Turning a surface superrepellent even to completely wetting liquids

See allHide authors and affiliations

Science  28 Nov 2014:
Vol. 346, Issue 6213, pp. 1096-1100
DOI: 10.1126/science.1254787


Superhydrophobic and superoleophobic surfaces have so far been made by roughening a hydrophobic material. However, no surfaces were able to repel extremely-low-energy liquids such as fluorinated solvents, which completely wet even the most hydrophobic material. We show how roughness alone, if made of a specific doubly reentrant structure that enables very low liquid-solid contact fraction, can render the surface of any material superrepellent. Starting from a completely wettable material (silica), we micro- and nanostructure its surface to make it superomniphobic and bounce off all available liquids, including perfluorohexane. The same superomniphobicity is further confirmed with identical surfaces of a metal and a polymer. Free of any hydrophobic coating, the superomniphobic silica surface also withstands temperatures over 1000°C and resists biofouling.

Undercutting the surface keeps liquids at bay

The shape of an umbrella is designed both to protect the holder from falling droplets and to have the collected rainwater flow away from the person underneath. Liu and Kim exploited the idea of an umbrella to make materials with a surface that repels almost any liquid.

Science, this issue p. 1096

The ability to understand the extraordinary liquid repellency of natural surfaces (1, 2) has affected a wide range of scientific and technological areas, from coatings (3), heat transfer (4), and drag reduction (5) to biomimetics (6). Whereas the wetting-resistant surfaces developed since the 1960s (710) used only surface roughness to trap gas with no interest in the apparent contact angles, superhydrophobic surfaces created since the late 1990s (1, 11, 12) combined the roughness with a hydrophobic material to superrepel water—that is, to display a very large apparent contact angle (θ* > 150°) and a very small roll-off angle (θroll-off < 10°). For low-energy liquids such as oils or organic solvents, a roughness with overhanging topology was necessary to make the hydrophobic material superoleophobic (13, 14), omniphobic (15), or superomniphobic (16, 17). Despite the use of the prefix “omni-” (6, 1518), however, no natural or man-made surface has been reported to repel liquids of extremely low surface tension or energy (i.e., γ < 15 mJ/m2), such as fluorinated solvents, which completely wet existing materials (10, 1921). Departing from the prevailing approach of roughening a hydrophobic material, we first propose that the material’s inherent wettability, depicted by the intrinsic contact angle θY, is irrelevant when dealing with a completely wetting liquid (θY = 0°). Focusing instead solely on the roughness details, we develop a surface that superrepels all available liquids, including fluorinated solvents—for instance, perfluorohexane (C6F14; namely, 3M Fluorinert FC-72), whose surface energy (γ = 10 mJ/m2) is the lowest known and which has never been observed to bead up, let alone roll off, on any surface.

To avoid the confusion with the petal effect (22), in which a droplet with large contact angles sticks to the surface, it helps to first clarify that “repelling” means that droplets not only bead on but also roll off the surface. To repel (i.e., θ* > 90° with a small θroll-off) or superrepel (θ* > 150° with θroll-off < 10°) a wetting liquid (θY < 90°) on a structured surface, two conditions must be met: (i) a successful liquid suspension on the roughness and (ii) a low liquid-solid contact. The microstructures should first be able to suspend the liquid, supporting a composite interface proposed by Cassie and Baxter (23). Once suspended, decreasing the liquid-solid contact would increase θ* and reduce θroll-off, hence increasing the repellency. In the rare cases of a highly wetting (θY < 10°) liquid beading (i.e., successful suspension and θ* > 90°) on a structured surface—for example, water on SiO2 (13, 24) or hexane on nickel (16)—the liquid stuck on the surface rather than rolled off; these surfaces are not considered to repel the liquid, despite the beading.

Liquid suspension by surface structures (or resisting liquid wetting by surface topologies with characteristic length smaller than the liquid’s capillary length) was proposed in the 1960s with θY as a critical parameter (8, 10). For θY > 90°, as in water or aqueous solutions on a hydrophobic material, a simple microstructure (Fig. 1A) would suspend the liquid to a Cassie state (1, 2, 8, 1012) (figs. S1 and S2). For θY < 90°, as in oils and organic solvents on a hydrophobic material or water on a slightly hydrophilic material, a reentrant microstructure (Fig. 1B) is required to suspend the liquid and resist it from wetting into the cavity (3, 8, 10, 1318, 25). From simple force balance, the reentrant topology of Fig. 1B would suspend a liquid even with θY ~ 0° in the absence of any positive liquid pressure. However, there is always a pressure in reality (e.g., hydrostatic, Laplace, or environmental perturbation), and once pushed into the cavity the liquid spreads spontaneously. So far, the most successful suspension was for liquids with surface energy as small as γ ~ 15 mJ/m2 [i.e., pentane (15, 16) and isopentane (16)], leaving many fluorinated solvents unresolved.

Fig. 1 Liquid suspension on surface structures of three different topologies.

(A) Simple structures require θY > 90° to suspend water. Δp, is the pressure difference between the liquid and air. (B) Reentrant structures allow θY < 90° to suspend oils or solvents. They would fail if θY ~ 0°, as surface tension acts parallel to the horizontal overhangs with little vertical component to suspend the liquid. (C) Doubly reentrant structures allow θY ~ 0° to suspend any liquid, as surface tension acts on the vertical overhangs with a substantial vertical component. If the liquid-solid contact fraction is small enough, the surfaces would also repel the liquids.

In addition to the reentrant microstructure, it has long been hypothesized that surface structures of a doubly reentrant topology (Fig. 1C) might provide a stronger resistance against wetting and retain suspension, even if θY ~ 0° (810, 15, 26). The mechanism of suspending such a completely wetting liquid on a doubly reentrant microstructure is reasoned as follows (see Fig. 1C): Upon contacting the surface, the liquid would wet the top surface and continue down along the sidewall of the vertical overhangs. The liquid would stop advancing at the bottom tip of the vertical overhangs, where the surface tension can start to point upward. Though this concept of suspending even highly wetting liquids on a doubly reentrant topology has been known (810, 15, 26) and confirmed with water (24), the liquid-solid contact fraction should be sufficiently low for the resulting surface to not only suspend but also repel the liquid. A highly wetting (θY < 10°) liquid suspended on the microstructures would still spread (i.e., θ* < 90°) on the composite surface unless the liquid rests mostly on air. To understand how far we are from being able to repel the highly wetting liquids, let us assess the contribution of air to the repellency.

The apparent contact angle θ* for a suspended droplet (i.e., in a Cassie state) is described by the Cassie-Baxter model (23) as cosθ* = fscosθYfg (1)where fs is the liquid-solid contact fraction [or “solid fraction” for short; i.e., the proportion of liquid-solid contact area (including the wetted regions inside the roughness) to the projected area of the entire composite interface], fg is the gas fraction similarly defined for liquid-vapor interface, and fs + fg ≥ 1 (27). If the liquid-solid and liquid-vapor interfaces are perfectly flat, neglecting any solid roughness and meniscus curvature—that is, the ideal Cassie state with fs + fg = 1—Eq. 1 simplifies tocosθ* = fs(1 + cosθY) – 1 (2)Although valid only for the ideal Cassie state, Eq. 2 allows us to qualitatively explore the relation between θ*, fs, and θY. In addition to the widely appreciated consequence that θ* can be greatly increased as fs decreases, we examine the role of the intrinsic contact angle θY by plotting Eq. 2 with θY as a parameter in Fig. 2. As is evident, the difference between the θ* values of a large θY and a small θY decreases as fs decreases. In other words, by minimizing fs the contribution of the material’s inherent nonwettability (described by the magnitude of θY) on the liquid repellency (described by the magnitude of θ*) diminishes. This diminishing trend suggests that a structured surface may repel extremely wetting liquids if fs is very small. For example, even a completely wetting liquid (θY = 0°) may be superrepelled (θ* > 150°) if fs < 6%. However, it should not be forgotten that this argument is valid only for the Cassie state (i.e., the suspended state), which is exceedingly difficult to achieve if fs becomes very small. Even for the reentrant topology of Fig. 1B, the suspension force becomes too small before fs becomes small enough to repel liquids with very small θY. This difficulty explains why superrepellency has been shown for liquids with surface tensions above ~20 mN/m (13) but not for those with surface tension values of 15 to 20 mN/m, such as pentane (15, 16) and isopentane (16), which have been suspended but not repelled.

Fig. 2 Relation between apparent contact angle θ* and solid fraction fs for ideal Cassie-state droplets with intrinsic contact angle θY as a parameter.

As fs decreases, the band of lines narrows, indicating that the influence of θY on θ* diminishes. If fs is below 6%, θ* is above 150° even if θY ~ 0°. The green, red, and blue lines represent the inherent wettability: nonwetting (e.g., water on a hydrophobic surface), moderately wetting (e.g., solvent on a hydrophobic surface), and highly wetting (e.g., fluorinated solvents on any surface or most liquids on clean SiO2), respectively. The three bold lines correspond to the three cases shown in Fig. 1.

From Figs. 1 and 2, one can now reason that a structured surface may repel any liquid if the microstructures are doubly reentrant and also of a low enough solid fraction. However, common doubly reentrant shapes in the literature (8, 10, 15, 18, 26) produce only a weak suspension and a moderate solid fraction insufficient to repel highly wetting liquids. To suspend completely wetting liquids on a surface with a minimal solid fraction, an entire surface should be uniformly covered with doubly reentrant structures having vertical overhangs as thin, vertical, and short as possible. As illustrated in Fig. 1C, such an ideal doubly reentrant structure minimizes the break-in force by the liquid pressure that wets the cavity and maximizes the surface tension force that suspends the liquid against wetting (eq. S1) (9). The thin and vertical geometry of the overhangs minimizes their projected area added to the solid fraction, and the shortness of the overhang keeps the increase of the solid fraction by the vertical surfaces at bay. Some superhydrophobic or superoleophobic surfaces described in the literature attempted to incorporate doubly reentrant structures but with little success. For example, only a few doubly reentrant structures were formed among predominantly simple or reentrant structures prone to wetting (3, 14, 25), and structures barely satisfying the doubly reentrant shape were replicated from springtail skin with only a moderate solid fraction (18).

To fulfill all the requirements reasoned above and quantified from Fig. 2, we designed a surface illustrated in Fig. 3A: an array of doubly reentrant structures consisting of microscale posts with nanoscale vertical overhangs. Posts were chosen over ridges or holes to minimize fs more easily. Also, the post array allows the air underneath the droplet to remain connected to the atmosphere so that the liquid is suspended only by surface tension and is not assisted by the pressure of the trapped air. We chose to form the surface structures from SiO2 for the following two reasons: First, clean SiO2 is highly wetted (i.e., θY < 10°) by most liquids (except liquid metals such as mercury), including water (1, 20). Because roughening of a SiO2 surface is supposed to amplify the liquid affinity to complete wetting (1), structuring a SiO2 surface to repel liquids should provide a stark contrast to the existing approach. Second, Si micromachining provides sophisticated equipment and techniques to process SiO2. With precisely controlled thermal oxidation of a shallow-etched silicon surface followed by three sequential etching steps on SiO2 and Si (fig. S3E) (28), we successfully fabricated a SiO2 surface (1.7 cm by 1.7 cm) with close-to-ideal doubly reentrant structures (Fig. 3, B to E). The inclined angle of the vertical overhang is measured to be ~85 ± 1° (Fig. 3E), providing a maximum suspension force that is 99.6% of the perfectly vertical overhang shown in Fig. 1C. In spite of the overall resemblance between the microposts in Fig. 3B and those of superoleophobic surfaces (13, 1517), it is the close-to-ideal nanoscale vertical overhangs in Fig. 3, C to E, that lead to an unprecedented liquid-repellency.

Fig. 3 Design and fabricated results of the SiO2 surface.

(A) Designed surface of microposts with doubly reentrant nano-overhangs. As key geometric parameters, D is the post top diameter, P is the center-to-center distance (i.e., pitch) between adjacent posts, and δ and t are the length and thickness, respectively, of the vertical overhang. To make fs small enough (fs < 6%), δ and t should be shrunk to extreme values. (B to E) Scanning electron micrographs of the fabricated surface. (B) Top-angled view of the square array of circular posts with D ~ 20 μm, P = 100 μm, δ ~ 1.5 μm, and t ~ 300 nm, resulting in fs ~ 5%. (C) Bottom-angled view of one post. (D) Cross-sectional view of one post. (E) Magnified cross-sectional view of the vertical overhang. Note the similarity with the ideal topology of Fig. 1C.

To evaluate the liquid repellency, we chose 14 different liquids (table S1) (29), including water, ionic liquid, acid, oils, and numerous polar or nonpolar organic or fluorinated solvents with surface tensions ranging from 72.8 mN/m (i.e., water) to the lowest known value of 10 mN/m (i.e., FC-72). As expected, a smooth SiO2 surface was highly wetted (θ* = θY < 10°) by all of the liquids (table S2). In contrast, our structured SiO2 surface successfully suspended and repelled all of the tested liquids, beading them into Cassie-state droplets (water, methanol, and FC-72 shown in fig. S4A and movie S1) and letting them roll around (fig. S4B and movies S2 and S3); that is, behaving superomniphobic in air.

To quantify the repellency of our surface, we measured the advancing and receding contact angles (Fig. 4A) and roll-off angles (fig. S4C) of all 14 liquids. For comparison, Fig. 4A also includes the other two liquid-repellent surfaces analyzed in Fig. 1: a superhydrophobic surface consisting of cylindrical posts (Fig. 1A and fig. S3B) and a superoleophobic surface consisting of posts with reentrant overhangs (Fig. 1B and fig. S3D), both of which were coated with a hydrophobic layer of C4F8. As expected, although the superhydrophobic surface with vertical posts could not suspend liquids with surface tension below ~40 mN/m, the superoleophobic surface with reentrant posts repelled liquids with lower surface tension (20 to 40 mN/m). However, liquids with even lower surface tension (<20 mN/m) could not be suspended, as they wicked between the reentrant posts. In contrast, on the surface with doubly reentrant posts, all 14 liquids formed large contact angles, even without any hydrophobic coating.

Fig. 4 Omniphobicity of the structured SiO2 surface.

(A) Apparent advancing (θ*A) and receding (θ*R) contact angles of the 14 liquids measured on three liquid-repellent surfaces: our omniphobic surface and two control surfaces of the same nominal solid fraction (fs ~ 5%). Data on the omniphobic surface are depicted by blue circles (solid and hollow); data on the control surfaces with reentrant and vertical topologies are represented by orange triangles and green squares, respectively. Each data point is an average of more than 100 measurements. Error bars are omitted here for clarity and are instead shown in fig. S6. (B) Relations of contact angles on a smooth surface (cosθY) and on a structured surface (cosθ*). The theoretical relations from the Wenzel and Cassie-Baxter models are plotted as solid black lines. Data near (1,–1) and (1,1) are shown in the zoomed-in boxes, revealing the difference between our structured SiO2 surface and the control surfaces, especially when liquids highly wet the material. (C) Robust repellency of the structured SiO2 surface demonstrated by bouncing FC-72 off the super omniphobic SiO2 surface with doubly reentrant posts of D ~ 10 μm, P = 50 μm, δ ~ 920 nm, t ~ 270 nm, and fs ~ 5% under Weber number ~0.42.

The extent to which wettability is modulated by surface roughness is shown in Fig. 4B, where the apparent wettability (cosθ*) is plotted as a function of the inherent wettability (cosθY). Data from our surface with doubly reentrant posts (i.e., blue circles) were populated at the lower right corner in the fourth quadrant near point (1,–1), exhibiting the exceptional ability to render a highly wettable material superrepellent. In contrast, although the two control surfaces permit a Cassie state with nonwettable or partially wettable material, they got soaked when the material was highly wetted by the liquids of very low surface tension (i.e., hexane and six fluorinated solvents), displaying θ* ~ 0° with data populated near point (1,1). These results are consistent with the theory schematically summarized in fig. S2.

In addition to repelling all 14 liquids (movie S2), our superomniphobic surfaces are also expected to sustain static (fig. S5 and movie S4) and dynamic (movie S5) pressures better than the existing superhydrophobic and superoleophobic surfaces (26). The doubly reentrant structures allow droplets to bounce on even extremely sparse posts (i.e., tens of micrometers of pitch and a solid fraction of only ~5%). With high-speed imaging, water, methanol, and FC-72 droplets were confirmed to bounce off the truly superomniphobic SiO2 surfaces (movie S5). Water (γ = 72.8 mN/m) and methanol (γ = 22.5 mN/m) droplets rebounded on a surface with microposts of 100-μm pitch; this pitch was much larger than those reported in the literature (3, 15). However, FC-72 (γ = 10 mN/m) droplets penetrated and wetted the above surface at impact. A surface with uniformly halved structures (i.e., fs remaining at ~5%) was further prepared to provide enough resistance against impalement and let FC-72 droplets rebound, as shown with snapshots in Fig. 4C. While resisting the physical intrusion of liquids, this surface has no defense against some other intrusion mechanisms such as condensation inside the cavity. The internal condensation would be a common issue to all existing superhydrophobic and superoleophobic surfaces (1), calling for a provision (5) for practical use.

Because the proposed superrepellency depends only on physical attributes, we further fabricated metal (i.e., tungsten) and polymer (i.e., parylene) counterparts based on the given SiO2 surface and confirmed that they possess the same superrepellency as expected (fig. S4A). The ability to repel fluorinated solvents may allow the electronic circuits to be cooled by nucleate boiling (i.e., the most efficient mode of cooling (4). Free of polymeric coating, the superomniphobic SiO2 surface can serve at high temperatures. The surface was found unaltered after storage at >1000°C and was used to demonstrate rolling-off of another FC liquid at 150°C and a nonvolatile liquid at >320°C (fig. S8 and movie S6). The polymer-free parts are expected to last longer in outdoor environment, where polymeric materials tend to degrade faster. Unaffected by the surface chemistry, the superomniphobic SiO2 surface also demonstrated prolonged repellency to biological fluids (such as the sheep serum we tested), whereas a regular superhydrophobic surface lost the repellency (fig. S9 and movie S7).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S9

Tables S1 and S2

References (3036)

Movies S1 to S7

References and Notes

  1. This general definitions of fs and fg follow Cassie and Baxter’s original paper (23), which included all of the nonflat (e.g., rough, curved) effects on the liquid-solid and liquid-vapor interface. In addition to the most simplified version of flat liquid-solid and flat liquid-vapor interfaces, which results in fs + fg = 1, a less simplified version of nonflat liquid-solid and flat liquid-vapor interfaces is often adopted in the literature.
  2. Materials and methods are available as supplementary materials on Science Online.
  3. These liquids are commonly used for applications such as electrochemistry, fuel cells, integrated circuits fabrication, microfluidic systems, heat transfer, etc.
  4. Acknowledgments: C.-J.K. was encouraged by D. Attinger to start this research. T.L. acknowledges W. Choi and K. Ding for discussion of the fabrication, L.-X. Huang for assistance with high-speed imaging, and K. Shih for help with roll-off angle measurements. C.-J.K. and T.L. thank an anonymous referee for advice on the biofouling test; D. Di Carlo and O. Adeyiga for biofluid selection; and B. Dunn, R. Freeman, and S. Chen for manuscript preparation. The data reported in the paper are tabulated in the supplementary materials. C.-J.K. and T.L. have filed a patent on this work (“Liquid-repellent surface made of any materials,” International Application no. PCT/US2014/57797).
View Abstract

Navigate This Article