Phototaxis is commonly observed in motile photosynthetic microorganisms. For example, green algae are capable of swimming towards a light source (positive phototaxis) to receive more energy for photosynthesis, or away from a light source (negative phototaxis) to avoid radiation damage or to hide from predators. Recently, with the aim of applying nanoscale machinery to biomedical applications, various inorganic nanomotors based on different propulsion mechanisms have been demonstrated. The only method to control the direction of motion of these self-propelled micro/nanomotors is to incorporate a ferromagnetic material into their structure and use an external magnetic field for steering. Here, we show an artificial microswimmer that can sense and orient to the illumination direction of an external light source. Our microswimmer is a Janus nanotree containing a nanostructured photocathode and photoanode at opposite ends that release cations and anions, respectively, propelling the microswimmer by self-electrophoresis. Using chemical modifications, we can control the zeta potential of the photoanode and program the microswimmer to exhibit either positive or negative phototaxis. Finally, we show that a school of microswimmers mimics the collective phototactic behaviour of green algae in solution.
Phototaxis is the ability of an organism to sense the direction of light and to displace itself towards or away from it1,2,3,4. Depending on the illumination intensity, nutritional state and photosynthetic activity, the microorganism can swim towards the light source (positive phototaxis) or away from it (negative phototaxis). This phototactic ability benefits microorganisms by allowing them to achieve more regulated light exposure or by offering them a greater probability of finding phototrophic organisms for food. Autonomous locomotion is also found in some synthetic Janus nanowires5,6,7,8. These Janus nanowires act as nanoscale engines and generate a propulsive force from surrounding energy sources, such as chemical fuels6,9,10,11, light12,13,14, an acoustic field15, magnetic field16 or electric field17. Because of their simplicity and intrinsically small size (usually a few hundred nanometres to a few micrometres), these micro/nanomotors are considered an ideal component to power nanorobotic systems for targeted drug delivery9,18 and non-invasive microsurgery16,19. However, because the unguided active nanowires only show enhanced diffusion20, direction control is necessary for many desired applications. For active drug delivery, the nanomotor should be able to orient towards specific chemical signals, such as those demonstrated in chemotaxis21,22,23,24,25 and pH taxis26. For certain applications such as non-invasive surgery and nanofabrication, it is desirable to remotely control the migration of the micro/nanomotor using an external field. To achieve this, ferromagnetic segments (such as nickel) have been incorporated into nanomotors27,28,29; these enable the nanomotors to align in a desired direction in an external magnetic field. Similarly, an optical signal can be used to guide the microscopic migration, as demonstrated previously using natural phototactic swimmers for cargo transport4. Some photoactive semiconductors such as silver chloride nanoparticles, titanium dioxide (TiO2) microparticles and TiO2 microtubes exhibit autonomous motion under illumination12,13,30. It has also been demonstrated that the illumination can suppress the migration of Ti/Cr/Pt catalytic microengines by decreasing the local fuel and surfactant concentration31, suggesting the feasibility of light-guided artificial micro/nanomotors.
Here, we demonstrate a light-driven microswimmer based on a Janus TiO2/Si nanotree structure32. This microswimmer incorporates both a photocathode and a photoanode. Under illumination, the light-driven photoelectrochemical (PEC) reaction generates anions and cations at opposite ends of the nanotree, resulting in an asymmetric distribution of generated ions, which propels the nanotree by self-electrophoresis. Moreover, similar to natural phototactic microorganisms2,3, the shading effect of the large TiO2 nanowire head enables ‘steering’ of the nanotree along the illumination direction. By controlling the zeta potential of the nanotree via chemical modification, we successfully programmed the microswimmers with both positive and negative phototaxis, which mimics the natural phototactic algae at both the individual nanomotor level and the macroscopic scale.
Self-propulsion of an artificial microswimmer
Figure 1a presents a schematic illustration of a microswimmer based on a Janus nanotree structure, in which TiO2 nanowire branches are grown on a p-type silicon nanowire trunk. In this nanotree, the TiO2 nanowires serve as the photoanode and the silicon nanowire serves as the photocathode32,33. We synthesized a large-scale Janus nanotree forest by silicon wet etching followed by TiO2 nanowire hydrothermal growth, using methods based on a previous study32 (Supplementary Section 1). On illumination, the PEC reactions promote reduction and oxidization reactions on the silicon trunk and TiO2 head, respectively, which leads to an asymmetric distribution of charged reaction products along the nanotree axis and propels the nanotree by electrophoresis. Figure 1b presents a scanning electron microscope (SEM) image of the as-prepared Janus nanotree forest composed of 7-µm-long silicon tails and 4-µm-long TiO2 nanowire heads (diameter of ∼2.6 µm). The silicon trunk and TiO2 nanowire are both single crystalline (Fig. 1c) so as to have good PEC activity and high electrical conductivity.
In the present experiment, a 365 nm ultraviolet light-emitting diode was used as the light source to power the PEC reaction. Two different electron/hole scavenger couple systems, hydrogen peroxide (H2O2) and 1,4-benzoquinone/hydroquinone (Q/QH2), were used to illustrate the PEC propulsion mechanism of our microswimmer. First, 0.1% hydrogen peroxide (H2O2) aqueous solution was used as both electron and hole scavengers because of its fast electrochemical reaction speed. Without illumination, hydrogen peroxide decomposition on the nanotree is negligible and the nanotree behaves as a Brownian particle (Supplementary Movie 1). Under illumination, the photoexcited holes in the TiO2 nanowire and the electrons in the silicon migrate to the semiconductor–electrolyte interface and promote PEC reactions as H2O2 – 2e– → O2 + 2H+ and H2O2 + 2e– → 2OH–, respectively. The PEC reaction was confirmed by a PEC cell measurement with TiO2 nanowires on fluorine-doped tin oxide (FTO) glass as the photoanode and a silicon nanowire substrate as the photocathode (Supplementary Section 2). The charge inside the Janus nanotree can be balanced quickly by recombination of electrons and holes on the interface between the silicon and TiO2 nanowires, and the low mobility of the ionic species in solution (3.6 × 10–3 cm2 V–1 s–1 for H+ in water, compared to ∼100 cm2 V–1 s–1 for holes in silicon) allows a local electric field to build up around the charged nanotree, leading to its autonomous swimming via electrophoresis in this self-generated electric field (Fig. 2a,c and Supplementary Movie 1). We observed that all pristine nanotrees migrate in a tail-forward direction (Fig. 2a), indicating that the overall zeta potential of the pristine nanotree is positive. This is expected, as the surface area of the positively charged TiO2 nanowire head is much greater than that of the negatively charged silicon tail. Because H2O2 decomposes spontaneously and can potentially provide energy for a chemically powered nanomotor6,9,34, we demonstrate that the PEC-driven microswimmer can harvest energy from light by replacing H2O2 with a classic reversible redox shuttle, benzoquinone/hydroquinone. The benzoquinone/hydroquinone redox couple is a well-studied model system in electrochemistry due to its perfect reversibility, and was previously used as a redox shuttle in a dye-sensitized solar cell35. Similar to the Janus nanotree in H2O2 solution, under illumination, OH− anions are generated on the silicon tail via a cathodic reaction Q + H2O + 2e– → QH2 + 2OH–, while H+ cations are generated on the TiO2 head via anodic reaction QH2 – 2e– → Q + 2H+. Autonomous migration of the Janus nanotree in the tail-forward direction is observed in a mixture solution of 1,4-benzoquinone and hydroquinone (1 mM:10 mM) (Fig. 2b and Supplementary Movie 3). To confirm that the propulsion mechanism of the Janus nanotree is electrophoresis rather than diffusiophoresis, we compared the migration of the Janus nanotree with a separated TiO2 nanowire shell, a silicon nanowire and an integrated nanotree with insulating SiO2 layer inserted between the silicon trunk and TiO2 nanowire branches (Supplementary Section 4). It was found that none of the control samples exhibited fast migration under our experimental conditions, verifying that the electrical current flow associated with the PEC reaction and electrophoresis mechanism is essential for the propulsion of the Janus nanotree. We also observed that the migration speed of the Janus nanotree scales linearly with solution resistivity, which is also in line with the proposed self-electrophoretic propelling mechanism6,36 (Supplementary Section 5).
Surface modification-dependent migration speed
Given that the photocurrent density J in the PEC reaction is proportional to the intensity of the incident light, the electric field E, as determined by Ohm's law (E = J/k) (where k is the electrical conductivity of the solution), is also proportional to the light intensity. As described in the Hückel equation37, the electrophoretic speed of a microparticle is proportional to both the applied electric field E and the particle's zeta potential ζ, which implies that the migration velocity of the nanotree, ve, also scales linearly with the light intensity as well as the nanotree's overall zeta potential. The linear dependence of the migration speed on light intensity in Fig. 3d indicates that the autonomous migration is indeed powered by the PEC reactions. To further confirm the self-electrophoresis mechanism, we altered the zeta potential of the Janus nanotrees38 by chemical treatment (see Methods and Supplementary Section 6) and studied its effect on the migration speed. Because the photocurrent depends on light intensity, the length of the nanotree and the TiO2 nanowires head percentage32 (the ratio of TiO2 head length over the total length of the nanotree), the migration speeds of nanotrees with similar geometries were compared (Supplementary Section 7). Here, we define the light-intensity-normalized migration velocity, which is the slope of the absolute migration speed as a function of light intensity, to benchmark the intrinsic migration ability of the nanotree. The light-intensity-normalized migration velocity is more reliable than the absolute migration speed and cancels out the influence of Brownian motion. Nanotree treatment with 3-[2-(2-aminoethylamino)-ethylamino]-propyltrimethoxysilane (AEEA) grafts positively charged amines on the silicon surface, which leads to a higher positive zeta potential38 and the tail-forward migration (Fig. 3a). As expected, the AEEA-treated nanotree shows significantly higher light-intensity-normalized migration velocity (95% confidence interval) than the pristine Janus nanotrees (95% confidence interval). Similarly, when negatively charged molecules are grafted onto a nanotree, which lowers the overall zeta potential, the nanotree should migrate more slowly or even in the reverse direction. Two negative modifications were tested in our experiments: (1) benzenesulfonic acid was grafted onto the silicon tail surface by treatment with 2-(4-chlorosulfonylphenyl)ethyltrimethoxysilane (CSPTMS) followed by hydrolysis in water and (2) negatively charged platinum nanoparticles were attached onto the TiO2 head by thermal decomposition of dropcast chloroplatinic acid (Supplementary Section 6). Both negatively modified nanotrees migrate in a head-forward direction (Fig. 3b,c) presented as negative speed in Fig. 3d, which implies that the overall zeta potential of the CSPTMS- or platinum-treated nanotree is negative.
Because the alignment of a nanotree is driven by the electric force on the TiO2 head, it is possible to program the alignment direction of the nanotree by chemical modification. A nanotree with a positively charged head (such as in a pristine and AEEA-treated nanotree) turns its head away from the light source, whereas a nanotree with a negatively charged head (such as the platinum-nanoparticle-decorated nanotree) turns its head towards the light source as the electric force direction is reversed.
For natural phototactic algae, such as Euglena gracilis and Chlamydomonas reinhardtii, chemical treatment could change the cytoplasmic redox poise of cells and regulate the sign of the phototaxis41. Analogously, our artificial swimmers could also be programmed to change the sign of the phototaxis by simple chemical treatment. The nanotree's swimming direction is determined by its overall zeta potential, while the surface charge of the TiO2 head determines whether it aligns along or against the illumination direction. Because the surface charge of different parts of the nanotree can be modified independently, various phototactic swimmers can be programmed. In general, if the overall nanotree and its TiO2 head are both positively or both negatively charged, the microswimmer will exhibit positive phototaxis. If the overall nanotree and its TiO2 head have different charge polarities, the nanotree will exhibit negative phototaxis (Supplementary Section 10). For pristine and AEEA-treated swimmers, the overall nanotree and its TiO2 head are both positively charged, resulting in ositive phototaxis (Fig. 5a,b). For CSPTMS-treated swimmers, the TiO2 head is still positively charged, as TiO2 quickly decomposes any grafted organic molecules under ultraviolet illumination, and the overall zeta potential of the CSPTMS-treated swimmer is dominated by the negatively charged benzenesulfonic acid grafted to the silicon tail, resulting in negative phototaxis (Fig. 5c). For the platinum-nanoparticle-decorated nanotree, the overall nanotree and its TiO2 head are both negatively charged, resulting in a positive phototactic migration in a head-forward direction (Fig. 5d and Supplementary Movie 6).
In nature, some prokaryotic green algae, such as Anabaena variabilis, can sense the light direction and exhibit phototactic behaviour42. However, prokaryotic phototaxis has been observed only on two-dimensional interfaces, because the interface plays a critical role in their phototactic migration. In contrast, eukaryotic algae with more complex structures, such as E. gracilis, have evolved with the ability of phototactic migration in three-dimensional open water. In our artificial microswimmer, the mechanism of sensing and rotating does not involve a substrate, which also enables three-dimensional phototactic migration. Here, we demonstrate the three-dimensional phototaxis of our artificial swimmers on a macroscopic scale and compare it with the natural phototactic green algae E. gracilis43. Janus nanotrees (∼1 mg) were dispersed in 3 ml 0.1% H2O2 solution in a UV–vis cuvette and illuminated by a commercial ultraviolet flashlight from one side. Positive phototaxis (swimming towards the light source) was observed for the pristine nanotree suspension (Fig. 6b) without supporting substrate, similar to natural E. gracilis (Fig. 6a). In contrast, the CSPTMS-treated nanotree suspension clearly exhibits negative phototaxis (swimming away from the light source) (Fig. 6c), as observed at the single swimmer level on a glass substrate (Fig. 5c), which confirms that the phototaxis of the swimmer is independent of the substrate and therefore truly mimics the more advanced eukaryotic phototactic algae (Supplementary Movie 7).
We have presented a light-controlled artificial microswimmer based on a Janus nanotree structure, which self-aligns with the direction of light propagation and mimics the phototaxis of natural motile algae. The advantages of the Janus nanotree structure are twofold. First, the three-dimensional TiO2 nanowires head enables controllable autonomous motion away from its body axis, which is critical for nanomotor navigation. Second, because the nanomotor is composed of different materials, orthogonal chemical modifications can be realized, thus offering large flexibility in nanomotor design. In our case, the Janus nanotree is ‘steered’ by its TiO2 head, while the overall zeta potential determines its migration speed and direction. By controlling the head and overall surface charges independently via chemical modification, we successfully programmed the artificial swimmer to show either positive or negative phototaxis.
One particularly desirable goal for micro/nanorobotics is to find use in healthcare applications such as non-invasive surgery and active drug delivery. With the development of optical techniques such as wavefront shaping, it is possible to direct tightly focused light deep into biological tissue44, which makes the optical navigation demonstrated here a feasible method to achieve high-precision in vivo manipulation. We believe that our design principle can be applied to more complicated micro/nanorobotic systems across different photoactive materials, such as narrow-bandgap semiconductors, photoactive polymers and proteins, so that harmful ultraviolet radiation can be replaced by the biocompatible visible or infrared illumination. Furthermore, the PEC-reaction-driven nanomotor offers more freedom for redox shuttle selection. We have demonstrated that the toxic H2O2 fuel can be replaced by the more biocompatible benzoquinone/hydroquinone redox shuttle, which suggests a wider biological application of PEC-reaction-driven nanomotors.
A solution-based synthesis process was developed based on a previous method32 to prepare a large-scale Janus nanotree array. After photolithographic patterning on the boron-doped silicon wafer, the silicon substrate was etched in aqueous solutions of 0.02 M silver nitrate (AgNO3) and 5 M hydrofluoric acid (HF). An array of vertical silicon nanowires was obtained after selectively removing thin silicon nanowires by wet oxidation and HF etching. Platinum nanoparticles were deposited on the silicon nanowires via electroless deposition to achieve ohmic contact between the silicon and TiO2. The silicon nanowire array was half-filled with polymethyl-methacrylate (PMMA) by dropcasting. After the wires were cleaned with oxygen plasma, 5 nm of TiO2 was deposited as a seeding layer on the exposed silicon nanowire tips by atomic layer deposition (ALD). The rest of the PMMA was removed in air at 450 °C. Hydrothermal growth of TiO2 nanowires was performed in an autoclave at 200 °C for 90 min. Fresh platinum nanoparticles were deposited on the silicon tail immediately before the motion test.
For chemical grafting, AEEA and CSPTMS were dissolved in absolute ethanol. Pristine Janus nanotrees on a silicon substrate were then placed into the solution, sealed and allowed to react for 15 h. The CSPTMS-treated sample was further reacted in deionized water for 2 h at 85 °C for complete hydrolysis.
For platinum nanoparticle decoration over the nanotree surface, a 10-nm-thick TiO2 protection layer was deposited on the nanotree via ALD. A small amount of chloroplatinic acid solution was dropcast onto the substrate and then decomposed into platinum nanoparticles at 600 °C in a vacuum.
Five 3 W 365 nm LED beads (LG Innotek) were attached to a customized black hollow box to serve as the light source. Each LED was controlled independently by a joystick for the nanotree navigation experiments.
For the macroscopic phototactic experiments, ∼1 mg of Janus nanotrees was dispersed in ∼3 ml 0.1% H2O2 solution in a UV–vis cuvette. The suspension was illuminated from one side by a 10 W commercial 395 nm ultraviolet flashlight. For comparison, E. gracilis (Carolina Biological Supply Co.) was used as a model natural phototaxis system. A tabletop incandescent lamp was used as an illumination source.
The authors thank S. Brittman from AMOLF and C. Liu from Harvard University for discussions. This work was supported in part by the Hong Kong Research Grants Council (RGC) General Research Fund (GRF17303015, ECS27300814), the University Grant Council (contract no. AoE/P-04/08), the URC Strategic Research Theme on New Materials and the URC Strategic Research Theme on Clean Energy (University of Hong Kong).
Supplementary Movie 1
Supplementary Movie 2
Supplementary Movie 3
Supplementary Movie 4
Supplementary Movie 5
Supplementary Movie 6
Supplementary Movie 7
About this article
Nature Communications (2018)