Fabrication of Pt Nanoparticle-Decorated Tubular Microengine
Figure 1a illustrates the experimental procedure for the fabrication of Pt nanoparticle-decorated microengine. The fabrication strategy was based on rolled-up technology using photoresist as a sacrificial layer (see the “Experimental Section” section for details) [12]. Briefly, bilayer nanomembranes with different thicknesses and thickness ratios (e.g., SiO/SiO2 5/20 nm, Ti/SiO2 20/10 nm, Ti/Co 10/10 nm, SiO2/Ti 10/20 nm) were deposited on photoresist patterns via electron beam evaporation. After selective etching of the sacrificial layer, the bilayer was set free and the strain gradient causes rolling of the bilayer nanomembrane into microtube [12]. Geometrical parameters such as the length, diameter, and shape of the microtubes can be tuned on one hand by changing the dimensions of the photoresist patterns and on the other hand by controlling the angles, rates, and thicknesses during the depositions of the nanomembranes [28]. After formation of microtubes, Pt nanoparticles were coated on the tube wall by ALD, where two self-limiting and complementary reactions are used in an alternating sequence [29]. On the first circle, the PtOx was produced during piping in a pulse of O2. Then, a pulse of methylcyclopentadienyl-(trimethyl) platinum(IV) ((MeCp)Pt(Me)3) is forced into the generator’s chamber, which reacted with the PtOx layer and O atoms are removed, leaving only Pt. On the next cycle, the unreacted precursor was removed and Pt surface was oxidized during the pulse of O2, preparing it for the next cycle [30, 31]. Due to its high surface energy, Pt deposition on supports proceeds via an island growth mechanism (Volmer–Weber mechanism) during the initial stages of ALD processes [30, 31]. Ultimately, after a sufficient number of exposure cycles, the islands will merge to form a film. However, for applications in catalysis, it is typically undesirable to obtain a continuous film: the island structure should be maintained because the islands/nanoparticles with a high surface-area-to-volume ratio should have better catalytic activity compared with flat layer [32, 33]. In current work, Pt nanoparticles were uniformly coated on the surface of the tube walls by precisely controlling the number of cycles adopted.
Figure 1b displays bird-view scanning electron microscopy (SEM) image of a typical 50-μm-long Pt nanoparticle-decorated SiO/SiO2 microtube under low magnification. A close examination of such tubular structure (Fig. 1c) reveals that, unlike common rolled-up microtube with a Pt smooth surface [34], the current microtube is covered by nanoparticles with average diameters of ~10 nm. As will be illustrated below, such Pt nanostructure leads to a dramatically increased catalytic surface area [35] and corresponding improved propulsion efficiency [36].
We further investigated formation of Pt nanoparticles on different microtubular structures. Figure 2a–d shows that the microtubes of well-defined lengths and geometries by rolling different nanomembranes can be arranged into ordered arrays. Such arrays can be mass-produced by normal photolithography and this makes it easier to prepare a large number of microengines simultaneously [37]. As demonstrated in our previous work [12], the diameter can be tuned by changing the layer thicknesses, the thickness ratios, and the built-in strain in the nanomembrane. In present case, the SiO/SiO2 microtubes have diameter of 5 μm and Ti/SiO2 microtubes have lager diameter of 12 μm due to different nanomembrane thicknesses and stress gradient therein. In order to illustrate the Pt nanoparticles on the inner tube wall for details, Fig. 2e–h shows the corresponding SEM images. It is found that nanometer-scale islands nucleate on the wall of microtubes after ALD cycles. The energy dispersive X-ray spectra of the samples (not shown) clearly prove the presence of Pt on the tube walls. However, the nanoparticles on different top layer of nanomembranes (inner tube wall) show different sizes and morphologies after the same ALD process. The nanomembranes with oxide top layers (i.e., SiO/SiO2 and Ti/SiO2 bilayers) exhibit very flat and smooth surface and the Pt nanoparticles on them appear in the form of irregular shapes like ellipses and bars (Fig. 2e, f). On the other hand, the nanomembrane containing metallic layers (Ti/Co and SiO2/Ti in present case) is relatively rough and uneven, and Pt nanoparticles in the form of small semi-spheres on the surface can be observed (Fig. 2g, h). We believe that the morphological difference in the bilayer nanomembranes is mainly due to different growth models and surface energies between oxide and metals during electron beam evaporation [38, 39]. In such incoherent growth condition, the growth of large particles/islands as a result of dissolution of small particles/islands can be explained by Ostwald ripening mechanism [38]. These factors also cause the change of the shapes of Pt nanoparticles when they are deposited on nanomembranes with oxide and metal top layer. However, it should be mentioned that, for the sake of simplicity, we suppose the nanoparticles are all in the shape of semi-spheres in the following model. This certainly introduces deviation in the model, but as we will discuss later, the experimental results can fit theoretical prediction well, suggesting that this simplicity is acceptable. Based on Fig. 2e–h, we have calculated the mean sizes of Pt nanoparticle on the inner wall (top layer). The results are 11, 10, 5, and 6 nm for nanoparticles on the surfaces of SiO/SiO2, Ti/SiO2, Ti/Co, and SiO2/Ti nanomembranes, respectively (Additional file 1: Figure S1). And the densities of nanoparticles are as high as 3.07 × 1015, 4.62 × 1015, and 1.85 × 1016, and 3.18 × 1016 m−2, respectively. It is clear that the Pt nanoparticles on the inner tube wall of SiO/SiO2 and Ti/SiO2 microtubes are larger than those on the inner tube wall of Ti/Co and SiO2/Ti microtubes, but the densities show the opposite result.
The Motion of Pt Nanoparticle-Decorated Tubular Microengine
Figure 3a–d shows time-lapse images of the movement of Pt nanoparticle-decorated SiO/SiO2 microengines in 10 % H2O2 (see also Additional file 2: Video 1). Oxygen bubbles ejected from one large end of microengine through the decomposition of H2O2 and propelled the microengine in opposite direction [40]. It is worth noting that both inner and outer surfaces are covered with Pt nanoparticles after Pt coating by ALD, but we observed no O2 bubbles generating on the outer surfaces of microengines. This indicates that O2 molecules have different nucleation behaviors on the inner and outer surfaces. The similar phenomenon had also been found in single-component metal oxide tubular microengines controlled by UV light [41]. It is considered that the geometries of the microengines have significant influence on bubble nucleation and generation. Generally, the bubbles can be formed on solid surfaces, if the gases reach heterogeneous nucleation energy [42]. Previous literature demonstrated that there are two factors determine the heterogeneous nucleation energy: the gas saturation concentration and the curvature of the surface [43]. The energy required for bubble formation on a flat surface is less than on a convex surface, and even less energy is required on a concave surface. It indicated that the gas produced on the concave surface of inner tube wall is much easier to nucleate compared than that on the convex surface of outer tube wall. In addition, different from other microengines such as Janus–motor [44] and Au-Pt nanorod [45], our microtubes can be used as a gas collecting chamber and O2 molecules produced inside the microtube will easily reach the supersaturation concentration for the bubble nucleation due to the accumulative effect of the inner confined space [46]. The accumulated O2 gas in the microtube can further facilitate the bubble nucleation. We noticed that the existence of Pt nanoparticles on the surface of the tube wall makes the catalytic decomposing reaction much more intense compared with smooth Pt layer, and high frequency bubble generation forms a long tail at the tube end. In our previous work [47], for a microtubular engine, we used the following Eq. (1) to calculate the oxygen productivity dV
O2/dt:
$$ \mathrm{d}{V}_{{\mathrm{O}}_2}/\mathrm{d}t=n{C_{{\mathrm{H}}_2}}_{{\mathrm{O}}_2}\pi \mathrm{R}\mathrm{L} $$
(1)
where n is O2 production rate constant which was experimentally estimated to be ≈9.8 × 10−4 ms−1 in our previous work [46] and \( {C_{{\mathrm{H}}_2}}_{{\mathrm{O}}_2} \) is the concentration of H2O2. This equation is considered to be valid for the microengines with a smooth inner surface. However, for the current microengines decorated with Pt nanoparticles, the inner surface area is much bigger than 2πRL. Apparently, the oxygen production is much higher than the microengines with smooth Pt layer, suggesting that Pt nanoparticle-decorated microengines can produce more oxygen. The corresponding bubble generation frequency makes the decorated microengines move in higher speed as we will explain in detail later. Detailed analysis of the video and time-lapse images demonstrates that the microengine was propelled at an ultrafast speed of around 3200 μm s−1 (Additional file 2: Video 1). According to the literature, for swimmer at low Reynolds number, the drag force (F) acted on the microengine is proportional to the motion speed (v) [47],
$$ F=-\frac{2\pi \mu Lv}{1\mathrm{n}(X)-0.72} $$
(2)
where X = 2 L/R is a geometrical parameter (L and R are the length and radius of the microengine, respectively) and μ is the fluid viscosity. The motion with faster speed means that the Pt-coated microengines need to overcome higher resistance. Moreover, the output power is proportional to the square of the motion speed since the output power is the product of the driving force and speed. In present case, one can deduce that the output power of Pt nanoparticle-decorated Ti/SiO2 microengines is also remarkably increased due to its ultrafast speed, and quantitative analyses of the speed promotion will be given below. We believe that the higher output power could enable this kind of microengines to accomplish more complex tasks in the future. For instance, we observed a powerful microengine spewing tiny bubbles off their back and push along a big bubble in the front (see the Additional file 3: Video 2), suggesting potential applications of powerful microengines in the field of microdelivery [48] or smart drug delivery [49]. It is worth noting that the performance enhancement is not limited to SiO/SiO2 microengine after decoration with Pt nanoparticles. Our results indicate that Pt nanoparticle decoration also leads to acceleration of other kinds of microtubular engines. In order to elucidate this phenomenon clearly, in Fig. 3e–f, we show the trajectories of four microengines moving in 10 % H2O2, extracting from the corresponding Additional files 4, 5, 6, and 7: Videos 3–6. One may note that the trajectories and the microbubble tails show unique geometries like linear, circular, and helical curves (Additional file 1: Figure S2). It is considered to be due to the imperfection in the microtubular structures, which generates a torque which is not parallel to the axis of microtubes resulting in different movement behaviors [40]. Quantitatively, the moving distances over a period of 0.5 s are 1489, 1121, 762, and 830 μm for Pt nanoparticle-decorated SiO/SiO2, Ti/SiO2, Ti/Co, and SiO2/Ti microtubes, respectively (Fig. 3). We found that particle distribution and size have a great influence on the surface area and therefore the performance of microengines. The surface area of SiO/SiO2/Pt, Ti/SiO2/Pt, Ti/Co/Pt, and SiO2/Ti/Pt microengines is 1.48, 1.80, 1.42, and 1.20 times larger, respectively, compared with smooth microtubular structure (Additional file 1: Figure S3), and thus, they demonstrated efficient catalytic effect, powerful propulsion thrust, and distinct moving trajectories, as shown in the Additional files 8, 9, 10 and 11: Videos 7–10. In addition, the enhanced surface area due to the existence of Pt nanoparticles also makes the microengines available to work in solution with low H2O2 concentration, and the motion of Pt nanoparticle-decorated microengines in 5 mL 10 % H2O2 after 24 h was shown in Additional file 12: Video 11. We experimentally found that the threshold H2O2 concentration for current Pt nanoparticle-decorated microengine can be as low as ~0.5 %. The time-lapse images in Additional file 1: Figure S4 display a Pt nanoparticle-decorated SiO/SiO2 microengine moving in a 0.5 % H2O2 solution. Although the oxygen bubble generation frequency is low, the microengine is nonetheless self-propelled at a speed of ~100 μm/s.
The Experimental Results and Theoretical Model
To investigate the motion of decorated microengines in more details, we have calculated average speed of the four types of microengines based on statistics of 10 microengines in each case. Figure 4a shows the average speeds of the four types of Pt nanoparticle-decorated microengines moving in 5 and 10 % H2O2 solution. It is obvious that the average speeds of all four types increase with the concentration of H2O2 due to higher O2 productivity (see below).
According to body deformation model [45], the bubbles propel the microengine in a stepwise manner and the average speed of the smooth tubular microengines (v) can be theoretically predicted from Eq. (3)
$$ v=\frac{9n{C}_H{{{}_{{}_2}}_O}_{{}_2}X}{6+X/\left(1\mathrm{n}X-0.72\right)}, $$
(3)
The above equation suggests that the average speed of a microengine is mainly determined by the geometrical parameter X under the certain H2O2 concentration, as plotted by the red dashed curve in Fig. 4b. One can see that the Pt nanoparticle-decorated SiO/SiO2, Ti/SiO2, and SiO2/Ti microengines exhibit higher speeds compared with the theoretical prediction (1.38, 2, and 1.18 times, respectively), mainly due to the increase in the surface areas. If the surface areas are normalized (red, blue, and black squares in Fig. 4b), the experimental results can fit theoretical prediction very well if one notices that the surface areas were calculated by a simple approach (Additional file 1: Figure S3). This further proves that the larger surface area of the Pt nanoparticles (as calculated before) is mainly responsible for the highly efficient propulsion behavior of microengines, although the nanoparticle geometry may also affect the catalytic activity [32, 33]. Whereas in the case of Pt nanoparticle-decorated Ti/Co microengine (green square in Fig. 4b), the motion speed is slower than the theoretical prediction if the surface area is normalized. The surface area increased 1.42 times, but the speed increased only 1.26 times compared to the theoretical calculation. We assign this deviation to different surface morphology: the surface of the Ti/Co is unflat compared to other three samples, especially those with pure oxide bilayer nanomembrane, as we have mentioned above (see Fig. 2g). This may significantly influence the nucleation of gaseous microbubbles in the tubular chamber during catalytic motion and may also influence the dynamics of the microengine when it moves with high speed at low Reynolds number. In addition, we cannot rule out the possibility of the existence of electrochemical process in the O2 production. The Ti/Co microengine is the only one in the current four samples with conductive tube wall. Although this needs further investigation, we consider that the electrochemical process therein may be one of the possible reasons leading to smaller O2 productivity and corresponding slow motion speed.