Say goodbye to stains


Editor's Introduction

Robust self-cleaning surfaces that function when exposed to either air or oil

annotated by

Did you ever wish that the water you just spilled had not wet your entire pants? Or that your clothes remained clean after that walk on the beach? Lu et al. found an easy yet effective way of making all kinds of materials water-repelling and self-cleaning, including cotton. They combine nanotechnology with a simple adhesive, just as you would find in Scotch tape. The coating even resists scratching and rubbing. Thanks to science, you may soon experience new freedom on your next hike through the woods or walk through the rain.

Paper Details

Original title
Robust self-cleaning surfaces that function when exposed to either air or oil
Original publication date
Vol. 347, Issue 6226, pp. 1132-1135
Issue name


Superhydrophobic self-cleaning surfaces are based on the surface micro/nanomorphologies; however, such surfaces are mechanically weak and stop functioning when exposed to oil. We have created an ethanolic suspension of perfluorosilane-coated titanium dioxide nanoparticles that forms a paint that can be sprayed, dipped, or extruded onto both hard and soft materials to create a self-cleaning surface that functions even upon emersion in oil. Commercial adhesives were used to bond the paint to various substrates and promote robustness. These surfaces maintained their water repellency after finger-wipe, knife-scratch, and even 40 abrasion cycles with sandpaper. The formulations developed can be used on clothes, paper, glass, and steel for a myriad of self-cleaning applications.


Artificial self-cleaning surfaces work through extreme water repellence (superhydrophobicity) so that water forms near spherical shapes that roll on the surface; the rolling motion picks up and removes dirt, viruses, and bacteria (1–3). To achieve near spherical water droplets, the surfaces must be highly textured (rough) combined with extremely low water affinity (waxy) (45). The big drawback of these artificial surfaces is that they are readily abraded (68), sometimes with little more than brushing with a tissue, and readily contaminated by oil (911). We report here a facile method for making superhydrophobic surfaces from both soft (cotton or paper) and hard (metal or glass) materials. The process uses dual-scale nanoparticles of titanium dioxide (TiO2) that are coated with perfluorooctyltriethoxysilane. We created an ethanol-based suspension that can be sprayed, dipped, or painted onto surfaces to create a resilient water-repellent surface. By combining the paint and adhesives, we created a superhydrophobic surface that showed resilience and maintained its performance after various types of damage, including finger-wipe, knife-scratch, and multiple abrasion cycles with sandpaper. This method can also be used for components that require self-cleaning and lubricating such as bearings and gears, to which superamphiphobic (repels oil and water) surfaces (911) are not applicable.

A paint was created by mixing two different size ranges of TiO2 nanoparticles (~60 to 200 nm and ~21 nm) in an ethanol solution containing perfluorooctyltriethoxysilane (12). Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) of the constituent particles of the paint (Fig. 1A) show the dual-scale nature of the TiO2 nanoparticles. X-ray photoelectron spectroscopy (XPS) (Fig. 1B) showed that the titanium dioxide particles were coated with perfluorooctyltriethoxysilane.


Fig. 1. Paint characterizations. (A) SEM (top) and TEM (bottom) of the constituent nanoparticles in the paint. Sizes varied from ~60 to 200 nm for the TiO2 nanoparticles (Aldrich), whereas ~21 nm in size refers to P25. (B) XPS of the paint, where “F” refers to perfluorooctyltriethoxysilane and “Ti” refers to TiO2. (C) XRD patterns of treated and untreated substrates compared with the respective standard patterns for TiO2 anatase (the P25 particles had a small rutile component, as expected).


1) What does the surface coating look like on the micro- and nanoscale? Is it of dual length- scale?

2) What’s the chemical composition of the surface of the coating?

3) What’s the chemical and structural composition of the interior of the coating?

Answer, Question 1

Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM) show the particle sizes on the surface (SEM) and within the coating (TEM). The SEM image reveals the roughness of the coating, whereas the TEM image shows the two different length scales of particles.

Answer, Question 2

X-ray photoelectron spectroscopy (XPS) graphs show the chemical composition at the surface of the coating. On the x-axis the binding energy—i.e., the energy that needs to be overcome for the electrons to leave the atom—is plotted, and the y-axis shows simply the counts of how many electrons left their respective atoms. The taller the peak, the more predominant that species is. Electrons in different elements, molecular groups, and electronic orbits have unique binding energies, thus each peak can be clearly identified. The upper image in “B XPS” shows the peaks for the silane, and the lower image shows the TiO2 peaks. Thus it can be concluded that the TiO2 nanoparticles were indeed coated with silane.

Answer, Question 3

X-ray diffraction (XRD) graphs show the crystallographic data of the coating on different substrates. XRD depends on the interplay between crystal plane distances within the material and the incident angle of the x-rays. The x-axis shows the different incident angles and the y-axis again the intensity, i.e., counts. In all of the four graphs the black line represents the peaks that “normal” TiO2 has. The red curves are the actual data points for the untreated substrates and the blue curves for the respective substrates with the overcoat. In all cases, the blue line has peaks in the locations where either the bare substrate had peaks and/or where TiO2 has its peaks. (It’s an overlay of substrate and coating peaks.) Thus it can be concluded that the coating is applicable to a wide range of substrates and that it retains its crystallographic characteristics.

We used many different coating methods to create the water-repellent surfaces, including an artist’s spray-gun to coat hard substrates such as glass and steel, dip-coating for cotton wool, and a syringe (movie S1) to extrude the paint onto filter paper. After allowing the ethanol to evaporate for ~180 s at room temperature, the treated areas of the substrates supported water as near spherical droplets, whereas the untreated parts were readily wetted (it required ~30 min for the ethanol to fully evaporate from cotton wool and filter paper at room temperature) (fig. S1). We used x-ray diffraction (XRD) (Fig. 1C) to analyze the coatings on hard and soft substrates. The diffraction peaks show the expected patterns for nanoscaled TiO2.

On a surface that shows water repellence, water droplets tend to bounce instead of wetting the surface (1314). However, for soft substrates, extreme superhydrophobicity is required to achieve the bouncing phenomenon because the water droplets tend to be trapped onto the threads of the substrates (cotton wool) (15). Shown in fig. S2 are the water dropping tests on untreated glass, steel, cotton wool, and filter paper, which were readily wetted (the contact moment of the water droplets and the solid surfaces is defined as 0). Shown in Fig. 2 is the water bouncing process on dip-coated glass, steel, cotton wool, and filter paper surfaces. Water droplets completely leave the surface without wetting or even contaminating the surfaces (the water was dyed blue to aid visualization), indicating that the surfaces were superhydrophobic. In movie S2, we compare the water-affecting behavior between untreated and treated glass, steel, cotton wool, and filter paper, respectively. The effect of artificial rain on the treated surfaces is shown in movie S3; the drop sizes varied with random impact velocities, and the droplets could not wet the treated surfaces.


Fig. 2. Time-lapse photographs of water droplets bouncing on the treated glass, steel, cotton wool, and filter paper surfaces. Droplet sizes, ~6.3 ± 0.2 μL.


1) Does the coating render the surface superhydrophobic?

2) What is the droplet behavior on the coated surface?

3) Is there a difference in droplet behavior between the different substrates?

Answer, Question 1

Figure 2 shows snapshots from a high-speed camera movie of the impact of a water droplet (approximately 4 mm droplet diameter, so macroscopic with the dimension much larger than the coating’s microstructures) onto the coated surfaces. Because the droplets are bouncing off after the impact and do not leave water residue behind on the surface, it can be concluded that the coating is indeed superhydrophobic.

Answer, Question 2

When the water impacts the surface it first forms a hemispherical cap and a rim that is moving outward. The kinetic energy from the point of impact gets converted to free surface energy as the surface-to-volume ratio of the droplet increases. At some point the kinetic energy is approximately zero (3.1 ms) and the rim starts receding again, converting the free surface energy back into kinetic energy. At 11.7 ms the droplet jumps off the surface. With the image sequences, the droplet shapes, and time evolutions the researchers can compare their finding with those of other researchers and confirm assumptions.

Answer, Question 3

The droplet sequences are shown on four different substrates: glass, steel, cotton, and paper, where the former two are solid substrates and the latter two are flexible and soft substrates. The droplet shapes at each time step are almost the same on the different samples, which leads to the conclusion that the substrate stiffness does not significantly influence the superhydrophobic behavior of the coating and that the coating can indeed be applied to any surface.

The paint had good self-cleaning properties when applied on various substrates, especially for soft porous materials, such as those used in making clothes and paper. The coated surfaces show water-proofing properties from the water-bouncing and artificial rain tests. Further tests on cotton wool and filter paper are shown in figs. S3 (the experimental scheme) and S4 (the experimental results). As shown in fig. S4, A and B, the dip-coated cotton wool inserted into the methylene blue–dyed water formed a negative meniscus on the solid-liquid-vapor interfaces because of hydrophobicity (16). The cotton wool was removed from the water and remained fully white with no trace of contamination by the dyed water (fig. S3). A dirt removal test when an artificial dust (MnO powder) was put on the spray-coated filter paper, which was then cleaned by pouring water, is shown in fig. S4, C and D. The untreated piece of filter paper (placed below) was wet and polluted by the dirt, whereas the treated piece stayed dry and clean (fig. S3). The self-cleaning tests on the dip-coated cotton wool and spray-coated filter paper are shown in movie S4; a time-lapsed video clip of water droplets (dyed blue) staying on the dip-coated cotton wool and syringe-coated filter paper for 10 min is shown in movie S5, and neither the cotton wool nor the filter paper had blue left after the droplets were removed. These tests indicate that the soft substrates (cotton and paper) gained the nonwetting and self-cleaning properties after treating with the paint. Dirt removal tests were also carried out on dip-coated glass and steel surfaces; as shown in fig. S4, E and F, the droplet took the dirt (MnO powder) away, and the surfaces were cleaned along the path of the water droplet movement. The self-cleaning property of dip-coated glass and steel surfaces is shown in movie S6 in a high-speed motion capture.

Very few reports have shown any self-cleaning tests in oil because superhydrophobic surfaces normally lose their water repellency when even partially contaminated by oil. This is because the surface tension of the oil is lower than that of water, resulting in the oil penetrating through the surfaces. Making superamphiphobic surfaces (that repel both water and oils) is an effective way to solve this problem (91017). However, there are many instances that require both self-cleaning from water repellency and a smooth coating of oil, such as lubricating bearings and gears; under these conditions, superamphiphobic surfaces cannot be used because they will also repel lubricating oils. The self-cleaning tests of the painted surfaces after oil (hexadecane) contamination and immersion are presented graphically in fig. S5. As shown in fig. S6, water droplets still formed “marbles” on the dip-coated surface when immersed in oil, rather than forming a two-layer system (fig. S5A), thus indicating that the surfaces will retain their self-cleaning properties after being immersed in oil. For example, on the untreated areas of a glass slide, water droplets spread and wet the surfaces. We show in movie S7 water dropped on the dip-coated and untreated surfaces immersed in oil. We show in Fig. 3A the side view of a water droplet that formed a sphere at the oil-solid interface without wetting a spray-coated surface; the droplet then rolled off from the surface. As shown in Fig. 3, B and C, the water droplets slipped off from the spray-coated surface that was contaminated by oil (hexadecane), indicating self-cleaning was retained even after oil-contamination (fig. S5B and movie S8). We show in Fig. 3, D to F, a dirt-removal test on the spray-coated surfaces both in oil and air. The treated surface was fully contaminated by oil and then partly inserted into oil; dirt (MnO powder) was also put partly in oil and air onto the surface. Water was dropped so as to remove the dirt both in air and oil (fig. S5C and movie S9). This was to test the dirt-removal properties of the oil-contaminated painted surface both in air and under oil. For further dirt-removal tests on oil-contaminated painted surfaces, we used soil, household dust, and cooking oil from actual conditions and repeated the experiments shown in fig. S5C. As shown in fig. S7, soil and dust were removed by water from the dip-coated surfaces immersed either in hexadecane or cooking oil.


Fig. 3. Self-cleaning tests after oil-contaminations. (A) Water droplet was repelled by the treated surface when immersed in oil (hexadecane). (B and C) The treated surface retained its water-repellent property even after being contaminated by oil (D to F) The dirt removal test in oil-solid-vapor interfaces. Dirt was put partly in oil and air, the surface was contaminated by oil, water was dropped onto the surface, and this removed the dirt both in air and oil.


1) Does the coating retain its superhydrophobicity even when immersed in oil?

2) Does the coating retain its superhydrophobicity even when contaminated with oil?

3) Does the coating retain its self-cleaning properties when immersed in or contaminated with oil?

Answer, Question 1

Image A shows a water droplet (dyed with blue color for better visualization) rolling down the coated surface while being immersed in oil. The three-phase contact line is now solid-water-oil (hexadecane) instead of solid-water-air, which changes the force balance at the interface. Yet the water droplet forms a near spherical shape even when immersed in oil, thus the authors conclude that the coating keeps its water-repelling characteristics even when immersed in oil.

Answer, Question 2

Images B and C show a sequence of water droplets sliding down a coated surface that has been contaminated with oil. Now, the contact line is some mixture of solid-water-air/oil. The air voids that usually act as cushions are now mostly filled with oil and water droplets are no longer near-spherical (i.e., the surface is not superhydrophobic), yet the coating retains its high droplet mobility. Thus from figures A through C it can be concluded that the coating remains self-cleaning even when in contact with oil.

Answer, Question 3

In images D through F the authors contaminated the sample with MnO powder (to represent dirt or dust) both inside and outside of the oil and tested whether the water would remove the powder (image D). The droplets were then released on the slippery side of the sample outside of the bulk oil, yet on the oil-contaminated part (image E) and slid down the inclined surface, cleaning everything—outside and within the oil (image F). It can thus be concluded that the coating retains its self-cleaning properties when immersed in or contaminated with oil.

When the treated surfaces were immersed in oil, the oil gradually penetrated into the surface, so the water droplets were supported by both oil and the surface structures and were still marble-shaped (fig. S6). In this condition, the self-cleaning behavior in oil is similar to that in air (1820); thus, the treated surfaces retained the water-repellent and dirt-removal properties when immersed in oil (Fig. 3, D to F). In air, when the treated surfaces were contaminated by oil, the surface structures locked the oil as a lubricating fluid, and a slippery state was then achieved (2124). Dirt was removed from the treated surfaces simply by passing water over the surface. For these reasons, the treated surfaces retained their self-cleaning properties when being contaminated by oil.

Low surface robustness is the main issue limiting the widespread application of superhydrophobic coatings because the surface roughness is usually at the micro- or nanoscale and is mechanically weak and readily abraded (25). This surface roughness is partially protected by soft substrates, such as cotton wool and filter paper, because of their inherent flexibility (626) and ability to reduce direct friction between the coating and the surface. However, on hard substrates such as glass, nanostructures are easily destroyed or removed. We developed a method to bond the self-cleaning coatings to the substrates by using adhesives so as to apply more sophisticated and robust adhesive techniques and overcome the weak inherent robustness of superhydrophobic surfaces. We show in fig. S8 the “paint + adhesive (double-sided tape/spray adhesive) + substrates” sample preparation methods (fig. S8, A and B) and the relevant robustness tests, including finger-wipe (fig. S8C), knife-scratch (fig. S8D), and sandpaper abrasion (fig. S8, E and F). We show in fig. S9 and movie S10 the finger-wipe tests that compare the untreated, paint-treated, and “paint + double-sided tape”–treated (PDT) glass and steel substrates, respectively. After the finger-wipe, the paint directly coated on substrates was removed, whereas the double-sided tape-bonded paint was still left on the substrates, and the surfaces retained superhydrophobicity. Although the inherent robustness of the paint is intrinsically as weak as most superhydrophobic surfaces, it is friendly to adhesives, from which the robustness was gained. A glass substrate was used as one example for further robustness tests with double-sided tapes (knife-scratch and sandpaper abrasion tests); as shown in movie S10, the glass bonded with double-sided tape, and the paint still kept dry and clean after the knife-scratch and then water drop. The sandpaper abrasion tests were carried out on the PDT glass. The PDT glass weighing 100 g was placed face-down to sandpaper (standard glasspaper, grit no. 240) and moved for 10 cm along the ruler (Fig. 4A); the sample was rotated by 90° (face to the sandpaper) and then moved for 10 cm along the ruler (Fig. 4B). This process is defined as one abrasion cycle (movie S11), which guarantees the surface is abraded longitudinally and transversely in each cycle even if it is moved in a single direction. The water contact angles after each abrasion cycle are shown in Fig. 4C, and it was observed that the static water contact angles were between 156° and 168°, indicating superhydrophobicity was not lost by mechanical abrasion. In order to test whether this superhydrophobicity was kept after abrasion on the whole area but not merely on some points (contact angle measuring points), water droplet was guided by a needle to travel on the PDT glass surface after the 11th, 20th, 30th, and 40th cycle’s abrasion, respectively (movie S12). The water droplet traveling after the 40th cycle is shown in Fig. 4D.


Fig. 4. Sandpaper abrasion tests. (A and B) One cycle of the sandpaper abrasion test. (C) Plot of mechanical abrasion cycles and water contact angles after each abrasion test. (D) Water droplet traveling test after 40th cycle abrasion.


1) How can the mechanical robustness of the coating be tested?

2) Is there a reduction in contact angles after abrasion?

3) Is there degradation in the superhydrophobic behavior after abrasion?

Answer, Question 1

In images A and B the authors show how the abrasion tests were performed. They put a coated sample upside down onto sandpaper (i.e., the coated side touching the sandpaper), then put a weight on it and moved the entire sample over the sandpaper both longitudinally and transversely (i.e., in x and y direction). The total length of one of these cycles was thus 20 cm (two times 10 cm). This procedure allowed for high repeatability of the abrasion test. Other tests not shown in this figure, but in the supplement, were finger-wipe tests and knife scratches.

Answer, Question 2

The plot in image C shows the contact angles on the y-axis as a function of the cycles of abrasion test on the x-axis. The small variation in contact angles (less than 10%—that’s good for these kind of experiments) might come from measurements on different locations (think inhomogeneities) and general uncertainties. In general, there is no reduction in contact angle after 40 cycles of mechanical abrasion.

Answer, Question 3

Contact angles alone do not tell the entire story about the superhydrophobicity: Apart from high contact angles, which were shown in image C to be true, we also need a very small hysteresis—i.e., the difference between the advancing and receding contact angle. In image D, the droplet is moved across the surface that has been subjected to 40 abrasion cycles. The arrow shows the direction of movement. The hysteresis is very small: The droplet has a nearly perfect spherical shape. Thus, it can be concluded that the coatings (with “tape treatment”) retain their superhydrophobicity even after abrasion and are thus mechanically very robust. The graph also implies that the entire surface retains its superhydrophobicity, as the spots where the contact angle is measured vary between measurements and will always be different (think of the statistical importance—also see tab 3).

To enlarge the application scale and broaden the types of substrates, the spray adhesive [EVO-STIK (Bostik, UK)] was also used to bond glass, steel, cotton wool, and filter paper substrates with the superhydrophobic paint. We show in fig. S10 and movie S13 the finger-wipe tests on untreated, paint-treated, and “paint + spray adhesive”–treated (PSAT) substrates, respectively. On hard substrates (glass and steel), PSAT surfaces retained water proofing, whereas the paint was just removed when directly applied; the case is different on soft substrates (cotton and paper), on which paint was protected by their porous structures, resulting in both paint-treated and PSAT cotton and paper being superhydrophobic after the finger-wipe. However, in a more powerful test (sandpaper abrasion of cotton), this “protection” is limited (fig. S11). As shown in fig. S12 and movie S14, the sandpaper abrasion tests on PSAT substrates and both hard and soft substrates became robust after the PSAT treatment. As shown in fig. S13 and movie S15, the PSAT substrates retained water repellency after knife-scratch tests. After different damages, the PSAT materials still remained superhydrophobic, indicating that this method could efficiently enhance the robustness of superhydrophobic surfaces on different substrates; it is believed that the idea of “superhydrophobic paint + adhesives” can be simply, flexibly, and robustly used in large-scale industrial applications.

The superhydrophobic surfaces show that a robust resistance to oil contamination and ease of applicability can be achieved by implementing straightforward coating methods such as spraying, dip-coating, or even simply extrusion from a syringe. The flexibility of the “paint + adhesives” combination enables both hard and soft substrates to become robustly superhydrophobic and self-cleaning. The surfaces can be readily implemented in harsh and oily environments where robustness is required.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S13

References (2728)

Movies S1 to S15

References and Notes

  1. W. Barthlott, C. Neinhuis, Planta 202, 1–8 (1997).

  2. R. Blossey, Nat. Mater. 2, 301–306 (2003).

  3. I. P. Parkin, R. G. Palgrave, J. Mater. Chem. 15, 1689 (2005).

  4. T. Onda, S. Shibuichi, N. Satoh, K. Tsujii, Langmuir 12, 2125–2127 (1996).

  5. L. Feng et al., Adv. Mater. 14, 1857–1860 (2002).

  6. J. Zimmermann, F. A. Reifler, G. Fortunato, L. C. Gerhardt, S. Seeger, Adv. Funct. Mater. 18, 3662–3669 (2008).

  7. X. Zhu et al., J. Mater. Chem. 21, 15793 (2011).

  8. Q. Zhu et al., J. Mater. Chem. A 1, 5386 (2013).

  9. A. Tuteja et al., Science 318, 1618–1622 (2007).

  10. X. Deng, L. Mammen, H. J. Butt, D. Vollmer, Science 335, 67–70 (2012).

  11. Y. Lu et al., ACS Sustainable Chem. Eng. 1, 102 (2013).

  12. Materials and methods are available as supplementary materials on Science Online.

  13. D. Richard, C. Clanet, D. Quéré, Nature 417, 811 (2002).

  14. J. C. Bird, R. Dhiman, H. M. Kwon, K. K. Varanasi, Nature 503, 385–388 (2013).

  15. Y. Lu et al., J. Mater. Chem. A 2, 12177 (2014).

  16. D. Vella, L. Mahadevan, Am. J. Phys. 73, 817 (2005).

  17. A. Tuteja, W. Choi, J. M. Mabry, G. H. McKinley, R. E. Cohen, Proc. Natl. Acad. Sci. U.S.A. 105, 18200–18205 (2008).

  18. A. Nakajima et al., Langmuir 16, 7044–7047 (2000).

  19. R. Fürstner, W. Barthlott, C. Neinhuis, P. Walzel, Langmuir 21, 956–961 (2005).

  20. B. Bhushan, Y. C. Jung, K. Koch, Langmuir 25, 3240–3248 (2009).

  21. T. S. Wong et al., Nature 477, 443–447 (2011).

  22. M. Nosonovsky, Nature 477, 412–413 (2011).

  23. A. Grinthal, J. Aizenberg, Chem. Mater. 26, 698–708 (2014).

  24. D. C. Leslie et al., Nat. Biotechnol. 32, 1134–1140 (2014).

  25. M. Im, H. Im, J. Lee, J. Yoon, Y. Choi, Soft Matter 6, 1401 (2010).

  26. B. Wang et al., ACS Appl. Mater. Interfaces 5, 1827–1839 (2013)