Reusable and Long‐Lasting Active Microcleaners for Heterogeneous Water Remediation

Self‐powered micromachines are promising tools for future environmental remediation technology. Waste‐water treatment and water reuse is an essential part of environmental sustainability. Herein, we present reusable Fe/Pt multi‐functional active microcleaners that are capable of degrading organic pollutants (malachite green and 4‐nitrophenol) by generated hydroxyl radicals via a Fenton‐like reaction. Various different properties of microcleaners, such as the effect of their size, short‐term storage, long‐term storage, reusability, continuous swimming capability, surface composition, and mechanical properties, are studied. It is found that these microcleaners can continuously swim for more than 24 hours and can be stored more than 5 weeks during multiple cleaning cycles. The produced microcleaners can also be reused, which reduces the cost of the process. During the reuse cycles the outer iron surface of the Fe/Pt microcleaners generates the in‐situ, heterogeneous Fenton catalyst and releases a low concentration of iron into the treated water, while the mechanical properties also appear to be improved due to both its surface composition and structural changes. The microcleaners are characterized by scanning electron microscopy (SEM), X‐ray photoelectron spectroscopy (XPS), nanoindentation, and finite‐element modeling (FEM).

During the reaction chain, Fe 2+ oxidizes to Fe 3+ and Fe 2+ is regenerated back from Fe 3+ (Equations ( 1) and ( 2) ). One of the main disadvantages of the classical Fenton reaction is that at

Reusable and Long-Lasting Active Microcleaners for Heterogeneous Water Remediation
Jemish Parmar , Diana Vilela , Eva Pellicer , Daniel Esqué-de los Ojos , Jordi Sort , and Samuel Sánchez * Self-powered micromachines are promising tools for future environmental remediation technology. Waste-water treatment and water reuse is an essential part of environmental sustainability. Herein, we present reusable Fe/Pt multi-functional active microcleaners that are capable of degrading organic pollutants (malachite green and 4-nitrophenol) by generated hydroxyl radicals via a Fenton-like reaction. Various different properties of microcleaners, such as the effect of their size, short-term storage, long-term storage, reusability, continuous swimming capability, surface composition, and mechanical properties, are studied. It is found that these microcleaners can continuously swim for more than 24 hours and can be stored more than 5 weeks during multiple cleaning cycles. The produced microcleaners can also be reused, which reduces the cost of the process. During the reuse cycles the outer iron surface of the Fe/Pt microcleaners generates the in-situ, heterogeneous Fenton catalyst and releases a low concentration of iron into the treated water, while the mechanical properties also appear to be improved due to both its surface composition and structural changes. The microcleaners are characterized by scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), nanoindentation, and fi nite-element modeling (FEM).

Introduction
It is well known that organic and industrial wastewater poses a serious threat to the environment and, if the wastewater is released untreated, it can damage aquatic life and be harmful to human health. [1][2][3] Since the last century, signifi cant efforts have the end of the treatment the iron ions need to be removed from the solution. Iron salt removal requires a high amount of chemicals for precipitation and produces a large amount of sludge. Further sludge removal is an expensive process and requires a lot of energy. In addition, the non-reusability of iron salt as a catalyst and the energy required for mixing result in extra costs for the treatment. To overcome the disadvantages of this classical homogeneous Fenton reaction, signifi cant efforts have been made to develop heterogeneous Fenton catalysts. [ 43 ] To develop more practical microcleaners and to overcome the limitation of the Fenton reaction, we developed microcleaners that can be reused several times for batch cleaning, swim continuously for hours, and be stored for weeks for later use, while at the same time also minimizing the iron release into the solution by generating an in-situ heterogeneous catalyst on the iron surface. The effect of different microcleaner sizes on the organic dye degradation rate, the chemical composition after the cleaning cycles, and mechanical properties of the microcleaners were studied to understand the system thoroughly. We also extended the applicability of microcleaners to another model of organic contaminant, i.e., 4-nitrophenol, demonstrating their versatile remediation functionalities.

Size Effect of Microcleaners on Dye Degradation and Reusability
Pre-strained nanomembranes of iron and platinum were sequentially evaporated by e-beam on photoresist in squared patterns of different sizes. The nanomembranes were selectively lifted off from the glass substrate and rolled up into micro tubular structures (Movie S1, Supporting Information), which led to the formation of the microcleaners as explained in the Experimental Section. The effect of the size of the Fe/Pt microcleaners on the degradation rate was studied using three different sizes. For all sizes, an equal area of previously designed photoresist patterns was deposited to keep the amount of catalytic material the same in each experiment even though the total number of tubes was different. Three sizes of microcleaners (200 µm, 300 µm, and 500 µm long) were fabricated by rolling up Fe/Pt nanomembranes ( Figure 1 A,B) and used for the degradation of the dye. The experimental parameters for dye degradation are presented in the Experimental Section. The classical Fenton reaction is highly oxidative in nature because of the production of hydroxyl radicals during the reaction of Fe 2+ ions with hydrogen peroxide, which are capable of completely oxidizing organic molecules. Degradation of a model pollutant dye, malachite green, via a classical Fenton reaction has been studied in detail using iron salt (ferrous sulfate) as the source of Fe 2+ ions. [ 44 ] However, the external mixing needed for the degradation reaction and the removal of the sludge produced by the precipitation of Fe 2+ ions after completion of the reaction make the process expensive. Under similar experimental conditions, the active microcleaners showed similar results but without the need for external mixing and a lower amount of iron was released from the surface into the treated water. The microcleaners thus act as multipurpose agents, whereby the platinum layers inside the microcleaner act as the engine to decompose H 2 O 2 into O 2 and H 2 O. The oxygen bubble trail produces a thrust that propels the microcleaner, which additionally provides micro-mixing and enhances mass transfer. [ 45,46,55 ] The iron layer on the outside of the microcleaners reacts with the H 2 O 2 to produce hydroxyl radicals via a Fenton-like reaction that degrades the organic compound. The pH was adjusted to 2.5 using sulfuric acid (the reported optimum pH is between 2-3 for the Fenton reaction catalyzed by zero valent metallic iron [ 47,48 ] ) and the initial concentration of malachite green was kept to 50 µg mL −1 in all the experiments. During the dye degradation experiments, the dye concentration was periodically measured by UV-vis spectrometer and the microcleaners were left swimming in the contaminated dye solution until a steady-state degradation was observed after 60 minutes. Figure 1 C shows the degradation curves of 200 µm, 300 µm, and 500 µm microcleaners and control experiments without microcleaners. The microcleaners degraded more than 80% of the malachite green in 60 minutes; furthermore, complete degradation was achieved over longer times (not shown). After 60 minutes of degradation, the malachite green degradation was measured for all three sizes of microcleaners. A one-way analysis of variance (ANOVA) was calculated for all measured data points for the different sizes of microcleaners to verify the statistically signifi cant difference between them. No signifi cant difference was found in the amount of dye degraded by the three different sizes of microcleaners at the P = 0.9850 ( n = 5) level. The degradation of malachite green dye is due to the oxidation reaction facilitated by hydroxyl radicals produced while the iron containing microcleaners are swimming in wastewater containing hydrogen peroxide. Hydroxyl radicals have very strong oxidation potential (2.8 V), just below the oxidation potential of fl uorine (3V); therefore, if enough time is given, hydroxyl radicals can mineralize organic molecules into carbon dioxide without leaving any toxic byproducts. Hydroxyl radicals oxidize malachite green into a fi nal byproduct of oxalic acid, before mineralizing into carbon dioxide [ 56 ] .
Clearly, the Fe/Pt microcleaners showed a higher degradation rate compared to the control experiments without microcleaners as shown in Figure 1 C. Fe/Pt microcleaners have already been shown to outperform various controls with respect to the dye degradation rate, such as (a) Fe tubes only, (b) noniron containing motors, i.e, Ti/Pt and (c) immobilized Fe/Pt microcleaners. [ 26 ] Figure 1 D shows the dye degradation by a single microcleaner considering that all microcleaners present in the solution contribute equally to the total degradation. The Figure reveals that a larger microcleaner of 500 µm is more effective than a 300 µm or a 200 µm microcleaner. Provided that the amount of total rolled up catalytic material present in the solution is equal, that is, 0.64 cm 2 in all cases, the differences in size of the microcleaners do not give an added advantage and have a limited effect on the degradation of dye in the studied size range. The total amount of catalytic material plays a more important role than the size of the microcleaners. Different experimental parameters-such as the effect of H 2 O 2 concentration and the addition of H 2 SO 4 -were further studied by using 500 µm micro-cleaners fabricated in a new batch. Figure 1 E shows the percentage of degradation of 50 µg ml -1 malachite green by micro-cleaners in 60 minutes at the different concentrations of H 2 O 2 . Above 15% H 2 O 2 the degradation percentage does not increase signifi cantly, reaching a plateau. Figure 1 F shows the absorbance spectrum of malachite green after 60 minutes of degradation by microcleaners with and without the addition of sulfuric acid for 2.5 pH maintenance. Interestingly, the effect of sulfuric acid addition on the degradation percentage is almost negligible, meaning that in future applications, addition of acidic media is no longer required for the degradation of organics using microcleaners.
A reusable catalyst is important for the cost effectiveness of the Fenton-based advanced oxidative processes. The reusability performance of the microcleaners was studied, as shown in Figure 2 A. All three sizes of microcleaners were tested for reusability to verify if the performance remained comparable in later cycles. In each cleaning cycle, the microcleaners were fi rst left swimming in the malachite-dye-contaminated water for dye degradation. After this, the microcleaners were collected using a permanent magnet, cleaned with ultrapure water three times, and then reused in subsequent cleaning cycles. The time interval between two cycles was chosen to be incremental in order to capture both the short-term and long-term changes and the effect on the degradation rate. The fi rst fi ve cleaning cycles were performed consecutively from 1 to 5 hours, then the next cycles were performed after 18 hours and 24 hours of storage in sodium dodecyl sulfate (SDS) containing water without hydrogen peroxide to study the changes after shortterm storage. The following cycles were performed after a 1-week interval between each cycle.
Degradation of the dye from 1 to 5 hours, when microcleaners were reused continuously without storing them, and at 18 hours and 24 hours after short-term storage was between 68-86%, as shown in Figure 2 B. After long-term storage (one to fi ve weeks), the degradation was slightly reduced to 56-67%, as presented in Figure 2 C. The percentages of degradation were very similar for the different sizes of microcleaners for all dye degradation cycles after both short-term and long-term use, which shows that the size of the microcleaners was also of no infl uence in terms of reusability.
Previously, iron layers have been used for magnetic steering and guiding purposes. [ 49 ] Here, we exploited the ferromagnetic nature of the Fe layer as an added functionality to recover the microcleaners, along with their Fenton-reaction capability. The microcleaners can be magnetically recovered and reused several times without signifi cant changes in the dye-degradation effi ciency, even after weeks of storage.
After each reusability cycle, the swimming behavior of the microcleaners was observed under an optical microscope to assess the motility and bubble-production activity. We observed that from second cycle onwards the microcleaners were producing bubbles more vigorously because of the selfcleaning and activation of platinum surface in the fi rst cycle. The microcleaners remained active after 5 weeks (including both short-term and long-term intermediate storage, see Movie S5, Supporting Information). The structural integrity of the microcleaners was also observed to be very good during the initial cycles but in the later cycles, some of the longer microcleaners were broken into two pieces or broken layers became visible, whereas some shorter microcleaners broke into even smaller pieces without any tubular geometry. The damage in the structure could be due to i) multiple exposure of the microcleaners to the external magnetic fi eld of a strong neodymium-iron-boron magnet during the recovery process after every cycle, and ii) internal pressure of bubbles generated while swimming. Damage in the structural integrity could be one of the reasons for the observed decrease of dye degradation percentage in the later cycles after long-term storage (Figure 2 C).
A separate continuous swimming experiment was carried out to understand if it is possible to use microcleaners for continuous longer swimming applications or many batch-wise shorter cleaning cycles. All three sizes of microcleaners were left swimming in H 2 O 2 (15% v/v) solution for 24 hours and swimming was monitored periodically under the microscope (Movie S2 to S4, Supporting Information). Figure 3 shows that all 200-µm, 300-µm, and 500-µm microcleaners were swimming even after 24 hours of continuous motion. Thus, it is indeed possible to use them for long-term swimming activities. Although some microcleaners were broken into smaller pieces after a few hours of swimming, they were still active. Changes in the diameter were also observed after a few hours of swimming, as seen in Figure 3 . Namely, a decrease in the diameter was visible for longer microcleaners in the images taken after 24 hours. The opposite effect was observed for the 200-µm microcleaners; some of them had opened up and broken into pieces. This difference is due to the presence of fewer windings in the microcleaners fabricated from the smaller photoresist patterns. As the same thickness of Fe/Pt nanomembranes were rolled up from different sizes of photoresist patterns, similar diameters between 40 to 60 µm (and thus, different number of The ferromagnetic microcleaners were collected with an external magnet and the treated water was changed with Millipore water to wash the surface and the beaker itself. After cleaning the surfaces of the microcleaners, new dye-contaminated water was added for the next degradation cycle. B) Reusability performance of different sizes of microcleaners for 5 consecutive degradations over periods from 1 to 5 hours and at 18 hours and 24 hours after short-term storage. One degradation cycle involved 60 minutes of swimming of the microcleaners in polluted water. C) Reusability performance of microcleaners in each cycle after 1 to 5 weeks of storage.
windings in the rolled-up tubular microcleaner structure) were achieved.

Heterogeneous Catalytic Shift in the Fenton Reaction and Surface Characterization
It is widely accepted that the zero-valent-iron-mediated Fenton reaction is mainly related to the ferrous ions generated from the iron surface in acidic pH. Fe 2+ ions that have leached from the surface into the solution play an important role in the reaction kinetics, whereby Fe 2+ is oxidized into Fe 3+ ions (Equation ( 1) ). The regeneration rate of Fe 2+ ions from Fe 3+ (Equation ( 2) ) is the rate-limiting factor for the classical Fenton reaction and the presence of a metallic surface is believed to help the reduction of the Fe 3+ ions to Fe 2+ , thus maintaining the Fenton reaction rate. [ 48 ] The iron released from the surface of the microcleaners in the solution was measured by inductively coupled plasma optical emission spectroscopy (ICP-OES). Measurements were performed after 60 minutes of degradation cycle for up to 8 cycles. The measured iron concentrations for the 200-µm, 300-µm, and 500-µm size microcleaners after the fi rst cleaning cycle were around 2.10, 2.15, and 2.20 µg mL −1 , respectively. The concentration of iron in the solution after the fi rst cycle was thus about the same for all sizes ( Figure 4 A), which further proves that the initial dye degradation rate for the different size motors was similar (Figure 1 C). In the subsequent cycles, the concentration dropped sharply, and remained low in the subsequent cycles, as shown in Figure 4 A.
The initial ferrous ion concentration in the reaction mixture greatly affects the kinetics of the Fenton reaction. As reported by Hameed et al., an iron concentration above 2 µg mL −1 is suffi cient to carry out the classical homogeneous Fenton degradation of malachite green. [ 44 ] However, if the Fe 2+ concentration in the solution is below 1 µg mL −1 , the malachite green degradation rate will not be higher than the rate observed in the control experiment without Fe 2+ . The dye degradation in the fi rst cycle can be attributed to the released iron from the surface of the microcleaners but from the second cycle onwards, the iron concentration was below 1 µg mL −1 . In spite of having an iron concentration that was below 1 µg mL −1 , the percentage of degradation only changed marginally. This result suggests a shift in the reaction pathway towards a heterogeneous Fenton reaction. This implies the formation of an in-situ heterogeneous Fenton catalyst on the surface of the microcleaners to achieve a dye degradation effi cacy as high as in the fi rst cycle. Also, the motion of the microcleaners can keep on regenerating the active surface and thus increase the mass transfer to help maintaining the percentage of dye degradation.
To study the surface changes that occurred after the Fenton reaction, the microcleaners were analyzed by X-ray photoelectron spectroscopy (XPS) using a PHI 5500 multitechnique system spectrometer, equipped with a monochromatic X-ray source. XPS was carried out on the microcleaners before the Fenton reaction, after Fenton reaction for 5 hours, and after 5 weeks of storage. The microcleaners were washed with water and then dried in an ethanol-CO 2 critical point dryer before measurements (to dry them without damaging the structure). Critical point drying is necessary to avoid mechanical stresses that are generated when the surface tension changes when the solvent on and around the microcleaners is drying.
Fe is mostly present in an oxidized form on the outermost surface already before the Fenton reaction takes place, as evidenced by the existence of a Fe 2p doublet located at 709.8 and 723.9 eV, which can be assigned to Fe 2+ (Figure 4 C). [ 50,51 ] It is plausible that the in situ generation of a Fe x O y heterogeneous catalyst [ 43 ] at the surface of the microcleaners reacts with hydrogen peroxide to yield a reactive oxidative species in the Fenton-like reaction after fi rst use. In fact, the Fe 2p doublet is slightly shifted toward higher binding energies after 5 weeks of storage, indicating the presence of Fe 3+ . According to the literature, the peak positions of Fe shift towards higher binding energies as the oxidation state of Fe increases. [ 52 ] Although the difference in binding energy between the Fe 2+ and Fe 3+ oxidation states is very small (therefore, it is diffi cult to determine the relative amount of Fe 2+ and Fe 3+ in the microcleaners), it is clear that the surface becomes more oxidized as time passes. It should be noted that the shoulder observed at around 706 eV both before and after 5 hours of Fenton reaction, which can be attributed to metallic Fe (2p 3/2 ), [ 53 ] weakens after 5 weeks. Hence, a complex mixture of iron oxides (FeOOH, Fe 3 O 4 , or Fe 2 O 3 ) is probably present at the surface of the microcleaners after 5 weeks. Also, a slight shift in the Pt 4f doublet is observed after 5 weeks of Fenton reaction (Figure 4 B). This might indicate oxidation of metallic Pt, but to a much lesser extent than Fe owing to the noble nature of Pt. Regarding the O 1s corelevel spectra, a complex, broad signal with several maxima is observed (Figure 4 D). After 5 hours of Fenton reaction the contribution from lattice O 2− (529 eV) relatively increases, indicating again that the surface is more oxidized. Likewise, the peak at 530.7 eV has been attributed to non-stoichiometric oxides in the surface region (oxygen defi ciencies). [ 54 ] After 5 weeks, the O 1s signal was dominated by the contributions from hydroxyl groups. Moreover, the Fe/Pt ratio markedly Adv. Funct. Mater. 2016, 26, 4152-4161 www.afm-journal.de www.MaterialsViews.com diminished after Fenton reaction: 1.51 before Fenton; 1.37 after Fenton for 1 h; 0.90 after Fenton for 5 weeks, indicating that Fe undergoes a leaching process, which is in agreement with the ICP analyses.

Mechanical Behavior of Fe/Pt Microcleaners
In order to assess the mechanical robustness and integrity of the microcleaners, nanoindentation experiments were performed on the rolled tubular microcleaners obtained from the 500 µm × 500 µm Fe/Pt fi lms. Experiments were carried out i) before the Fenton reaction, ii) after 5 hours of Fenton reaction, and iii) after 5 weeks of storage. Figure 5 A (left panel) shows the applied load ( P ) versus penetration depth ( h ) curve of a microcleaner before the Fenton reaction (i.e., an unused microcleaner). The test revealed a smooth loading behavior up to a load of about 0.1 mN, where a pronounced pop-in (i.e., sudden displacement burst) could be observed. This displacement was associated with a cracking event of the material, which could further be verifi ed through optical microscopy. The center panel of Figure 5 A shows the image of the tubular microcleaner before indentation, whereas the right panel shows the same microcleaner after indentation. The arrows in the right panel indicate a layer of microcleaner that was chipped away during indentation and, most likely, corresponds to the cracking event shown in the left panel of Figure 5 A. All other investigated microcleaners showed a similar behavior before Fenton reaction, accompanied by a certain barreling of the microcleaners.
A representative nanoindentation curve of the microcleaners after 5 hours of Fenton reaction is shown in Figure 5 B (left panel). The maximum penetration depth attained after the Fenton reaction is smaller than before Fenton. Namely, h decreased from around 35 µm (before Fenton) to around 23 µm (after 5 hours), respectively. This means that the Fenton reaction induced an increase in strength of the microcleaners. Cracking events and exfoliation of the microcleaners were also seen to take place during the indentation tests performed after Fenton reaction, although only at loads close to 0.2 mN (see Figure 5 B (center and right panel)). In summary, before Fenton reaction the microcleaners appear to be more ductile, with a higher attained penetration depth than after Fenton reaction for a given value of maximum applied load (compare Figure 5 A and  B). Both, before and after 5 hours of Fenton reaction, indentation tends to cause a certain barreling of the tubes (particularly before Fenton reaction), which in the end induces cracking and exfoliation of the outer shells of the tubes. As aforementioned, after 5 hours of Fenton reaction the microcleaners appear to be mechanically stiffer mainly because: i) tightening up, reducing the diameter of the microcleaner and increasing the number of layers (i.e., their thickness) and ii) the formation of iron oxides on the outer surface of the microcleaners as seen in the XPS analysis in Figure 4 C. Figure 5 C shows the results of nanoindentation on a microcleaner after 5 weeks in storage. In this case, the penetration Adv. Funct. Mater. 2016, 26, 4152-4161 www.afm-journal.de www.MaterialsViews.com depth attained for an applied load of 0.2 mN was around 10 µm and no cracking events were observed for this maximum applied load. In order to assess whether exfoliation of the microcleaners took place at higher loads, we also performed nanoindentation experiments with P Max = 1 mN. As can be observed in Figure 5 C (left panel), in this case a clear cracking event occurred around P = 0.6 mN. This critical load for cracking was therefore higher than the ones observed in Figure 5 A and B, suggesting an increase in the mechanical resistance of the microcleaners with usage. Typical optical microscopy images of these tubes before and after indentation with P Max = 1 mN are shown in Figures 5 C (center and right panel, respectively). Table 1 shows the energy analyses performed during indentation of the microcleaners for the three investigated conditions. Remarkably, the elastic recovery (i.e., the ratio between the elastic energy, U el , and the total energy, U tot ) after 5 weeks of storage was clearly larger than before Fenton or after 5 hours of usage. Hence, from a mechanical point of view, the tubes were stronger after Fenton than before, as fracturing was clearly delayed and the elastic recovery was enhanced by more than a factor of 2 with respect to the as-prepared microcleaners before Fenton.
Additionally, nanoindentation fi nite-element simulations were performed using commercial software (ABAQUS) in order to shed further light on the mechanical performance of the microcleaners. The chosen geometry for the simulations was a cylinder with a wall-to-diameter aspect ratio similar to the investigated microcleaners before and after 5 hours of Fenton reaction. The mesh used during the simulations consisted of Adv. Funct. Mater. 2016, 26, 4152-4161 www.afm-journal.de www.MaterialsViews.com Figure 5. A-C) Representative load ( P ) -displacement ( h ) curves and optical microscopy images of microcleaners before indentation (center) and after indentation (right) corresponding to the microcleaners before Fenton (A), after 5 hours of Fenton (B), and after 5 weeks of storage (C). The arrows indicate chipped-off layers and cracks of the microcleaners that occurred during indentation and, most likely, are associated to the cracking event shown in the respective load-displacement curve. fully integrated brick-shape elements, the Berkovich indenter was considered to be a perfectly rigid body, and the cylinders were perfectly elastic with a Young's modulus equal to 200 GPa. The boundary conditions were chosen such as to prevent the vertical displacement of the cylinder during indentation. The von Mises yield criterion was used to study the differences in the mechanical performance of the microcleaners before and after 5 hours of Fenton reaction. The diameter of the microcleaners decreased after the reaction ( Figure 6 A,C) due to the tightening of the layers, likely because the pressure pulses generated during bubble development and release promoted the release of residual strain from the layers. The simulations revealed that the tube after Fenton reaction (Figure 6 D) accumulated a higher stress directly beneath the indenter tip for a given applied load than the tube before Fenton reaction (Figure 6 B), indicating that it is mechanically harder. Concomitantly, for a certain applied load, the overall deformation of the tube before Fenton reaction was higher than in the simulated tube after Fenton. The results of this simple simulation (which does not take into account the multiwalled structure of the microtubes) agreed qualitatively well with the experimental observations.

Degradation of Phenolic Compound
In order to demonstrate the remediation capabilities of microcleaners to other organic pollutants, we performed a degradation experiment for a phenolic compound (4-nitrophenol) using 500 µm microcleaners. 4-nitrophenol is one of the most common organic pollutant molecules present in industrial wastewater. Degradation of 4-nitrophenol is challenging using bacteria, yet hydroxyl radicals are capable of completely mineralizing it into carbon dioxide. [57][58][59] Figure 7 shows that using H 2 O 2 as oxidant alone cannot degrade 4-nitrophenol, while micro-cleaners can degrade around 30% in 60 minutes. The difference in the percentage of degradation for malachite green and 4-nitrophenol is due to the different reaction kinetics of hydroxyl radicals for different organic molecules. Microcleaners degraded ≈18 µg of 4-nitrophenol in 10 minutes and ≈41 µg in 60 minutes from 3 mL of contaminated water containing 150 µg of initial amount (50µg mL -1 ). A longer duration is required to achieve complete degradation, but the addition of larger amounts of microcleaners could achieve faster oxidation and even total degradation.

Conclusions
We demonstrated reusable, self-propelled Fe/Pt microcleaners that can carry out a Fenton-like reaction with high activity and without the need for external mixing. We found that the variation in the length of microcleaners does not affect the performance if the amount of catalytic material used is kept constant.
The reusability results showed that the microcleaners can be recovered using magnets and reused for multiple times within a short duration of less than a week without any decrease in their organic-degradation performance. Even longer term storage for several weeks is possible without sacrifi cing much of the activity. The microcleaners can also be used for continuous swimming applications for at least 24 hours. Although the iron released into the treated water from the second cycle onwards was much less compared to that from the fi rst cycle, the activity of the microcleaners remained constant. We observed that the surface of the microcleaners oxidized to produce in situ iron oxides that act as a heterogeneous catalyst.  and malachite green proved the possibility of using microcleaners for wide range of organic pollutants. The experiments presented here evidence the long-term reusability of very active microcleaners, which will be benefi cial towards lowering the cost of the water treatment using this advanced technology. Further experiments should be driven towards the remediation of other pollutants in real wastewater samples and in confi ned pipes or places diffi cult to reach by traditional methods.

Experimental Section
Fabrication of the Microcleaners : Microcleaners were fabricated by rolling up nanomembranes of iron and platinum metal deposited on square patterns of photoresist. Positive photoresist patterns (200 µm, 300 µm, and 500 µm) were developed using standard photolithography techniques. For this a positive photoresist (ARP 3510) was spin-coated (3500 rpm for 35 s) on previously cleaned glass wafers (18 mm × 18 mm) to make a layer with uniform thickness (2.4 µm) and exposed to UV light under a chromium mask with the respective sizes of the patterns confi ned in a 1 cm 2 area by a mask aligner. The photolithographic patterns on the glass substrates were developed (using 1:1 water/ AR 300) and dried by blowing nitrogen before depositing the metal nanomembranes. A custom-built e-beam evaporator was used for the deposition. Two layers of iron (100 nm) were evaporated at different deposition rates (at 0.30 nm s −1 and 0.06 nm s −1 respectively); a third layer, this time of platinum (5 nm), was evaporated (at 0.02 nm s −1 ). All three layers were deposited at a glancing angle (65°), which led to a non-deposited window in each pattern. The photoresist wall adjacent to the non-deposited window remained exposed which was required for the controlled directional rolling of the nanomembranes. A mixture of dimethyl sulfoxide (DMSO) and acetone (1:1) was used to selectively etch the photoresist from the exposed wall. The nanomembranes were rolled up from the side of the exposed wall to the unexposed wall in the shape of tubular microcleaners.
Size Effect, Reusability, and 4-Nitrophenol Degradation Experiments : Three different sizes of microcleaners (200 µm, 300 µm, and 500 µm long with a diameter ranging from 40-60 µm) were fabricated from the nanomembranes that were deposited on the photoresist patterns confi ned in the 1 cm 2 area on the glass substrate. The number of microcleaners rolled up from a constant amount of catalytic material present in a 0.64 cm 2 area (including all square patterns) was different for the different pattern sizes (around 1600, 729, and 256, respectively, for the different sizes in increasing order). After being rolled up, the microcleaners were fi rst transferred into sodium dodecyl sulfate (SDS) water (0.5% w/v) and then used for the degradation experiments, carried out in a beaker containing a total of 3 mL of polluted water consisting of malachite green (50 µg mL −1 ), hydrogen peroxide (15% v/v), and SDS (0.5% w/v) at an acidic pH (2.5). The dye concentration was measured using a spectrophotometer (Specord 250, Analytical Jena) at 0, 10, 30, and 60 minutes during the experiments to study the size effect. New batches of 500 µm micro-cleaners were fabricated and used to study the effect of hydrogen peroxide concentration (5%, 10%, 15%, 20% and 25%) on the degradation of malachite green in 60 minutes. Degradation of 4-nitrophenol (50 µg ml -1 ) was carried out using 500 µm micro-cleaners in the same experimental condition used for malachite green degradation.
A different batch of microcleaners of all sizes was fabricated, using the same parameters that were used for the size-effect experiments, to study the reusability. All three sizes of microcleaners were reused both after short-and long-term storage. The short-term experiments were carried out at varying time intervals; fi rst fi ve cycles were carried out at 1 to 5 hours continuously changing polluted water after the end of the 60 minutes of a degradation cycle. After the end of each cycle, the microcleaners were confi ned in a corner of the beaker using a strong neodymium-iron-boron hard magnet and the treated water was replaced with pure water (Millipore water) to clean the surfaces of the microcleaners, the cleaning step was repeated twice and then a new batch of polluted water solution was added for the next cycle. The composition of the polluted water was kept constant as in the size-effect experiments. After 5 hours, the microcleaners were cleaned and stored in SDS water (0.5% w/v) before using in the next cycles at 18 hours and 24 hours from the fi rst cycle. In a similar way, long-term storage experiments were carried out using the same microcleaners after 1 week of intermediate storage between two cycles, and up to 5 weeks from the fi rst cycle. The dye concentration after each cycle was measured using a UV-vis spectrophotometer. After each cycle, the treated water was collected and further analyzed by ion coupled plasma optical emission spectroscopy (ICP-OES) to measure the iron concentration that had leached out from the surface of the microcleaners.
Continuous Swimming and Video Recording : An upright microscope (Leica DFC3000G camera) was used to record the videos of the rolling-up of different sizes of microcleaners whereas an inverted microscope (Leica DMI300B) was used to study the swimming behavior of the microcleaners after each cleaning cycle. A custom-designed 3D-printed microscope stage was fabricated to record the swimming of the microcleaners directly in the beaker where the degradation experiment was going on. During the continuous swimming experiment, the microcleaners were observed under the inverted microscope at 1, 5, and 24 hours.
Surface Characterization : X-ray photoelectron spectroscopy (XPS) analyses were carried out on a PHI 5500 Multitechnique System (from Physical Electronics) spectrometer, equipped with a monochromatic X-ray source (Kα Al line with an energy of 1486.6 eV and power of 350 W), placed perpendicular to the analyzer axis and calibrated using the 3d 5/2 line of Ag with a full width at half maximum (FWHM) of 0.8 eV. The analyzed area was a 0.8 mm diameter disk surface for each sample. Any charging effects were corrected for by referencing the binding energies to that of the adventitious C 1s line at 284.5 eV.
Mechanical Properties : The microcleaners were dried using an ethanol-CO 2 critical point dryer before doing the nano-indentation experiments. Typical load-displacement measurements were conducted on the microcleaners before Fenton reaction, after 5 h of Fenton reaction and after 5 weeks. For the sake of simplicity, microcleaners obtained from the 500 µm × 500 µm Fe/Pt layers were selected for the mechanical tests. These experiments were performed in a load-control mode, using a UMIS instrument from Fischer-Cripps Laboratories equipped with a Berkovich pyramidal-shaped diamond tip. The maximum applied load values ranged between 0.2 mN and 1 mN. To ensure statistically meaningful results, at least 10 indentations were performed for each type of microcleaners and the representative average behavior is reported. The elastic ( U el ) and plastic ( U pl ) energies during indentation were assessed from the areas enclosed between the unloading segment and the displacement axis ( U el ), and between the loading and unloading segments ( U pl ). The total indentation energy is then U tot = U el + U pl and corresponds to the area enclosed between the loading segment and the displacement axis. The ratio U el / U tot is related to the elastic recovery of the tubes after having been indented.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.