Photochemically activated motors: from electrokinetic to diffusion motion control

: Self-propelled micro/nano-motors that can transform chemical energy from the surrounding environment into mechanical motion are cutting edge nanotechnologies with potential applications in biomedicine and environmental remediation. These applications require full understanding of the propulsion mechanisms to improve the performance and controllability of the motors. In this work, we demonstrate that there are two competing chemomechanical mechanisms at semiconductor/metal (Si/Pt) micromotors in a pump configuration under visible light exposure. The first pro-pulsion mechanism is driven by an electro-osmotic process stemmed from a photoactivation reaction mediated by H 2 O 2 , which takes place in two separated redox reactions at the Si and Pt interfaces. One reaction involves the oxidation of H 2 O 2 at the silicon side, and the other the H 2 O 2 reduction at the metal side. The second mechanism is not light responsive and is triggered by the redox decomposition of H 2 O 2 exclusively at the Pt surface. We show that it is possible to enhance/suppress one mechanism over the other by tuning the surface roughness of the micromotor metal. More specifically, the actuation mechanism can be switched from light-controlled electrokinetics to light-insensitive diffusio-osmosis by only increasing the metal surface roughness. The different actuation mechanisms yield strikingly different fluid flow velocities, electric fields and light sensitivities. Consequently, these findings are very relevant and can have a remarkable impact on the design and optimization of photoactivated catalytic devices and, in general, on bimetallic or insulating-metallic motors.

In the last decade, many efforts have been devoted to increase motor sophistication by endowing them with multiple functionalities and tasks.However, there are still many fundamental questions related to the chemomechanical propulsion mechanisms that remain open, even more when an additional parameter such as light is incorporated.Unraveling the complex interplay of the different physical-chemical processes involved in the chemomechanical actuation is crucial to improve the levels of control and to optimize the desired applications [45][46] .
Evidences of the poor understanding of the propulsion mechanism can be grasped with some typical bubble-free motors in H2O2 solution.Common examples of these motors are the bimetallic rods or relatively small Janus microparticles composed of half metal and half insulator.The motors are propelled through phoretic processes triggered by the decomposition of H2O2 at their surfaces.In the case of bimetallic motors there is a big consensus that the chemomechanical actuation is rooted in an electrophoretic process [45][46][47][48] .The H2O2 decomposition involves separate oxidation half reaction at one segment and reduction half-reaction at the other segment, with generation and consumption of protons at those segments.A local electric field is self-generated by the proton current from one segment to the other, being especially strong at the interface between both metals.However, motors made of insulator and metal segments have opened an ongoing debate.In those cases the fuel decomposition takes place only at the metal side.It was initially suggested a neutral diffusiophoresis mechanism driven by the decomposition of H2O2 into H2O and O2 at the metal catalyst without intervention of ionic products [49][50] .In that case the asymmetric distribution of reaction products would create a pressure gradient across the particle.A fluid flow would be directed towards the high pressure (high solute concentration) side inducing particle motion 51 .However, recent investigations revealed that the details of the propulsion mechanism are quite complex and a relevant presence of ions and electrophoretic effects (and even bulk reactions) should be taken into account to provide a more complete description of the catalytic actuation [52][53][54] .
A more complex scenario is found when motors are composed of a semiconductor and a metal catalyst.We recently reported highly efficient metal/semiconductor pumps, which could be controlled by visible light 44 .These pumps can be considered as the inverse and immobilized counterpart of swimmers.Micropumps constitute ideal platforms to acquire relevant information related to the actuation mechanisms by analyzing the fluid motion with respect to the motor surface.Our previous studies were focused on pumps made of silicon and a metal (Pt, Au).We argued that light creates electron/hole pairs at the silicon and that electrons are transferred to the metal.As a result, light activates two separated redox reactions at the motor surface: i) H2O2 mediated oxidation at the semiconductor part, and ii) H2O2 reduction at the metal catalyst.The separated reactions give rise to a proton current from the semiconductor to the metal, thereby building up an electric field that triggers the fluid flow by electroosmosis 44 .
In this work, using a similar semiconductor/metal pump configuration, we show that it is possible to enhance/suppress the photoactivated and separated H2O2 redox reactions and convert them into light insensitive direct electrochemical decomposition of H2O2 on Pt.Such change in the chemical pathway can be achieved by just tuning the surface roughness of the metal.In the previous work detailed in ref. 44, we only detected the photoactivated chemomechanical mechanism because pumps were made of very smooth platinum discs.In this paper we demonstrate that the increase of metal roughness triggers a striking change in the flow velocity and generated electric fields.That is due to a switch of the actuation mechanism from a dominant light-controlled electrokinetic process to a light-insensitive diffusio-osmotic one.These findings are very relevant, as they can have a high impact on the design and optimization of light activated catalytic motors.

Results and Discussion
Micropumps were fabricated by patterning Pt discs on p-doped silicon surfaces using electron beam lithography and platinum evaporation.The metal deposition was performed either by sputtering or electron beam evaporation to tune the surface roughness.The devices were then subjected to a mild oxygen plasma treatment to clean the surface (See Supplementary information for fabrication details).Fig. 1 illustrates a schematic picture of the micropumps under study.The figure shows atomic force microscopy images of the Pt part prepared with both deposition techniques.Devices with sputtered Pt (Fig. 1b) exhibit a roughness factor 14-fold larger than that obtained through e-beam deposition (Fig. 1a).Scanning electron microscopy (SEM) characterization also indicated a higher surface roughness on the sputtered Pt films, as can be observed in the SEM images of the Supplementary Information (Fig. S1).From now on, we will refer to the micropumps as r-Pt/Si for the rough sputtered ones, and s-Pt/Si for the smooth e-beam evaporated ones.The s-Pt/Si devices present the expected behavior shown in our previous study 44 , which pointed out a photoactivated electrochemical actuation through an electroosmotic process.According to previous results, the pumping mechanism for s-Pt/Si is illustrated in Fig. 2 a and b.The light absorbed in the Si part generates electron-hole pairs.In the vicinity of the Pt disc, the holes oxidize the H2O2 at the silicon/fluid interface, and the resulting electrons are transferred to the metal where the reduction of H2O2 takes place.During the oxidation reaction, protons are produced at the silicon region, which are consumed at the metal side.During this process a proton current is produced that generates an electric field pointing towards the Pt side, thereby dragging the fluid in the same direction.Such fluid motion pointing in the same direction as the electric field is expected for surfaces with negative zeta potentials, as the Si/Pt case 44 .
The impact of Pt roughness on the pump performance is elucidated by comparing the results of fluid motion in smooth e-beam evaporated and rough sputtered Pt pumps illuminated by high intensity white light, and using 1% H2O2 as chemical fuel.The fluid motion is followed by tracking the motion of tracer particles of opposite surface charge (zeta potentials of ξ-= -83 mV and ξ+=46 mV for negative and positive colloids, respectively).More details of the particle tracers can be found in the Supplementary Information.Despite of their similar size and Pt thickness, the interaction of positive and negative tracers with the r-Pt/Si and s-Pt/Si devices is strikingly different.In the case of r-Pt/Si, positive tracers exhibit maximum radial velocity of -8 μm/s.These particles move towards the Pt disc, and after crossing the Pt-Si edge, their trajectory bends upwards in the direction perpendicular to the disc, (see Fig. 3a and Movie 1).Such behavior is totally different to that previously encountered on smooth s-Pt/Si devices for similar positive tracers (Fig. 3 b), which move towards the Pt disc at much higher speed (maximum velocity of -65 μm/s) and stick to it (Movie 2).Even stronger differences between rough and smooth pumps can be observed for negative tracers (Figs.3c and 3d).In rough pumps, negative tracers move towards the Pt disc and they either cross the Pt edges and stick on the Pt surface or they just stick at the rims of the Pt disc (Movie 3).This behavior is surprisingly different to that found in s-Pt/Si motors, in which a repulsion band is generated at the Pt disc, thereby leaving the Pt surface and surrounding Si completely free of negative tracers (Fig. 3d and Movie 4).We also evaluated the influence of photoactivation on r-Pt/Si and s-Pt/Si devices by varying the light intensity.Figure 4 compares the tracer velocities at three different light intensities (100%, 40%, and below 10% with a neutral light attenuation filter).In the case of s-Pt/Si motors the radial component of the velocity of the positive tracers significantly drops as the light intensity is attenuated.In contrast, the velocity of positive tracers in the r-Pt/Si system only exhibits a very weak reduction with light intensity attenuation, and negative tracers do not show any significant variation with light intensity.
To rule out the potential influence of contaminants in the pump behavior, we have analyzed by X-ray photoelectron spectroscopy (XPS) the chemical features of the Pt films deposited by both techniques.The XPS spectra of the sputtered and electron beam evaporated Pt show similar profiles (see Supplementary Information, Fig. S2), thereby ruling out the effect of impurities in their different catalytic actuation behavior.In addition, we have analyzed the zeta potential of the sputtered Pt, giving a value of -35 mV (See Supplementary Information), which is within the expected zeta potential range in noble metals (Au or Pt) 45,46 .Such value also discards possible contaminants that could generate positive zeta potentials and induce the change of the fluid flow direction in presence of an electric field.
All these findings suggest that the actuation in the semiconductor/metal pumps is a combined effect of two different mechanisms.In the case of the s-Pt/Si pumps the catalytic actuation is dominated by electro-osmosis.The catalytic reaction generates an electric field pointing towards the Pt disc that is drastically amplified by the incident light.Such electric field triggers a fluid flow in the same direction and, in addition, induces a strong attractive force towards the Pt surface on positive tracers, and repulsive force on the negative ones (Fig. 3).In contrast, the increase of Pt roughness seems to favor the direct H2O2 decomposition at the entire Pt surface area, whose process is light insensitive.In that case, the liquid motion is driven by the solute concentration gradient created by such reaction at the Pt surface.This actuation behavior suggests the generation of a diffusion-osmotic process that totally changes the interaction of charged tracers with the pump.
where  is the fluid permittivity, ξp the particle zeta potential, Er the radial component of the electric field, and η the fluid viscosity.Both Vrf and Er can be obtained by analyzing the velocities of the tracers with opposite charge (eq.1).The results collected in Table 1 show the pronounced decrease of both Er and Vf in the rough pumps compared to the smooth ones.To gain more insight into these competing mechanisms, we suppressed the photoactivated reactions, and hence their associated electric field contribution, by fabricating rough and smooth pumps on insulating SiO2.In the s-Pt/SiO2 pumps, we could not capture any tangible directional motion of the positive or negative tracers in the presence of 1% H2O2.However, the behavior of positive and negative tracers is dramatically different in r-Pt/SiO2 pumps (Fig. 5).Positive particles move towards the Pt disc at maximum speed of -(4.4±0.6)μm/s, and they drift upwards from the surface after crossing the Pt edges (Movie 5).Similar behavior is also observed with quasi-neutral particles (ξo= -12 mV), which approach the Pt disc at maximum velocities of -(5.7±0.5)μm/s.On the contrary, negative particles move towards the Pt disc with maximum velocities of -(6.9±0.5)μm/s, and then they stick to it (Movie 6).These different responses suggest the generation of an electric field in the Pt disc pointing outwards.However, the value of this electric field is too weak to enable repelling the positive tracers from the Pt surface.The weaker repulsive force is also due to the lower absolute value of the zeta potential in positive tracers as compared to the negative ones.As a result, the tracer motion is dominated by the fluid flow component, and is less sensitive to the generated electric fields.The quantification of Er and Vrf in the rough Pt/SiO2 pumps by using Eq. 1 yields 31.7 V −1 and -5.4 m -1 , respectively (Table 1).This analysis demonstrates the existence of an electric field in the opposite direction to that in smooth Pt/Si pumps, i.e. pointing outwards the Pt disc.Therefore, the electric field direction is opposite to the expected direction in electro-osmotically driven pumps whose elements exhibit negative zeta potentials, as is the case of Si and Pt.In pumps with negative zeta potentials, the electric field should drive the fluid flow in the same direction according to the equation, where ξs is the zeta potential of the surface (<-40mV in our case due to the strong contribution of the negative surface potential of silicon) 44 .
The weak positive electric field in rough Pt/SiO2 pumps cannot be the responsible of triggering a fluid flow in the opposite direction.Therefore, the catalytic actuation in the r-Pt/SiO2 pumps implies the generation of diffusioosmosis processes induced by the direct decomposition of H2O2 at the rough Pt surface.As a result, the fluid moves towards the region where the reaction takes place, i.e. the Pt disc where the solute concentration and the pressure are higher.
The existence of an electric field in the metal/insulator pumps is in line with previous findings in metal/insulator Janus particles, which pointed out the generation of an electric field induced by charged reactions and ionic ef-fects [52][53][54] .The origin of such ionic effects has been attributed to different sources.It was proposed that asymmetries in the metal catalyst, such as different metal thicknesses, could yield to reactions involving charged intermediates 53 .That could be a plausible explanation for rod and sphere like structures capped by the catalyst, in which a thickness gradient with thicker catalyst layer at the pole than at the equator sides is expected after metal evaporation.However, such thickness asymmetries are expected to be negligible in the flat configuration of our pumps.Other studies suggested the potential dissociation of H2O2 and water in ions at the metal surface and in the bulk solution as additional sources for a self-generated electric field 54 .However, the generated electric field in the r-Pt/SiO2 pumps is not the driving force of the fluid flow, as it should generate the electrokinetic fluid flow in the opposite direction to what it is observed.Therefore, the electrokinetic propulsion mechanism can be discarded in the metal/insulator pump.The chemical origin of the electric field is not well known at this moment but future studies including photoemission experiments are pursued to better understand the Si/Pt interface and the precise chemical reaction schemes that could yield to an electric field in such conditions.

Conclusion
The findings in insulating SiO2/Pt pumps demonstrate that at least two chemo-mechanical mechanisms are behind the liquid propulsion in Si/Pt pumps.The first one is an electro-osmotic process due to a photoactivated reaction taken place in two separated redox reactions: the oxidation of H2O2 at the Si side and the reduction of H2O2 at the Pt side.The induced proton current from Si towards Pt in these reactions generates the electric field towards Pt that drags the fluid towards the metal structure.In contrast, the second propulsion mechanism features diffusio-osmotic characteristics, with H2O2 decomposition taking place mainly at the Pt region.This decomposition process triggers the fluid flow towards the Pt structure to compensate the higher solute concentration and higher osmotic pressure.This latter mechanism is not light responsive.The dominance of one chemomechanical mechanism over the other depends on the catalyst roughness.In smooth Pt/Si pumps the photoactivated electro-osmotic mechanism prevails, whereas the diffusioosmotic and light insensitive one dominates in rough Si/Pt.As a consequence, to develop highly efficient light sensitive engines, the generation of separated redox interfacial reactions between Si and Pt should be maximized by minimizing metal roughness and increasing the Si/Pt perimeter.These results are then crucial for the design of new photoactivated nanoswimmers/pumps controlled by visible and near infrared light.More specifically, these findings can have direct impact on light controlled motion of semiconductor/metal micro/nanostructures, in which the metal can act as a single catalyst by itself, or as part of a coupled catalytic process with the semiconductor . .Survey XPS spectra of platinum films deposited on silicon substrates by sputtering (continuous blue line) and e-beam evaporation (continuous red line), respectively.The samples were measured as-received and without previous exposure to oxygen plasma.The experiments were performed with a PHOIBOS150 hemispherical analyser hosted in a UHV system with a base pressure of 5x10-10 mbar and using monochromatic AlKa radiation (1486.6 eV).The spectra reveal carbon contamination (C1s line at 285 eV) arising from the unavoidable exposure to the atmosphere and a weak nitrogen signal at about 400 eV, again due to exposure to air during transport from the deposition equipment to the XPS system.Both samples were less than 30 minutes to the atmosphere.No other contaminants can be detected in spite of the high surface sensitivity of the XPS technique.

Zeta potential of Pt deposited by sputtering
A rough estimation of the zeta potential of sputtered Pt was obtained with dynamic light scattering (Malvern ZetaSizer Nano ZS Instrument) using an aqueous dispersion of silicon rods coated with sputtered Pt.

Figure 1 .
Figure 1.Schematics of the pumps with (a) smooth ebeam evaporated Pt and with (b) rough sputtered Pt, together with their respective topographic AFM images taken at the same z scale.Sputtered Pt exhibits a root mean square roughness (Rq) of 5.5±0.3nm whereas Rq for electron beam deposited Pt is 0.40±0.05nm

Figure 2 .
Figure 2. a) Schematic picture illustrating the photochemical activated process with oxidation and reduction taking place at the Si and Pt sides, respectively.b) Schematic representation of the proton generation and consumption as a consequence of the light activated redox processes together with the generated electric field and fluid flow direction.

Figure 3 .
Figure 3. Interaction of positive tracers with r-Pt/Si (a) and s-Pt/Si (b) pumps.Interaction of negative tracers with r-Pt/Si (c) and s-Pt/Si (d) pumps.The schematics of the particle trajectory are also included for each pump configuration.

Figure 4 .
Figure 4. Influence of the photoactivation on the s-Pt/Si and r-Pt/Si pumps represented by the variation of the radial component of the tracer velocity as a function of light intensity.The s-Pt/Si devices exhibit strong light responsiveness, whereas the r-Pt/Si ones are almost insensitive to light intensity variation.The figure includes a zoom of the r-Pt/Si pump as inset, for more accurate comparison.The fluid flow and electric field values for the s-Pt/Si and r-Pt/Si pumps can be extracted by considering that the radial component of the velocity of the tracer Vrp is the sum of two contributions: i) an electrophoretic term (Vef) induced by the particle electric charge, and ii) the radial fluid flow velocity (Vrf):

Table 1
Maximum values of the radial fluid velocity (Vrfmax) and Electric field (Ermax) for the different micropumps.Negative sign denotes radial inward direction of the electric field and fluid flow.

Figure 5 .
Figure 5. Interaction of positive (a) and negative (b) tracers with the r-Pt/SiO2 system.Schematics of the trajectories for positive and negative tracers at the rough Pt/SiO2 pump are also included.c) Schematic depicting the fluid flow and electric field direction.
Figure S2.Survey XPS spectra of platinum films deposited on silicon substrates by sputtering (continuous blue line) and e-beam evaporation (continuous red line), respectively.The samples were measured as-received and without previous exposure to oxygen plasma.The experiments were performed with a PHOIBOS150 hemispherical analyser hosted in a UHV system with a base pressure of 5x10-10 mbar and using monochromatic AlKa radiation (1486.6 eV).The spectra reveal carbon contamination (C1s line at 285 eV) arising from the unavoidable exposure to the atmosphere and a weak nitrogen signal at about 400 eV, again due to exposure to air during transport from the deposition equipment to the XPS system.Both samples were less than 30 minutes to the atmosphere.No other contaminants can be detected in spite of the high surface sensitivity of the XPS technique.