second (Figure S17, (Figure 4C). Supporting Information). In this context, the magnetic gradi- After the release, the helical micromachines were maneuvered ent field-based propulsion of the superstructures demonstrates using rotating magnetic fields, as illustrated in Figure 4D. While a clear superiority in terms of velocity over the corkscrew motion some helical micromachines exhibited adhesion among them- induced by rotating magnetic fields in individual helical micro- selves, individual propulsion speeds were characterized as a func- machines. The faster travel of the superstructures, when scaled tion of the rotational field strength and frequency (Figure 4E; to their body length, underscores the effectiveness of using mag- and Movie S2, Supporting Information). At 15 mT, a clear step- netic field gradients for efficient navigation, particularly in sce- out frequency was observed at 45 Hz, beyond which the forward narios that require rapid transportation of the microrobotic struc- velocity decreased drastically. A similar step-out frequency was tures over larger distances. observed at 40 Hz when actuated at 10 mT. However, we could Finally, implementing the proposed delivery strategy of inter- not observe an obvious step-out frequency for helical microma- locking the multiple helical micromachines into a superstruc- chines actuated at 5 mT. Although the application of a gradient ture significantly improved targeted delivery. Through the in- field generates faster motion (Figure S16, Supporting Informa- creased magnetic volume, the superstructure enables the helical tion) compared to using a corkscrew motion, this locomotion al- microswimmers to swim at higher speeds than their step-out fre- lows for more precise control in confined spaces. Furthermore, quencyallows. Employing a superstructure as a carrier of helical Adv. Mater. 2023, 2310084 2310084 (6 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de machines enhanced the velocity by up to 4 times compared to scribe GmbH) and the slicing/hatching distances were set to 0.3 μm. The using a single helical micromachine in the gradient field (Figure laser power was set to 23% of the maximum laser power of the laser and S16, Supporting Information). the scan speed was set to 7500 μm s−1 . The strong reflection of the un- derlaying copper layer enabled the detection of the copper/photoresist in- terface, which is crucial for further fabrication steps. The microstructure had two types of channels, A and B. Channel A let the electrolyte reach 3. Conclusion the copper substrate leading to electrical contact, while channel B did not connect to the substrate. This setup allowed selective plating in channel A In this study, we demonstrated an efficient microscale tar- and filling channel B with gelatin. After light exposure and baking at 110 °C geted delivery approach using microrobotic superstructures that for 1 min. Last, the substrate was developed in AZ MIF 726 to remove the adapt their size to navigate through channels of varying di- exposed regions for 13 min to form the channels (Figure S18, Supporting mensions. These superstructures were fabricated by interlock- Information). Then, the template was immersed in water to remove the ing electroplated-iron helical micromachines with a microcasted developer. The channels of the micromold were kept in water to keep the channels wet. thermally responsive magnetopolymer chassis. Notably, this Iron Deposition: An electrolyte consisting of 250 g L−1 of FeSO4 ·7H2 O, method enables batch fabrication of superstructures without ad- 42 g L−1 of FeCl2 ·4H2 O, and 20 g L−1 of NH4 Cl mixed with 200 mL of DI ditional assembly steps. The chassis, composed of thermally re- water was used. The solution was thoroughly stirred at 250 rpm until the sponsive gelatin infused with magnetic IONPs, which function iron salts were dissolved. The pH value of the electrolyte was adjusted to a as heat transducers, allowing for the controlled dissolution of the pH of 2 by adding sulfuric acid and the electrolyte was heated to 50 °C for gelatin matrix on-command by means of high-frequency mag- iron deposition. A platinum sheet was used as an inert counter electrode and a double junction Ag/AgCl electrode was used as a reference elec- netic fields. To this end, we have used a combination of PAA- trode (Figure S19, Supporting Information). The substrate was quickly im- IONPs and PEG-IONPs to reach a tradeoff between nanoparticle mersed into the electrolyte to prevent the fast drying of the microchannels, dispersion within the gelatin matrix, while enhancing the ther- ensuring a complete infiltration of the electrolyte within the channels and mal gelatin dissolution. By remotely instructing the dissolution good electrical contact with the conductive copper substrate. The iron was of the gelatin, we enable the release of nested microhelices, facil- electrodeposited potentiostatically by means of a potentiostat/galvanostat itating navigation through narrower vessels. (PGSTAT204, Metrohm AG) by applying a constant potential of −0.95 V. The presented manufacturing method can be extended to Typical potentiostatic curves for this process showed an initial sharp spike of current (capacitive charging of the double layer), which was followed by other platable magnetic materials and other polymers or com- a steady current step. When the deposit reached the exit of the microchan- posites containing various functional nanoparticles or contrast nel, a sharp decrease of the current was detected. At this moment, the pro- agents, which could be of interest in the biomedical domain. We cess was stopped, and the sample was removed and thoroughly rinsed in envision the adoption of this approach for the realization of more water to avoid the degradation of the plated iron structures caused by the complex superstructures capable of performing intricate delivery remaining acidic electrolyte. After rinsing, the sample was set to dry at procedures within the human body. Examples include accessing room temperature. At this stage, the type-A microchannel was selectively filled with metallic iron and the microchannels, which were not connected small capillaries, ducts in various organs, aneurysms, or the sub- to the substrate, remained unfilled. arachnoid space, among others. Nanoparticle Synthesis: Magnetic iron oxide particles were synthe- sized according to a previous recipe with minor adjustments. In a typical synthesis, 530 mg iron(III) acetylacetonate (Fe(acac)3 ), 199.6 mg sodium oleate, and 1.664 g oleic acid were dispersed in a mixture of benzyl ether 4. Experimental Section (7 mL), 1-octadecene (15 mL), and tetradecene (3 mL). The dispersion was Materials: Iron(II)-sulfate heptahydrate (FeSO4 · 7H2 O), iron(II) chlo- degassed at 60 °C under vigorous stirring and then heated under constant ride tetrahydrate (FeCl2 · 4H2 O), ammonium chloride (NH4 Cl), sulfuric N2 flow by ramping the temperature at 3 °C min−1 until reaching reflux acid (H2 SO4 ), acetone, isopropyl alcohol, and gelatin from porcine skin at 294 °C for 90 min. After the reaction was completed, the slurry was (gel strength 300, type A) were purchased from Sigma-Aldrich. AZ IPS cooled at room temperature. To remove organic impurities from the as- 6090 photoresist and AZ 726 MIF developer were purchased from Micro- synthesized particles, the crude product was subsequently washed twice chemicals GmbH. Borosilicate substrate glass slides and Immersol 518f with a mixture of chloroform (25 mL) and acetone (75 mL). Subsequent (Carl Zeiss Microscopy GmbH) were purchased from VWR International. purification was conducted by redispersing the particles in chloroform Conductive Copper tape was purchased from Distrelec. All the other chem- followed by precipitation in a mixture of methanol (50 mL) and acetone icals were purchased from Sigma-Aldrich. (50 mL) twice. Template Fabrication: For the preparation of the template a recent re- Ligand Exchange of Nanoparticles with Polyacrylic Acid: We conducted port was used as a basis. As substrate for all the subsequent fabrication a ligand exchange procedure to substitute the oleic acid (hydrophobic) steps a borosilicate glass slide (30 mm diameter) was used. A conductive from the surface of the as-synthesized particles with polyacrylic acid (hy- substrate was deposited on the glass a 20 nm titanium layer was sput- drophilic) as reported in previous protocols.[41] Briefly, 80 mg of synthe- tered onto the glass slide for adhesion before depositing an 80 nm Cu sized particles were dispersed in 8 mL of THF (tetrahydrofuran). This dis- layer onto the titanium to enable the electrodeposition onto the glass sub- persion was added dropwise into 72 mL of dry THF containing 2.88 g strate. Next, a positive tone AZ IPS 6090 photoresist was spin coated on of PAA under vigorous stirring for 72 h under a N2 blanket to enable a the copper side of the glass slide for 17 seconds at 2000 rpm s−1 with an ac- complete ligan exchange. Afterward, the particles functionalized with poly- celeration of 300 rpm s−1 to ensure a resist thickness of 45 μm. The coated acrylic acid were collected by centrifugation and washed 5 times by redis- substrate was baked on a hotplate at 125 °C for 5 min by ramping the tem- persion and subsequent centrifugation in DI water. Finally, particles were perature from room temperature at 8 °C min−1 to evaporate the solvent. dispersed in 20 mL DI water, deprotonated by the addition of 5 μL of NaOH A commercial two-photon lithography system (Nanoscribe Photonic Pro- (0.5 m), washed again by centrifugation, and redispersed in DI water at the fessional GT, Nanoscribe GmbH) was used to write on the photoresist desired concentration. the microchannels. The design of the microstructure was imprinted into Ligand Exchange of Nanoparticles with Polyethylene Glycol: Within a N2 the photoresist by using a 63× objective (Carl Zeiss AG) and a refractive filled glovebox 500 mg of mPEG-phosphate was dissolved in 72 mL of dry index matching oil (Immersol 518f, Carl Zeiss Microscopy GmbH). The de- toluene inside a Teflon reactor. Subsequently 80 mg of oleic acid covered sign of the microstructure was processed by the DeScribe software (Nano- particles dispersed in 8 mL of toluene were added, before the addition of Adv. Mater. 2023, 2310084 2310084 (7 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de 8 mL of dry Ethanol under constant stirring. Finally, the Teflon reactor was Microfluidic Chip Fabrication: The microfluidic chips were fabricated sealed, sonicated for 10 min within an ultrasonic bath and the mixture was using photolithography and PDMS casting. A film mask was prepared con- allowed to react overnight at 90 °C. After reaction the PEG functionalized taining the design of a chip with the length of 16.5 mm and channel widths particles were collected via extraction with DI-water and washed 5 times ranging from 80 to 400 μm. 250 μm of SU-8 100 (Kayaku Advanced Mate- by centrifugation and redispersed in DI-water. rials) photoresist was used as a mold. Then, a 1 mm thick layer of PDMS Nanoparticle–Gelatin Infiltration: For the nanoparticle–gelatin com- was cast and subsequently cured at 80 °C for 2.5 h. Additionally, a micro- posites, 25 μL of PEG-coated nanoparticles were pipetted into an Ep- scopic glass slide was also coated with a layer of PDMS and cured with pendorf tube. The aqueous nanoparticle solution was centrifuged at the same curing parameters. Next, the PDMS chip was released from the 14 500 rpm for 10 min and the supernatant was discarded. 25 μL silicon wafer. Additionally, the PDMS coated microscopic glass was cut to of PAA-IONPs were added and mixed with the PEG-IONPs resulting the size of the PDMS chip. The PDMS-coated glass slide and the PDMS in a 25 μL nanoparticle solution with 7.3 mg mL−1 PAA-IONPs and chip were then plasma bonded and stored at 50 °C over night. After bond- 3.6 mg mL−1 PEG coated nanoparticles, leading to an aqueous solu- ing, the entry channels were punched into the chip allowing the transfer tion with 10.9 mg mL−1 nanoparticles. The nanoparticles were redis- of the superstructures into the chip. The use of a PDMS-coated substrate persed in a sonicator for 10 min and heated up to 50 °C before adding decreased the sticking of the gelatin to the chip. 2.5 mg of type A gelatin from porcine skin. The nanoparticle–gelatin Actuation: Electromagnetic systems (MFG-100 and MFG-100-I, Mag- mixture was sonicated for an hour at 50 °C before pipetting the solu- nebotix AG) were used to magnetically actuate the microrobots. For the ac- tion onto the template. For the infiltration of the nanoparticle–gelatin tuation of the iron helices rotating magnetic fields and for the actuation of solution, 5 cycles of 0.05 MPa vacuum at 65 °C were applied to en- the interlocked superstructures, magnetic gradients were generated with sure successful infiltration without the evaporation of the aqueous part field strengths between 5 and 15 mT. Prior to actuating the microrobots, of the solution. After the infiltration, the sample was stored at 5 °C over they were set free by stripping the photoresist and the sample was directly night. transferred into DI water. The microrobots were then transferred into a Magnetometry: The magnetic characteristics of the particles were as- PDMS microfluidic chip with the use of a micromanipulator stage. The sessed using vibrating sample magnetometer (VSM) measurements on microfluidic chip was then placed in the working space of the magnetic powder samples. The measurement was conducted in a magnetic field coil setups and the magnetic actuation was characterized through the op- range of up to 40 kOe. Hysteresis loop measurements were conducted at tical magnifiers. room temperature with a maximum field strength of 20 kOe. Magnetic Hyperthermia: To externally release the helical microma- Transmission Electron Microscopy (TEM): TEM images were captured chines from the gelatin chassis, a magnetic hyperthermia setup was used. using an FEI Talos F200X (Chem S/TEM) operating at 200 kV, equipped The hyperthermia setup (NAN201003 Magnetherm, nanoTherics Ltd.) with an X-FEG emitter and CETA camera (16 m pixel CMOS Camera). Spec- was placed in an incubator (Galaxy 170 S, New Brunswick) to thermo- imens for imaging were prepared by dispersing particles in a diluted so- statize the coil of the system at 37 °C during the heating experiments and lution of deionized water and depositing them onto a carbon-coated 400 ensuring that heating on the specimens was only due to hyperthermia gen- mesh TEM grid. The average diameter and standard deviation of particles erated by the nanoparticles. The capacitor mounted was 22 nF, a resonant were determined by measuring a minimum of 50 particles. frequency of 510 kHz was set on the function generator and the voltage X-Ray Diffraction (XRD): The crystalline structure of the samples was was ramped up to 20 V with a power supply. Prior to setting the output, the analyzed using X-ray diffraction with an Empyrean instrument (Malvern PDMS microfluidic chip containing the interlocked structure was placed in Panalytical) operating at room temperature. The X-ray source utilized cop- a petri dish filled with DI water preheated to 37 °C. The petri dish contain- per radiation (𝜆 = 1.5406 Å), and a PIXcel detector was employed. Sample ing the interlocked structure was placed in the center of the coil to get preparation involved drop-casting a particle dispersion in ethanol onto a maximum magnetic field exposure. After switching on the output, the cur- silicon holder. rent in the coils increased to around 14 A. Dynamic Light Scattering and 𝜁 -potential: The particle’s hydrodynamic diameter and 𝜁 -potential were determined using a Zetasizer Nano-ZS. Fourier-Transform Infrared Spectroscopy (FTIR): FTIR spectra were Supporting Information measured using a Bruker Tensor 27 spectrometer across the range of Supporting Information is available from the Wiley Online Library or from 4000–400 cm−1 with a resolution of 4 cm−1 . the author. Thermogravimetric Analysis (TGA): Thermogravimetric measure- ments were conducted using a Mettler Toledo TGA/DSC 3+ Star System. Measurements were taken at temperatures of up to 1173 K under constant Acknowledgements airflow. F.C.L. and V.G. contributed equally to this work. This project has re- Stripping: To release the interlocked structures from the substrate, the ceived funding from the European Union’s Horizon 2020 Proactive photoresist template was first stripped by placing it in a specific holder at Open program under FETPROACT-EIC-05-2019 ANGIE (No. 952152), a 45° angle, which was carefully immersed in a glass beaker filled with the Swiss National Science Foundation under Project No. 198643, acetone for 5 min. After the photoresist was clearly dissolved, the sample and ETH under Grant No. 22-2 ETH-040. M.K. acknowledged par- holder was carefully transferred to a glass beaker filled with DI water. After tial financial support by the Swiss National Science Foundation under 2 min, the sample was carefully extracted from the glass beaker and was Project No. 200021L_197017. J.P.-L. acknowledged the Agencia Estatal quickly wetted with DI water with a pipette. At this point, the sample was de Investigación (AEI) for the María de Maeztu, Project No. CEX2021- always kept wet to avoid the dehydration of the gelatin. The sample was 001202-M and the Grant PID2020-116612RB-C33 funded by MCIN/ AEI transferred into a petri dish filled with DI water. A PDMS microfluidic chip /10.13039/501100011033. J.L.-W. acknowledged his Ph.D. grant in the was immersed in the same petri dish and placed next to the sample. The framework of the project BeMAGIC funded by the European Union’s interlock structures were carefully detached from the substrate and were Horizon 2020 Research and Innovation Programme under the Marie transferred to the PDMS microfluidic chip using a microprobe while al- Skłodowska-Curie Grant Agreement No. 861145. The authors would also ways being fully immersed in DI water. Once the interlocked structure was like to thank the Scientific Center for Optical and Electron Microscopy released into the microfluidic chip, the chip was transferred to another (ScopeM) and the FIRST laboratory at ETH for their technical support, and petri dish. The microfluidic chip remained immersed in water throughout the Cleanroom Operations Team of the Binning and Rohrer Nanotechnol- all the experiments. ogy Center (BRNC) for their help and support. Further Prof. Herman of the Scanning Electron Microscopy: Scanning electron microscope images Nanoparticle Systems Engineering Lab and Prof. Pratsinis of the Particle were obtained using a Zeiss ULTRA 66 at 5 kV with a 30 aperture and EDX Technology Laboratory are acknowledged for the access to various charac- elemental mapping was done at 20 kV, 60 μm aperture, and 10 mm working terization equipment. distance. Adv. Mater. 2023, 2310084 2310084 (8 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202310084 by Readcube (Labtiva Inc.), Wiley Online Library on [11/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Conflict of Interest [14] G. Go, A. Yoo, K. T. Nguyen, M. Nan, B. A. Darmawan, S. Zheng, B. Kang, C.-S. Kim, D. Bang, S. Lee, K.-P. Kim, S. S. Kang, K. M. Shim, Prof. Bradley Nelson is a shareholder of Magnebotix AG. S. E. Kim, S. Bang, D.-H. Kim, J.-O. Park, E. Choi, Sci. Adv. 2022, 8, https://doi.org/10.1126/sciadv.abq8545. [15] O. Ergeneman, G. Chatzipirpiridis, J. Pokki, M. Marin-Suárez, G. A. Sotiriou, S. Medina-Rodriguez, J. F. F. Sanchez, A. Fernandez- Data Availability Statement Gutiérrez, S. Pane, B. J. Nelson, IEEE Trans. Biomed. Eng. 2012, 59, 3104. The data that support the findings of this study are available from the [16] C. C. Mayorga-Martinez, J. Vyskocil, F. Novotný, P. Bednar, D. Ruzek, corresponding author upon reasonable request. O. Alduhaish, M. Pumera, Appl. Mater. Today 2022, 26, 101337. [17] L. Kong, A. Ambrosi, M. Z. M. Nasir, J. Guan, M. Pumera, Adv. Funct. Mater. 2019, 29, https://doi.org/10.1002/adfm.201903872. Keywords [18] Z. Yang, L. Zhang, Adv. Intell. Syst. 2020, 2, 2000082. [19] B. J. Nelson, S. Gervasoni, P. W. Y. Chiu, L. Zhang, A. Zemmar, Proc. magnetic hyperthermia, magnetic navigation, microrobots, targeted IEEE Inst. Electr. Electron. Eng. 2022, 110, 1028. delivery, two-photon lithography [20] J. G. Lee, R. R. Raj, N. B. Day, C. W. Shields, ACS Nano 2023, 17, 14196. Received: September 28, 2023 [21] A. Aziz, S. Pane, V. Iacovacci, N. Koukourakis, J. Czarske, A. Revised: November 23, 2023 Menciassi, M. Medina-Sánchez, O. G. Schmidt, ACS Nano 2020, 14, Published online: 10865. [22] M. T. Manzari, Y. Shamay, H. Kiguchi, N. Rosen, M. Scaltriti, D. A. Heller, Nat. Rev. Mater. 2021, 6, 351. [23] D. Rosenblum, N. Joshi, W. Tao, J. M. Karp, D. Peer, Nat. Commun. [1] T.-Y. Huang, H. Gu, B. J. Nelson, Annu. Rev. Control Robot Auton. Syst. 2018, 9, 1410. 2022, 5, 279. [24] P. Fischer, A. Ghosh, Nanoscale 2011, 3, 557. [2] H. Shen, S. Cai, Z. Wang, Z. Ge, W. Yang, Mater. Des. 2023, 227, [25] M. Medina-Sánchez, V. Magdanz, M. Guix, V. M. Fomin, O. G. 111735. Schmidt, Adv. Funct. Mater. 2018, 28, 1707228. [3] J. Jiang, Z. Yang, A. Ferreira, L. Zhang, Adv. Intell. Syst. 2022, 4, [26] K. E. Peyer, L. Zhang, B. J. Nelson, Nanoscale 2013, 5, 1259. 2100279. [27] T.-Y. Huang, M. S. Sakar, A. Mao, A. J. Petruska, F. Qiu, X.-B. Chen, S. [4] X.-Z. Chen, M. Hoop, F. Mushtaq, E. Siringil, C. Hu, B. J. Nelson, S. Kennedy, D. Mooney, B. J. Nelson, Adv. Mater. 2015, 27, 6644. Pané, Appl. Mater. Today 2017, 9, 37. [28] T. Kroupa, S. Hermanová, C. C. Mayorga-Martinez, F. Novotný, Z. [5] X. Z. Chen, B. Jang, D. Ahmed, C. Hu, C. De Marco, M. Hoop, F. Sofer, M. Pumera, Langmuir 2019, 35, 10618. Mushtaq, B. J. Nelson, S. Pané, Adv. Mater. 2018, 30, https://doi.org/ [29] J. Yu, L. Zhang, IEEE ASME Trans. Mechatron. 2019, 24, 154. 10.1002/adma.201705061. [30] J. Yu, L. Yang, L. Zhang, Int. J. Rob. Res. 2018, 37, 912. [6] D. Jang, J. Jeong, H. Song, S. K. Chung, J. Micromech. Microeng. 2019, [31] C. C. J. Alcântara, S. Kim, S. Lee, B. Jang, P. Thakolkaran, J. Y. Kim, H. 29, 053002. Choi, B. J. Nelson, S. Pané, Small 2019, 15, https://doi.org/10.1002/ [7] S. Jeon, S. Kim, S. Ha, S. Lee, E. Kim, S. Y. Kim, S. H. Park, J. H. Jeon, smll.201805006. S. W. Kim, C. Moon, B. J. Nelson, J. young Kim, S. W. Yu, H. Choi, Sci. [32] C. C. J. Alcântara, F. C. Landers, S. Kim, C. De Marco, D. Ahmed, B. Robot. 2019, 4, https://doi.org/10.1126/scirobotics.aav4317. J. Nelson, S. Pané, Nat. Commun. 2020, 11, 5957. [8] H. Lee, D.-I. Kim, S.-H. Kwon, S. Park, ACS Appl. Mater. Interfaces [33] L. Zhang, J. Abbott, L. Dong, K. E. Peyer, B. E. Kratochvil, H. Zhang, 2021, 13, 19633. C. Bergeles, B. J. Nelson, Nano Lett. 2009, 9, 10. [9] S. Fusco, M. S. Sakar, S. Kennedy, C. Peters, R. Bottani, F. Starsich, A. [34] A. Sakai, Y. Murayama, K. Fujiwara, T. Fujisawa, S. Sasaki, S. Kidoaki, Mao, G. A. Sotiriou, S. Pané, S. E. Pratsinis, D. Mooney, B. J. Nelson, M. Yanagisawa, ACS Cent. Sci. 2018, 4, 477. Adv. Mater. 2014, 26, 952. [35] E. S. Gil, R. J. Spontak, S. M. Hudson, Macromol. Biosci. 2005, 5, 702. [10] M. Hu, X. Ge, X. Chen, W. Mao, X. Qian, W.-E. Yuan, Pharmaceutics [36] S. Sun, H. Zeng, J. Am. Chem. Soc. 2002, 124, 8204. 2020, 12, 665. [37] P. Y. Furlan, S. A. Scott, M. H. Peaslee, Spectrosc. Lett. 2007, 40, 475. [11] D. Liu, T. Wang, Y. Lu, Adv. Healthcare Mater. 2022, 11, https://doi. [38] Y. A. Arfat, J. Ahmed, N. Hiremath, R. Auras, A. Joseph, Food Hydro- org/10.1002/adhm.202102253. coll. 2017, 62, 191. [12] E. Kim, S. Jeon, H. K. An, M. Kianpour, S. W. Yu, [39] S. Sahraee, B. Ghanbarzadeh, J. M. Milani, H. Hamishehkar, Food J. Y. Kim, J. C. Rah, H. Choi, Sci. Adv. 2020, 6, Bioproc. Tech. 2017, 10, 1441. https://doi.org/10.1126/sciadv.abb5696. [40] K. E. Peyer, L. Zhang, B. E. Kratochvil, B. J. Nelson, Proc. IEEE [13] X. Wang, X. Z. Chen, C. C. J. Alcântara, S. Sevim, M. Hoop, A. Int. Conf. Robot Autom. 2010, 96, https://doi.org/10.1016/j.colsurfb. Terzopoulou, C. de Marco, C. Hu, A. J. de Mello, P. Falcaro, S. 2020.111385. Furukawa, B. J. Nelson, J. Puigmartí-Luis, S. Pané, Adv. Mater. 2019, [41] L. Nie, P. Chang, C. Ji, F. Zhang, Q. Zhou, M. Sun, Y. Sun, C. Politis, 31, https://doi.org/10.1002/adma.201970192. A. Shavandi, Colloids Surf., B 2021, 197, 111385. Adv. Mater. 2023, 2310084 2310084 (9 of 9) © 2023 The Authors. Advanced Materials published by Wiley-VCH GmbH