Article Cite This: Langmuir 2018, 34, 4173−4179
pubs.acs.org/Langmuir
How to Achieve Reversible Electrowetting on Superhydrophobic Surfaces Michail E. Kavousanakis,†,§ Nikolaos T. Chamakos,†,§ Kosmas Ellinas,‡ Angeliki Tserepi,‡ Evangelos Gogolides,‡ and Athanasios G. Papathanasiou*,† †
School of Chemical Engineering, National Technical University of Athens, Athens 15780, Greece Institute of Nanoscience and Nanotechnology, NCSR “Demokritos”, Aghia Paraskevi 15341, Greece
‡
Downloaded via UNIV OF THE SUNSHINE COAST on June 27, 2018 at 11:12:28 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Collapse (Cassie to Wenzel) wetting transitions impede the electrostatically induced reversible modification of wettability on superhydrophobic surfaces, unless a strong external actuation (e.g., substrate heating) is applied. Here we show that collapse transitions can be prevented (the droplet remains suspended on the solid roughness protrusions) when the electrostatic force, responsible for the wetting modification, is smoothly distributed along the droplet surface. The above argument is initially established theoretically and then verified experimentally.
■
impaled by the solid roughness).16 The latter corresponds to a local minimum of the free-energy landscape, suggesting that an energy barrier has to be surpassed for a reverse dewetting transition. 17,18 In particular, Cassie and Wenzel states correspond to (meta)stable steady states (local minima of the free-energy landscape), which are separated by an intermediate unstable (experimentally intractable) steady state. Then, the energy barrier to induce a wetting or dewetting transition between the two (meta)stable states is determined by subtracting their free energy from the unstable state (saddle) that separates them. Several attempts have been made to prevent collapse transitions (Cassie to Wenzel) on structured surfaces, induced by electrowetting. In particular, Lapierre et al.19 achieved a moderate contact angle reversibility interval (160−130°) by using specially designed substrates covered by two layers of packed silicon nanowires. Dense and tall structures seem to help prevent Cassie to Wenzel transitions and increase the reversibility range. However, such structures feature poor mechanical properties, rendering them unsuitable for various applications. McHale et al.20 have demonstrated the use of liquid marbles, where hydrophobic silica particles and grains of lycopodium are used as a conformable skin on the water droplet in order to achieve reversible electrowetting. However, reversibility is guaranteed only over a limited range of applied voltages (up to approximately 200 V) and a contact angle modification not higher than ∼10°. At higher voltages, collapse transitions were encountered, and if the voltage was increased too high, the marble burst. A fully reversible wettability
INTRODUCTION Droplets on microscopically structured surfaces can exhibit fascinating properties, with an antagonistic impact, including wetting states featuring very high (>150°) or very low ( 150°), we selected surfaces with high impalement resistance (in terms of critical hydrostatic or hydrodynamic pressure for preserving Cassie states) and low contact angle hysteresis. Low CA hysteresis is important for reversibility even in the Cassie state, as our computations suggest. High impalement resistance is important in order to preserve the Cassie state when high voltage is applied, in order to achieve a large CA modification. And since it is expected that electrowetting easily induces Cassie to Wenzel transitions, we selected an exceptionally robust surface type in terms of the impalement resistance and low CA hysteresis, and we made substrates with different dielectric thicknesses. Such a surface type consists of submicrometer (quasi-)ordered pillar arrays fabricated by means of colloidal lithography and plasma etching. (See Figure 1a for a scanning
Figure 1. (a) SEM image of the solid surface topography displaying submicrometer (quasi-)ordered pillar arrays fabricated on the 1.4 μm (400 nm TEOS + 1 μm PMMA), 12 μm, 57 μm, and 75 μm (SU8) dielectric layers. The pillars’ typical height, width, and spacing (pitch) are 600, 125, and 300 nm, respectively. (b) SEM image of the random micro/nano-textures fabricated on the 188 μm COP film. electron micrograph of the topography.) These structures exhibit extreme resistance against droplet impalement, reaching 30 atm for water-based mixtures.31 The fabricated dielectric thicknesses range between 1.4 and 75 μm. The dielectrics underneath the surface structures are stacks consisting of tetraethoxysilane (TEOS), poly(methyl methacrylate) (PMMA), or SU8 photoresists. In particular, we fabricated the following: a thin stack featuring 400 nm tetraethoxysilane (TEOS) and a 1 μm poly(methyl methacrylate)
(4) 4175
DOI: 10.1021/acs.langmuir.7b04371 Langmuir 2018, 34, 4173−4179
Article
Langmuir
Figure 2. (a) Apparent contact angle vs electrowetting number, η, for a water droplet on a pillared surface with a 5 μm dielectric thickness. Dashed lines (blue and red) depict unstable equilibrium states. The dashed gray line is drawn as a guide to the eye, showing the ACA variation by increasing the applied voltage. (b) Enlarged bifurcation diagram showing the coexistence of Cassie, unstable partially penetrating, and Wenzel states. (c) Apparent contact angle dependence on the electrowetting number for the thick dielectric case. The gray shaded area illustrates the region of fully reversible transitions. Dashed lines depict unstable equilibrium wetting states. (d) Droplet free surface and electric potential distribution (dimensionless) at the gap formed by the surface pillars for the two different dielectric thicknesses and the same electrowetting number (η = 0.884). The radius of curvature, Rc, for the thin dielectric case is approximately 4 times lower than that for the thick dielectric case.
■
(PMMA) layer and three stacks featuring 12-, 57-, and 75-μm-thick layers of the SU8, respectively. Finally, in order to show that even a surface structure with a moderate resistance to impalement but with low contact angle hysteresis can perform reversible electrowetting if the dielectric thickness is large enough, we prepared random dual-scale micro/nano-textured topography on a 188-μm-thick cyclic olefin copolymer (COP) film32 (see Figure 1b). Details of the fabrication method are included in the Supporting Information. The apparent contact angle of a water droplet (at zero voltage) is 158° for the submicrometer (quasi-)ordered surface and 156° for the random micro/nano-textured COP surface, respectively. The contact angle hysteresis is ∼5° for both surfaces.31,32 The apparent contact angle, θapp, was measured using real-time image processing software that was developed in house and is based on the drop shape analysis technique.9
RESULTS AND DISCUSSION Computational Findings. Our computational predictions here concern the equilibrium of droplets sitting on multistriped dielectric surfaces of different thicknesses. We chose orthogonal stripes with rounded edges because this is a simple and representative geometry that can be used to highlight the effects of the dependence of the liquid surface curvature on the dielectric thickness. Trapezoidal or inverse trapezoidal geometries (i.e., with overhangs) could be also used, like the ones used in Kavousanakis et al.,18 without affecting the qualitative features of our results. The optimization of the surface geometry for improved impalement resistance is beyond the scope of the current work. Striped surfaces (two-dimensional geometries) instead of pillared surfaces (three-dimensional geometries) are considered since the computational requirements for 3-D geometries could become prohibitive. Notice 4176
DOI: 10.1021/acs.langmuir.7b04371 Langmuir 2018, 34, 4173−4179
Article
Langmuir
liquid viscosity and surface oscillations due to noise, which are not accounted for here. What is predicted in our diagram is the possible configurations that the droplet can attain in equilibrium. In the case of the critical, η = 0.884, plenty of Wenzel states can be observed with ACA from 120 to 90°. Such behavior is usual in experiments where one can observe various Wenzel states, even with a relatively high ACA. Another important aspect of the diagram in Figure 2a is that the solution families corresponding to Cassie or Wenzel states (i.e., blue or red zigzag-like curves cross the zero-η vertical axis without featuring turning points on their left). Such a solution space configuration denotes infinite hysteresis with decreasing η. Practically, this means that for the thin dielectric case, electrowetting can induce spreading but not receding (i.e., dewetting) even for the Cassie states. Such a solution space structure implies pinning on the tops of the structures in the Cassie state. This pinning, due to infinite hysteresis, is also calculated for the thick dielectric case (the non-gray-shaded part of Figure 2c). In order to achieve receding (i.e., reversibility), the existence of turning points on the left part of the solution curves is required and, in particular, positive values of η. This requirement is fulfilled in the thick dielectric case. In Figure 2c, one can observe two important features: (a) only Cassie states are found and (b) there is a distinct region, shadowed with gray, where the wetting modification is reversible. The corresponding range of reversibility is between 121 and 78° (corresponding to points P2 and P1, respectively). The range of reversibility, obtained by the structure of the solution space shown in Figure 2a,c, depends on the thickness of the dielectric and the surface topography. A thorough parametric study of finding the critical thickness in order to obtain a certain range of reversibility, as a function of the geometry of surface features, is beyond the scope of this article. The exclusive existence of Cassie states can also be attributed to the invariability of the curvature (or the radius of curvature, Rc) of the liquid surface at the contact with the solid surface, as Figure 2d shows. It is worth comparing the curvature radii of the thick and the thin dielectrics for the same electrowetting number, η = 0.884. One can see that Rc for the thick dielectric case (picture shown on the right in Figure 2d) is almost 4 times larger than that of the thin dielectric (picture shown on the left in Figure 2d); notice that the apparent contact angle for the thick dielectric is lower. The electric potential distribution, u, at the gap formed by the surface pillars, which is depicted in Figure 2d, also exhibits significant differences between the two dielectric cases. Note that our computational methodology is not based on any simplifications regarding the actual shape of the droplet and the field distribution. Experimental Results. Electrowetting experiments were performed on each surface using 7.5 μL water droplets, and the results are presented in Figure 3, where the dependence of the
that in our computations we compute the entire droplet shape along with the coupled field distribution assuming only translational symmetry in the direction normal to the xy plane. In particular, we find the following: 1. A thin dielectric case, where the dielectric material covers the trenched surface with a uniform thickness of 5 μm. 2. A thick dielectric case, where the surface protrusions are on top of a dielectric slab of 100 μm. In this case, the surface protrusions are also considered to be dielectric with the same dielectric strength (i.e., the thickness of the dielectric ranges from 100 to 100 + h μm, where h is the protrusion height). The height and width of the protrusions are 31 and 18.6 μm, respectively, and the gap between protrusions is 18.6 μm wide. The dielectric in both cases is SiO2 with ed = 3.8. As a top hydrophobic coating, we consider a material with a modified Young’s angle of θMY = 140°, which corresponds to wLS = 3.089 (eq 3). Such a high contact angle can be achieved by a secondary roughness scale (at the nanoscale). The chosen liquid is water in air ambient, and the corresponding interfacial tension is γLV = 0.072 N m−1. The characteristic length in our computations is R0 = 0.62 mm (if we consider a droplet with a volume of 1 μL). In Figure 2a,c, we present the dependence of the equilibrium apparent contact angle (calculated by fitting a circle to the water/air interface above the solid surface asperities) on the dimensionless electrowetting number, defined as η =
ϵoϵdV 2 , 2dγLV
where ϵoϵd is the capacitance per unit area. Considering that a d specific electrowetting number results, according to Lippmann’s equation, in the same electrowetting effect for both a thin and a thick dielectric layer (with high capacitance and low voltage in the thin case and low capacitance and high voltage in the thick case, respectively), these two cases can be conveniently compared in Figure 2a,c. The main feature of this diagram is the solution multiplicity which is interpreted as wetting states multiplicity (e.g., for a given η, a considerable number of states (solutions of the governing equations) are computed). Continuous lines correspond to stable states, whereas dashed ones correspond to unstable states. Computed Cassie-type states are marked with blue; Wenzel states are marked with red. It is worth mentioning that for the thin dielectric case (Figure 2a) even a small value of the electrowetting number is enough to realize Wenzel states; for example, for η = 0.25 we compute an extended range of apparent wettabilities, ranging from superhydrophobic Cassie states (ACA > 150°) to hydrophilic Cassie or Wenzel states (ACA < 90°). Starting from a superhydrophobic droplet (e.g., point C in Figure 2a) and gradually increasing η, the ACA decreases as the droplet spreads over an increasing number of surface protrusions. The advancing of the droplet contact line from protrusion to protrusion, as the droplet spreads, is associated with instabilities at specific turning points in the diagram. An important wetting transition described in the diagram of Figure 2a is the Cassie to Wenzel wetting transition; what signals this transition is an instability at the turning point, marked with CW, for η = 0.884. An enlargement of the diagram is shown in Figure 2b. At the turning point, CW, the radius of curvature, Rc, of the droplet free surface becomes sufficiently small (Figure 2d), initiating the collapse transition where the liquid enters the surface protrusions. The resulting ACA modification depends on various parameters that affect the droplet dynamics such as the
apparent CA on the electrowetting number, η =
cV 2 , 2γLV
is shown.
An effective Lippmann equation is fitted (solid line in Figure 3) as a guide to the eye (also see the figure legend for fitting details). 0 cos θapp = cos θapp +η
(6)
θ0app
where is the static CA at zero voltage. The experiments are performed as follows: A dc voltage is applied in a stepwise manner. In each step, the desired voltage is increased linearly in time (on the order of ∼2 s), the resulting equilibrium CA is measured, and then the voltage is switched off. In the next step, 4177
DOI: 10.1021/acs.langmuir.7b04371 Langmuir 2018, 34, 4173−4179
Article
Langmuir
Figure 3. Apparent contact angle vs electrowetting number, η =
Figure 4. (a) Droplet snapshots showing an electrowetting experiment on a thin dielectric layer (400 nm TEOS and 1 μm PMMA). An irreversible wetting transition is observed in this case. (The ACA does not change by removing the applied voltage, 65.3 V or η = 0.59, in (a, iii).) (b) Snapshots demonstrating the contact angle reversibility (from 119 to 156°) when the voltage is switched off from 484 V (η = 0.44) on a 75-μm-thick SU8 substrate. A video clip of the later experiment is included in the Supporting Information (S1).
cV 2 , 2γLV
for a water droplet on structured surfaces featuring different dielectric thicknesses. The red points (400 nm TEOS + 1 μm PMMA, 12 and 57 μm SU8) correspond to irreversible whereas the blue points (75 μm SU8 and 188 μm COP) correspond to the reversible modification of ϵ ϵ wettability. The capacitance per unit area, c = 0d d , is determined by fitting an effective Lippmann curve (eq 6) to the experimental measurements. In particular, c = 2 × 10−5 F m−2 for the thin dielectric substrate (400 nm TEOS + 1 μm PMMA), c = 1.9 × 10−6 F m−2 for the 12 μm SU8, c = 3.5 × 10−7 F m−2 for the 57 μm SU8, c = 3.1 × 10−7 F m−2 for the 75 μm SU8, and c = 7.8 × 10−8 F m−2 for the 188 μm COP film, respectively. A schematic of the experimental setup is presented in the inset.
pinning that limits the wettability modification in this case. In particular, at a certain voltage threshold (η = 0.16 for the COP substrate), the droplet starts to randomly oscillate and finally is ejected from the upper electrode. The above limitation cannot be attributed to the saturation phenomenon3 since the latter usually appears at ∼60° for a water droplet in air ambient. A possible explanation is that the electric stresses along the macroscopic three-phase contact line are asymmetric (the electric charge is not evenly distributed) due to ever present chemical or morphological heterogeneities of the solid structure, thus leading to droplet bulk motion on the superhydrophobic surface. The above justifies the narrower range of wettability modification (from 157 to 138°) in the case of the random micro/nano-textured substrate (COP film), as a result of the larger structure heterogeneity. Such behavior does not appear in the thinner dielectric cases due to the strong pinning of the droplet on the solid structure (liquid impalement). In this case, the droplet does not move even if the surface is tilted by 90°. The studied topographies, one with quasi-ordered pillared structures having extreme impalement resistance and one with random dual-scale roughness with moderate impalement resistance, both having low CA hysteresis, showed considerable EW reversibility when the dielectrics are made thick enough. Thus, we suggest that the dielectric thickness is important in realizing reversible EW on the surface with geometric patterns. Overall, we report that the thickness of the dielectric layer, which controls the action range of the electrostatic pressure,24 is a key factor in fabricating surfaces with fully controllable wettability. This finding, along with our theoretical analysis, explains the reversibility that Lapierre et al.19 reported in electrowetting experiments. Lapierre et al. used very tall silicon nanowires (almost 40 μm in height) which effectively provide a composite (air−silicon) thick dielectric, a suitable choice for reversibility, though the importance of the thickness was not highlighted therein.
the voltage attains a higher value and so on. In each sample, we proceeded by increasing the applied voltage until we observed a CA modification from the previous step (i.e., the experiment is terminated when no CA modification is observed). The red markers in Figure 3 correspond to experiments where the droplet spreads when the voltage increases but does not recede when the voltage is switched off from any value (i.e., no reversibility is observed). Blue markers correspond to experiments where the droplet reversibly retracts back to the initial (zero-voltage) CA when the voltage is switched off, respecting, though, the CA hysteresis (i.e., a small difference from the initial CA is attributed to the inherent CA hysteresis of the surface). We measured that the CA can be modified almost 50° without any sign of reversibility for dielectrics having a thickness below 57 μm, except when we used very low voltages corresponding to η < 2 × 10−2. Droplet snapshots showing the irreversible ACA modification, for the thinnest dielectric layer (400 nm TEOS + 1 μm PMMA), are demonstrated in Figure 4a. Interestingly enough, in the case of thicker dielectrics (75 μm SU8 and 188 μm COP) we observed reversibility with a considerable CA modification depending on the surface topography. In particular, the maximum reversible modification of CA, observed in our experiments, is ∼37° from 156 to 119° in the case of the 75 μm SU8. (See the corresponding droplet snapshots in Figure 4b.) We also note that in the final (reverted) state the droplet rolls off easily by tilting the substrate by only ∼5°, thus the droplet mobility has not been affected. The COP surface (i.e., the one with arbitrary roughness but with a large thickness (188 μm)) performed a reversible CA modification of ∼20°, namely, from 157 to 137°. It is worth mentioning, however, that unlike the thinner dielectric layers (
