Dynamical Force and Imaging Characterization of Superhydrophobic

Nov 14, 2011 - Elegant Shadow Making Tiny Force Visible for Water-Walking Arthropods and Updated Archimedes' Principle. Yelong Zheng , Hongyu Lu , Wei...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/Langmuir

Dynamical Force and Imaging Characterization of Superhydrophobic Surfaces Tuck Wah Ng* and Yohannes Panduputra Laboratory for Optics, Acoustics & Mechanics, Monash University, Clayton VIC 3800, Australia

bS Supporting Information ABSTRACT: We devised a dangling cantilever optical lever setup with imaging that permits dynamical studies of superhydrophobic surfaces without the effects of gravitational acceleration for better insight into the mechanics. The setup enabled us to ascertain liquid loss and ascribe it to the interaction of liquid that just touched the superhydrophobic surface as it translated at various constant lateral speeds. At lower speeds (2060 μm/s), the interactions were characterized by a strong initial liquid pin (at up to 0.6 nN force) and depin followed by a series of smaller force pin and depins before sufficient liquid loss led to total liquid detachment from the surface. At higher translation speeds (80100 μm/s), the interactions were characterized by liquid pinning and depinning processes at a sustained force (around 0.7 nN) in which liquid loss was low enough to engender a much later liquid detachment (beyond 100 s). A linear reduction of the receding contact angle with time, but not with the advancing contact angle, was found up to the point of first liquid depinning. This suggested a stronger role played by the receding contact line in establishing liquid adherence to the superhydrophobic surface. The detachment process from the surface was also characterized by a liquid bridge driven to rupture by way of liquid being conveyed away from the bridge.

’ INTRODUCTION A surface that allows water to displace easily from it so that it remains dry is normally known as superhydrophobic. Such an ability is found in nature,13 and surfaces tailored to mimic this behavior for applications46 and to reveal unexpected characteristics7,8 are actively reported. The often used condition for a surface to qualify for superhydrophobicity is for the effective contact angle of a liquid droplet resting on it to exceed 150°. This generally presupposes a linkage of this condition with low surface adhesion. Nevertheless, it is possible for a superhydrophobic surface to also be adhesive due to the manner of wetting.9 For a solid surface that possesses high roughness on the micrometer and nanometer scale, liquid is able to impregnate the surface totally (the Wenzel state) or the true liquidsolid contact replaced by a highly energetic liquidvapor interface such that the liquid interface suspends on a gas cushion (the Cassie or fakir state). It is now believed that both wetting states are able to coexist even on a highly structured micromanufactured surface (with pillars etc.),10 and thus should more readily occur on one that is not as structured. Understanding the dynamical rather than the static behavior of liquid on a superhydrophobic surface is vital, as this will be the main mode of its operation. Currently, this is studied by means of allowing a droplet to roll/slide off an incline.11,12 With this method, the effect of gravitational acceleration cannot be eliminated. This prevents the ability to monitor the mechanics at constant and slow speeds. Furthermore, as the movement of a droplet on a superhydrophobic surface is rapid, control is difficult and recording using a high speed camera is often needed. We describe here a r 2011 American Chemical Society

means to circumvent this. It permits interaction forces to be directly measured. In the process, this technique also yields important information on the detachment of liquid from the surface by way of the formation and eventual destruction of a liquid bridge. Measurement Technique. For liquid that is dispensed between two surfaces, one wetting (above) and another superhydrophobic (below), the strong wettability of the top surface will ensure that the liquid remains pinned there even when the superhydrophobic surface is moved laterally (in the y-direction) (see Figure 1a). The upper surface (in the form of a rod) can be attached to a cantilever flexible foil (see Figure 1b). By attaching the rod to a cantilever flexible foil, any lateral movement will result in its bending. If the length of the foil is L, the point force F is related to the lateral movement δ there via13 δ ¼ FL=3EI

ð1Þ

where E and I are the Young’s modulus and second moment of area of the foil. respectively. At some distance x from the free end, the slope of foil is given by θðxÞ ¼ FðL2  x2 Þ=2EI

ð2Þ

Hence the slope is related to the lateral displacement of the droplet θðxÞ ¼ 3δðL2  x2 Þ=2L

ð3Þ

Received: September 22, 2011 Revised: November 10, 2011 Published: November 14, 2011 453

dx.doi.org/10.1021/la203732g | Langmuir 2012, 28, 453–458

Langmuir

ARTICLE

Figure 1. Description of (a) a liquid placed between a wetting rod and superhydrophobic surface and (b) the rod attached to a flexible foil cantilever.

which indicates a direct relationship between both parameters. As the deformed foil is reflective, its change in slope at any point along its length can be accurately monitored with a laser beam incident there and reflecting off to a quadrant photodiode sensor.14 If the distance between the point on the foil to the photodiode l is relatively large, the laser beam on the photodiode will displace according to Δ = 2lθ. The quadrant photodiode can be read out in which the voltage registered is a linear function of Δ provided that the required measurement precautions are made.15 By eliminating θ in eq 3, we have

Figure 2. Schematic description of the dangling cantilever setup with optical lever sensing to study the dynamical behavior of superhydrophobic surfaces. adhesive. The laser used was a continuous HeNe (Melles Griot 25 LHP 991230) with a wavelength of 632.8 nm which was directed onto the foil at a distance of 152 mm from the free end. The laser beam was delivered to a commercially available quadrant photodiode (Pacific Silicon QP506SD) located 139 mm away from the foil. The signals from the quadrant photodiode (interfaced via a LabJack U12 Legacy) and video camera were sent to a personal computer for analysis and processing. Test surfaces of interest were placed on a linear actuator (Zaber T-LS-13SM motorized linear stage) with positional and speed resolutions of 0.1 μm and 0.001 mm/s, respectively. An optomechanical stage of (10 μm resolution allowed this surface to move up to toward the PMMA rod. The system was calibrated by placing a small piece of putty on a glass slide and locating it on the optomechanical stage. The putty was then raised so that it was just to the left of the rod. This material was used to ensure a positive contact with the rod. Movement of the linear stage to the right then brought the rod together with it. The actuator was translated at 20 μm/s as the output from the quadrant photodiode was monitored. The dynamical behavior of liquid on a superhydrophobic surface was studied by placing a 2.5 μL water droplet manually pipetted (Eppendorf 0.110 μL) to attach to the base of PMMA rod. The superhydrophobic substrate prepared was then placed on the linear actuator. The optomechanical stage brought the superhydrophobic surface up to just contact the liquid, leaving a 1.2 mm distance between the PMMA rod and superhydrophobic surface. Images confirming liquid behavior were recorded using a video camera (Moticam 2000) with a 1/3 in. CCD sensor comprising 2 megapixels. The lens used was an InfiniProbe (8 magnification) with a focus length of 18 mm to infinity.

3lðL2  x2 Þδ ð4Þ L Relation 4 tells us that the voltage registered should be a linear function of δ. Suppose that the voltage registered is V and a calibration process of plotting V against δ is made such that the gradient is given by K. Incorporating eq 1, the force developing at any instant can be determined from the voltage using Δ¼

F ¼

3EIV LK

ð5Þ

For a rectangular foil of width w and thickness t, the second moment of area is given by I = wt3/12.

’ METHODS Superhydrophobic Surface Preparation. The copper substrate was produced using an electroless galvanic deposition process in which the polished surface was first cleaned using absolute ethanol and then allowed to air-dry. It was then immersed in a 24.75 mM aqueous solution of AgNO3 for 1 min. After this, the surface was thoroughly rinsed with distilled water followed by absolute ethanol before being allowed to air-dry. Once dried, it was immersed in a 1 mM solution of the surface modifier CF3(CF2)7CH2CH2SH in absolute ethanol for 5 min. After removal, it was again rinsed with copious amounts of distilled water, followed by absolute ethanol, and then dried. Superhydrophobicity was confirmed by taking a picture of a sessile droplet and measuring its contact angle. The structure of the sample was checked by taking images using a scanning electron microscope (FEI Quanta 3D FEG). Setup and Measurements. The setup shown in Figure 2 was devised wherein a rectangular cantilever of length 100 mm and width 5 mm was shaped from aluminum foil of 25 μm thickness (AL000420, GoodFellow). A poly(methyl methacrylate) (PMMA) rod of diameter 2 mm and length 5 mm was attached to the base of the foil using some

’ RESULTS AND DISCUSSIONS The sessile drop contact angle was measured to be 158° (σ = 0.80°), confirming that the surface was indeed superhydrophobic. Images of the sample obtained using a scanning electron microscope are shown in Figure 3. It can be seen that the deposition process resulted in the growth of dendritic fractal structures which increased the roughness of the surface, thereby affecting its wettability. 454

dx.doi.org/10.1021/la203732g |Langmuir 2012, 28, 453–458

Langmuir

ARTICLE

Figure 3. Images of the superhydrophobic surface obtained using a scanning electron microscope (FEI Quanta 3D FEG; accelerating voltage = 10 kV; working distance = 9.6 mm). The surface roughness exists as a dendritic fractal pattern.

Figure 5. Force versus time traces of the superhydrophobic surface translated at (a) 30 μm/s, (b) 90 μm/s, and (c) 70 μm/s. Figure 4. Calibration curve of voltage registered by the quadrant photodiode sensor and translation of the rod. The linear trend permits the slope to be determined and thus correlates the voltage recorded to force.

interaction in which the force reaches a maximum and then remains constant (which gives rise to the possibility of establishing a dynamical friction coefficient) after that. Eventually, the liquid lost contact with the surface altogether (from point III onward, or at 70s after the start of the trace). In order to confirm that this was not due to the surface being tilted, we moved the translator in the opposite direction and ascertained that there was no liquid reattachment to the surface. The ability to detach altogether allowed us to establish that there was liquid loss by way of deposition onto the surface, notwithstanding that it was superhydrophobic. We conducted another simple experiment by leaving a droplet on the rod without surface interaction and recorded its shape evolution over time. From the negligible change found, we were able to confirm that evaporation was not a contributing factor to liquid loss in our experiments. In a minor digression, we note that when detachment occurred (see inset plot for Figure 5a) there were periodic force oscillations before eventual settling. The frequency of this oscillation was around 6.7 Hz which should correspond with the natural frequency of the system comprising the foil, rod, and droplet. These oscillations were not discernible from the video footage. Also, the force just prior to detachment was small (about 0.1 nN). Under a Wenzel wetting mode, liquid interaction loss should occur in the form of liquid being “peeled away” to leave behind remnants (Figure 6a),18 whereas in the Cassie mode this should assume the form of broken liquid bridges at the peaks19 (Figure 6b). While not able to conclusively establish this, we posit that it was the Wenzel wetting mode that predominantly dictated contact line attachment and thus the hysteresis in our case. This is supported

Figure 4 presents the voltage versus displacement trace of the calibration process. A highly linear trend (in which R2 = 0.988) is evident. Based on the plot, the value of K was found to be 0.32 V/μm. Using the dimensions of the foil, I = 6.5  1018 m4. The other parameters to determine the force in relation to voltage in eq 4 were E = 70 GPa (as the foil material was aluminum) and L = 50 mm. A typical trace of the force interrogated from the bending of the cantilever foil against time as the droplet interacted with the superhydrophobic surface translating at a constant lateral speed of 30 μm/s is given in Figure 5a. It can be seen that the resisting force starts off from minimum (point I) and builds up to a maximum (point II) which is indicative of the effect of contact angle hysteresis in action. This maximum corresponds to a force of around 0.6 nN. Essentially, the liquid was still pinned on the surface up until this point. Based on the video footage, we observed the advancing contact line being breached first, which is consistent with previous observations using large droplets on an incline.16,17 Beyond this point, the liquid underwent a series of depinning and repinning to try to re-establish adhesion to the surface. This is clearly detectable from the force trace. It should be noted that negative values are obtained at times which correspond to the cases where the liquid body was able to swing past the equilibrium position (when the surface was stationary) to the other side when a depinning event occurred. More importantly, the overall behavior was in contrast to a typical solidsolid 455

dx.doi.org/10.1021/la203732g |Langmuir 2012, 28, 453–458

Langmuir

ARTICLE

to some extent by the mechanism of liquid bridge rupture reported under Cassie wetting that is dependent on tension as a droplet rolls.19 Such an event should not happen here as the surface was translated laterally. It appeared that, each time the liquid repinned to the surface, some degree of Wenzel wetting was restored. Hence, a series of these processes (pinning and depinning) finally resulted in enough liquid loss to detach the liquid altogether from the surface. A markedly different situation occurred when the translator speed was increased to 90 μm/s. The forcetime trace (Figure 5b) revealed a similar resisting force buildup to a maximum (point II), indicative of the liquid still pinned onto the surface. Beyond this, the liquid again underwent a series of depinning and repinning in order to re-establish adhesion to the surface. However, despite this, the average force was maintained nearly constant near the force level at point II, allowing the liquidsolid interaction to now behave almost like a solidsolid interaction. It will appear that a faster-moving surface engendered a lower level of liquid loss on the surface from each depin, thus allowing each repin then to re-establish a degree of wetting adhesion force to the surface that was close to the original. This finding establishes a link between the ability to sustain a restoring force to the process of liquid loss from the surface, with the latter being influenced by the surface translation speed. Nevertheless, over a longer period of time, there should be sufficient liquid loss from interaction (or from evaporation) that will eventually cause a detachment. With the existence of two such distinct characteristics, a natural question to be raised is the manner of how they would transition. Figure 5c is a typical forcetime trace at 70 μm/s, which suggests a somewhat competitive mode between the ability to repin on the

surface and the inability to do so due to interaction liquid loss. Based on the experiments conducted at different speeds, we found the transitioning behavior to occur in the range of translation speeds between 60 μm/s and 80 μm/s in our case. Due to the stochastic nature of the surface, the delineation was not sharp. The video footage recorded in tandem with the forcetime traces permitted the forces to be relatable to the evolution of contact angle hysteresis. Figure 7a and c presents the images recorded at points I and II of the forcetime trace of Figure 5b, while Figure 7b is the image recorded at an intermediate point in between (i.e., 3.3 s after I). From these image sequences, it can be seen that there was a reduction in the receding contact angle (defined as associated with the contact line on the right of the liquidsolid interface) from I to II. No such trend, however, could be established for the advancing contact angle (which is taken to be associated with the contact line on the right of the liquidsolid interface). This is corroborated by quantitative measurements of the advancing and receding contact angles (see Figure 8) which were taken from images at time intervals of 0.3 s between points I and II in Figure 5b. With the receding contact angle distribution, an almost linear reducing trend starting from 160° (which corresponded almost exactly with the equilibrium contact angle) to 120° was evident before the liquid first depinned. With the advancing angle, we see the value hovering around 170°. This infers the advancing contact line depinning very regularly from the surface as it moved. That the receding

Figure 6. Under Wenzel wetting mode A, liquid loss will occur in the form of liquid being “peeled away” to leave behind remnants, whereas in the Cassie mode B, this may assume the form of broken liquid bridges at the peaks.

Figure 8. Plots of the measured advancing and receding contact angles based on time lapse from point I on Figure 5b. The receding angle plot is reducing but the advancing angle plot is constant (trend lines provided). The plots refer to the cases up to point II in Figure 5b.

Figure 7. Image sequences of the liquid interaction with the surperhydrophobic surface at (a) point I on Figure 5b and after (b) 3.3 s, and (c) 6.6 s later stages when the surface was translated at 90 μm/s. They permitted the advancing and receding contact angles to be determined. A video sequence of this is provided as Supporting Information. 456

dx.doi.org/10.1021/la203732g |Langmuir 2012, 28, 453–458

Langmuir

ARTICLE

Figure 9. Image sequences during the later stages when the surface was translated at 30 μm/s which showed the transition from liquid adhesion to detachment from the surface. A video sequence of this is provided as Supporting Information.

contact angle, but not the advancing contact angle, possessed a linear reducing trend indicates that the adhesion of the liquid on the surface was primarily determined by the pinning of the receding contact line. At this point, it is appropriate to review the currently accepted understanding of the ability of liquid to move in relation to the contact angles.20 Suppose we express γSV, γSL, and γLV as the surface free energies of the solidvapor, solidliquid, and liquid vapor interfaces, respectively. One could always expect a change in the shape of the liquid before it depinned. This shape change can be regarded as an activation barrier to motion. If the solid liquid contact area did not remain constant, the activation energy would be given by Ea = γLVΔALV + γSLΔASL + γSVΔASV; where ΔA denotes the areal changes. Logically, ΔASL and ΔASV can be either positive or negative but must sum to zero. It is not necessary for the advancing and receding events to be concerted, as we have demonstrated here. They can be very different processes with very different activation energies. Since advancing events might induce receding events or vise versa, there ought to be multiple different activation energies for synchronous or sequential events that occur around the perimeter of the contact line by various mechanisms. This suggests a complex relationship between the activation energy and contact angle hysteresis. In this vein, the liquidsolid contact shape will depend upon the relative rates of advancing and receding events occurring at the perimeter. The experiments that we have conducted through our setup enable an overall resisting force, which is more useful practically, to be possibly associated with the liquidsolid contact shape and the relative rates of advancing and receding events. At the other end of the interaction event, the experiments conducted with lower surface translation speeds shed light on the manner of liquid detachment from the surface. Despite the movement of the surface, the video footage did not reveal the formation of daughter droplets that normally accompany a strong momentum change in liquids.21,22 In fact, the behavior was more of one in which the liquid appeared to be stretched across the two surfaces followed by a relatively slow rupture of this liquid bridge at close proximity of the superhydrophobic surface end (see sequence in Figure 9). The rupture behavior of axially loaded liquid bridges has been well studied,23,24 and situations where liquid is dynamically introduced into the bridge as a continuous flow25 or more recently by simple drawing from a capillary26 have been reported. The slowly stretched liquid bridge rupture mechanism here appeared to infer the presence of a thin film of liquid forming on the superhydrophobic surface at all times to provide an opposite anchor for the liquid bridge. Due to the high contact angles at the point of detachment and surface movement, it was

difficult to ascertain if there was any directional peeling effect from the advancing to the receding contact line end. Such behavior would be expected if the advancing contact line was always the first to be breached. Nonetheless, the stretched behavior of the bridge suggests that the moving superhydrophobic surface was serving almost like an invisible conveyor draining out liquid from the bridge through adhesion to the micro and nano roughness of the surface (supporting the notion of a Wenzel wetting mode) to engender the liquid bridge above to be increasingly unstable until rupture is finally achieved. We posit that the relatively low wetting adhesion of the superhydrophobic surface permits the bridge to retain its form rather stably, despite some perturbation from the moving surface, until the point of its rupture. On a side note, one can draw some parallels of such a mechanism to the insightful investigations of evaporation affecting the stability of microcapillary bridges.27 Such small-scale bridges have important implications on the use of atomic force microscope tips to probe surfaces.28 Finally, we offer some caveats when applying the dangling foil optical lever setup for measurement. As the translator was driven by a stepping mechanism, the signal will naturally be overlaid by a periodic carrier. A low-pass filtering operation was found to remove this periodic carrier effectively. We found that it was necessary to locate the flexible foil to coincide with the axis of the rod. Failure to do this would engender rotation of the rod which affected accuracy. Lastly, care should be taken to avoid any crease from forming on the foil during its preparation. The presence of creases will violate the behavior of the foil as a cantilevered continuous beam.

’ CONCLUSIONS In summary, we have devised a dangling cantilever optical lever setup that allows dynamical studies of superhydrophobic surfaces without the effects of gravitational acceleration for better insight into the mechanisms at play. The setup enabled us to ascertain liquid loss and ascribe it to the interaction of liquid that just touched the superhydrophobic surface as it translated at a constant lateral speed. At lower surface translation speeds (2060 μm/s), the interactions were characterized by an strong initial liquid pin (at up to 0.6 nN force) and depin followed by a series of smaller force pin and depin processes before there was enough liquid loss to lead to total liquid detachment from the surface. At higher translation speeds (80100 μm/s), the interactions were characterized by liquid pin and depin processes at a sustained force (around 0.7 nN) in which the liquid loss was smaller, which would lead to a much later liquid detachment 457

dx.doi.org/10.1021/la203732g |Langmuir 2012, 28, 453–458

Langmuir

ARTICLE

(beyond 100 s). A transitional regime (at around 70 μm/s) showed a competitive mode between the ability to repin on the surface and the loss of this ability due to liquid loss due to interaction. The findings here impute a stronger influence by a Wenzel as opposed to a Cassie wetting mechanism. We have also been able to observe the evolution of contact angles which showed a linear reduction in the receding contact angle with time, but not with the advancing contact angle, up to the point of liquid depin. This finding suggests a stronger role played by the receding contact line in establishing liquid adherence to the superhydrophobic surface. The detachment process from the surface was also characterized by a liquid bridge driven to rupture by way of liquid being conveyed away from the bridge. The relatively low wetting adhesion to the superhydrophobic surface appeared to permit the bridge to retain its form rather stably until the point of rupture.

(19) Krumpfer, J. W.; Bian, P.; Zheng, P.; Gao, L.; McCarthy, T. J. Langmuir 2011, 27, 2166. (20) Gao, L.; McCarthy, T. J. Langmuir 2009, 25, 14105. (21) Hsu, C. F.; Ashgriz, N. Phys. Fluids 2004, 16, 1637036. (22) Hong, J. S.; Lee, B. S.; Moon, D.; Lee, J.-G.; Kang, I. S. Electrophoresis 2010, 31, 1357. (23) Fan, H.; Wang, G. F. J. Appl. Phys. 2003, 93, 2554. (24) Chen, Q. S.; Hu, W. R. Chin. Phys. Lett. 1999, 16, 822. (25) Lowry, B. J.; Steen, P. H. J. Colloid Interface Sci. 1995, 170, 38. (26) Schwalb, W.; Ng, T. W.; Lye, J. K. K.; Liew, O. W.; Cheong, B.H.-P. J. Colloid Interface Sci. 2012, 365, 314. (27) Maeda, N.; Israelachvili, J. N.; Kohonen, M. M. Proc. Natl. Acad. Sci 2003, 100, 803. (28) Wei, Z.; Zhao, Y.-P. J. Phys. D: Appl. Phys. 2007, 40, 4368.

’ ASSOCIATED CONTENT

bS

Supporting Information. Videos of Figures 7 and 9. This material is available free of charge via the Internet at http://pubs. acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT Portions of this work were made possible by support from the Australian Research Council Grant DP0878454, and the Monash NSMRF and ESG schemes. The technical assistance rendered by WYL Ling in preparing the superhydrophobic sample, and J. Fu in obtaining SEM images of the superhydrophobic surface are appreciated. ’ REFERENCES (1) Gao, X.; Jiang, L. Nature 2004, 432, 36. (2) Barthlott, W.; Neinhuis, C. Planta 1997, 202, 1. (3) Yu, Y.; Zhao, Z.; Zheng, Q. Langmuir 2007, 23, 8212. (4) Wu, S.-Z.; Wang, J.-N.; Niu, L.-G.; Yao, J.; Wu, D.; Li, A.-W. Appl. Phys. Lett. 2011, 98, 081902. (5) Wang, F.-C.; Yang, F.; Zhao, Y.-P. Appl. Phys. Lett. 2011, 98, 053112. (6) Guo, Z.; Liu, W. Appl. Phys. Lett. 2010, 97, 243701. (7) He, M.; Li, H.; Wang, J.; Song, Y. Appl. Phys. Lett. 2011, 98, 093118. (8) Ling, W. Y. L.; Lu, G.; Ng, T. W. Langmuir 2011, 27, 3233. (9) Liu, M.; Jiang, L. Adv. Funct. Mater. 2010, 20, 3753. (10) Koishia, T.; Yasuoka, K.; Fujikawa, S.; Ebisuzaki, T.; Zeng, X. C. Proc. Natl. Acad. Sci. 2009, 106, 8435. (11) Hao, P.; Lv, C.; Yao, Z.; He, F. EPL 2010, 90, 66003. (12) Sakai, M.; Kono, H.; Nakajima, A.; Zhang, X.; Sakai, H.; Abe, M.; Fujishima, A. Langmuir 2009, 25, 14182. (13) Beer, F. P.; Johnston, E. R.; DeWolf, J. T. Mechanics of Materials; McGraw Hill: Singapore, 2006. (14) Panduputra, Y.; Ng, T. W.; Neild, A.; Ling, W. Y. L. Opt. Lett. 2011, 36, 175. (15) Panduputra, Y.; Ng, T. W.; Neild, A.; Robinson, M. Appl. Opt. 2010, 49, 3669. (16) Ling, W. Y. L.; Ng, T. W.; Neild, A.; Zheng, Q. J. Colloid Interface Sci. 2010, 354, 832. (17) Ling, W. Y. L.; Ng, T. W.; Neild, A. Langmuir 2010, 26, 17695. (18) Boreyko, J. B.; Chen, C.-H. Phys. Rev. Lett. 2009, 103, 174502. 458

dx.doi.org/10.1021/la203732g |Langmuir 2012, 28, 453–458