This is an electronic reprint of the original article.This reprint may differ from the original in pagination and typographic detail.

Powered by TCPDF (www.tcpdf.org)

This material is protected by copyright and other intellectual property rights, and duplication or sale of all or part of any of the repository collections is not permitted, except that material may be duplicated by you for your research use or educational purposes in electronic or print form. You must obtain permission for any other use. Electronic or print copies may not be offered, whether for sale or otherwise to anyone who is not an authorised user.

Rontu, Ville; Jokinen, Ville; Franssila, SamiScalable Superomniphobic Surfaces

Published in:Journal of Microelectromechanical Systems

DOI:10.1109/JMEMS.2019.2950769

Published: 01/01/2020

Document VersionPeer reviewed version

Please cite the original version:Rontu, V., Jokinen, V., & Franssila, S. (2020). Scalable Superomniphobic Surfaces. Journal ofMicroelectromechanical Systems, 29(1), 54 - 61. https://doi.org/10.1109/JMEMS.2019.2950769

1

Scalable Superomniphobic SurfacesVille Rontu , Ville Jokinen, and Sami Franssila

Abstract— Superomniphobic surfaces that repel liquids ofextremely low surface tension rely on carefully fabricated doublyre-entrant topographies, typically made by silicon deep reactiveion etching technology. However, previously published processeshave depended on critically timed etching steps, which aredifficult to downscale. We present a scalable process that elimi-nates the critically timed etching steps. It is based on the useof silicon-on-insulator wafers and a silicon oxide foot of themicropillar, which makes the isotropic silicon release step non-critical. The process allows easy downscaling of pillars from20 μm to 10 μm and 5 μm. The downscaling increases thestability of the Cassie state. Based on the process, we are ableto create superomniphobic surfaces that sustain perfluorohexane(FC-72), which has the lowest surface tension of the known liquidsat room temperature (γlv = 11.91 mN/m at 20 ◦C), in the Cassiestate at droplet diameters down to 200 micrometers. These are thesmallest perfluorohexane droplets repelled to date. [2019-0207]

Index Terms— Cassie Wenzel transition, microfluidics,plasma etching, self-cleaning, silicon, superhydrophobic,superoleophobic.

I. INTRODUCTION

THE wettability of a material is affected by its surfacechemistry, topography and the properties of the liquid.

Wettability affects many natural and industrial processes suchas wicking [1], protein adsorption [2], and condensation [3].By careful design of the surface topography and chemistry,superhypdrophobic, superoleophobic and superamphiphobicsurfaces, which repel water, oil and both respectively, havebeen created. These surfaces have prospective applications indroplet control, heat exchange, antifouling, anti-icing and self-cleaning [4].

Wettability is commonly characterized by the contact angleand the Young’s equation:

γlv cos θ = γsv − γsl, (1)

where θ is the contact angle and γlv, γsv and γsl are the surfaceand interfacial energies of the liquid vapor, solid vapor andthe solid liquid interface respectively. On real surfaces, there

Manuscript received September 18, 2019; accepted October 26, 2019. Thiswork was supported in part by the Finnish Cultural Foundation and in part bythe Academy of Finland under Grant 297360. (Ville Rontu and Ville Jokinencontributed equally to this work.) Subject Editor R. Ghodssi. (Correspondingauthor: Ville Rontu.)

V. Rontu is with the Department of Chemistry and Material Science,Aalto University, 02150 Espoo, Finland, and also with Aalto Nanofab, AaltoUniversity, 02150 Espoo, Finland (e-mail: ville.rontu@aalto.fi).

V. Jokinen and S. Franssila are with the Department of Chemistryand Material Science, Aalto University, 02150 Espoo, Finland (e-mail:ville.p.jokinen@aalto.fi; sami.franssila@aalto.fi).

This article has supplementary downloadable material available athttp://ieeexplore.ieee.org, provided by the authors.

Color versions of one or more of the figures in this article are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JMEMS.2019.2950769

Fig. 1. a) Simple vertical structures are able to repel liquids with inherentcontact angle >90◦. b) Re-entrant geometry extends repellence to liquidswith inherent contact angle 0◦ < 90◦. c) Doubly re-entrant geometry is ableto repel even liquids with inherent contact angle of 0◦.

are two characteristic contact angles: the advancing contactangle θadv and the receding contact angle θrec. The differencebetween these two is called the contact angle hysteresis.

Contact angles observed on smooth surfaces range from0◦ to around 120◦ for water and from 0◦ to <90◦ for oils,depending on the surface chemistry. However, achieving con-tact angles higher than those possible for smooth surfaces ishighly desirable to create liquid-repellent surfaces. A suitablyrough surface with a low surface energy coating will exhibitsuperhydrophobicity [5]–[7], which is characterized by highwater contact angles, low contact angle hysteresis and lowsliding angles. Superhydrophobicity is based on the Cassie-Baxter state [8] where an air pocket called the plastron limitsthe contact between the liquid and the solid. The contact angleof a droplet in an ideal Cassie-Baxter state (perfectly flatliquid-solid and liquid-vapor interfaces at the contact area) isoften described by the equation:

cos θCassie = f (1 + cos θ) − 1, (2)

where θCassie is the contact angle observed on the roughsurface, f is the fraction of the surface that is in contactwith the liquid and θ is the inherent contact angle of thesame material as a smooth surface. Since the inherent contactangle can be >90◦, simple vertical structures (Fig. 1a) createa pressure barrier that stabilizes the plastron. A suitably lowsolid fraction will result in surfaces with very high advancingand receding contact angles and very low adhesion of waterdroplets.

For oils, however, simple vertical structures are not suitablesince the inherent contact angles are <90◦ so capillary wick-ing would make the plastron spontaneously fill with liquid.However, a metastable Cassie-Baxter state can be created byusing re-entrant (also called overhanging) geometries (Fig. 1b)[9]–[11]. Re-entrant geometries combined with fluorinatedcoatings were shown to lead to superoleophobic surfaces [11]that had high advancing and receding contact angles to severaloils, including octane (γlv = 21.62 mN/m at 20 ◦C) [12].Re-entrant oleophobic surfaces have been fabricated bylithography and silicon etching [11], hydrothermal ZnO

2

growth [13], electrospinning [14], and spray coating [15].Tuteja et al. [16] showed that the robustness of the oil-repellent state depends on the ability of the topography toresist liquid penetration either by sagging or by bypassingthe energy barrier created by the re-entrant geometry andthat downscaling leads to an increase in the robustness formicropillar surfaces. The limitation of these surfaces is thatin practice they still require the inherent contact angle to beclearly higher than 0◦, since otherwise the surface is unstableagainst any external pressure that is inevitably present (suchas hydrostatic or Laplace pressure). Therefore, liquids withextremely low surface tension, including fluorinated solvents,could not be repelled.

This problem was addressed by Liu and Kim [17] whodesigned doubly re-entrant surface geometries (Fig. 1c). Thesesurfaces were repellent to liquids even in cases where theinherent contact angle is 0◦, as demonstrated by the abil-ity of the surfaces to repel fluorinated solvents includingperfluorohexane (γlv = 11.91 mN/m at 20 ◦C) [12]. Sincethe surfaces work with any inherent contact angle, surfacechemistry modification is not needed, but fabrication of thedoubly re-entrant geometry is a challenge. The main method sofar has been to utilize lithography and silicon etching [17]–[19]although 2-photon polymerization based 3D printing has alsobeen reported [20], [21]. The problem with random structures,which would be easier to fabricate than lithographic, is thatit is not enough to have the surface doubly re-entrant inmost places, it needs to be doubly re-entrant everywhere asotherwise the non-doubly re-entrant areas will initiate a lateralcapillary filling process. Domingues et al. [18], [19] addressedthe problem of lateral capillary filling by using doubly re-entrant cavities that prevent spreading of the liquids in case ofa collapsed Cassie-Baxter state. However, this requires widespacing between the cavity walls to satisfy the low solidfraction requirement for superomniphobicity leading to lowbreakthrough pressure. A second problem is that the overallstability of the Cassie state on the doubly re-entrant surfacescan still be low due to the low surface tension. To address thisissue, it would be beneficial to fabricate the features at smallersize scales to scale up the breakthrough pressure. However,the silicon etching process presented by Liu and Kim dependson a critically timed etching step that will become more andmore challenging as the features are downscaled. Furthermore,while 3D printing by 2-photon polymerization can produceeven submicron structures [22], the process is extremely slowand is limited to millimeter scale areas.

Here we present a scalable lithography and silicon etchingprocess that can produce doubly re-entrant surfaces for liquidrepellency without critically timed etching steps. We show howthe surfaces repel low surface tension liquids and how thestability of the Cassie state improves as the feature size isdownscaled between 20 μm and 5 μm.

II. EXPERIMENTAL

A. Doubly Re-Entrant Pillar Fabrication

The pillars were fabricated on the 25-μm thick (100)oriented n-type (1-20 �·cm, phosphorus doped) device layer of

Fig. 2. Fabrication steps for doubly re-entrant structure. a) SOI wafer,b)–c) first lithography, d) silicon DRIE etching, e) resist removal, f) holefilling, g)–h) second lithography, i) oxide etching, j) Si etching for verticaloverhang, k) resist removal, l) conformal oxide deposition, m) spacer etching,n) isotropic etching of silicon.

a 150-mm silicon-on-insulator (SOI) wafer (Ultrasil), which isseparated from 600-μm thick n-type (1-20 � ·cm, phosphorusdoped) handle wafer by a 1-μm thick buried oxide, by theprocess shown in Fig. 2.

The wafers (Fig. 2a) were cleaned in 5:1:1H2O:NH3:H2O2(H2O de-ionized (18 M� · cm), NH3min. 25 % VLSI grade (Honeywell), H2O2 30 w-% VLSIgrade (Honeywell)) solution according to SC-1 cleaningprotocol. Then photolithography with Süss MA-6 at 365-nmwavelength was done to pattern 2-μm holes on AZ5214Ephotoresist (Microchemistry) (Fig. 2b–c). It was used as an

3

TABLE I

DRIE PROCESS LOOP PARAMETERS

etch mask in the following DRIE step through the devicelayer (Fig. 2d) using Bosch type process in Aviza Omegai2L etcher using parameters shown in Table I creating 2-μmdiameter and 25-μm deep holes in silicon.

After DRIE and removing the photoresist (Fig. 2e), the holeswere filled by growing a 1-μm thick SiO2 (Fig. 2f) by CVDmethod at 710 ◦C and 250 mTorr pressure in CentrothermE1200 HT 260-4 using a tetraethyl orthosilicate (TEOS)(99.99 %, Versum Materials) based process. After this, anotherlithography (Fig. 2g–h) was performed to define 5-, 10- and20-μm diameter caps on AZ5214E photoresist, which werealigned with the former 2-μm holes. This photoresist wasused as an etch mask in the subsequent anisotropic SiO2 RIE(Fig. 2i) using Surface Technology Systems Advanced OxideEtcher (STS-AOE) and C4F8 and H2 chemistry. The etch timewas 4 min 10 s with C4F8 flow of 10 sccm, H2 flow of 8 sccm,coil power of 1150 W and platen power of 250 W. The basepressure was 0.5 Pa, but pressure strike to 3.3 Pa was used.After etching through the SiO2, anisotropic Si etching wascontinued (Fig. 2j) in STS Advanced Silicon Etcher (STS-ASE) with a mixed process for 5 min using 40-sccm SF6flow, 120-sccm C4F8 flow, 5-sccm O2 flow and ramping upthe C4F8 flow by 2 sccm/min and decreasing platen powerfrom 30 W by 1 W/min resulting in smooth sidewalls and Sietch depth of 1 μm. The coil power was 600 W and pressure at1.6 Pa. After the Si etching, the photoresist and polymer pas-sivation from the sidewalls was removed (Fig. 2k) by 30-minO2 plasma treatment in Oxford Instruments PRS900 plasmastripper using 320-sccm O2 flow and 600-W RF power at113-Pa pressure. Following this the surface was covered with212-nm thick SiO2 film (Fig. 2l) by atomic layer deposition(ALD) (1200 cycles) in Beneq TFS-500 tool at 120 ◦C usingcommercial AP-LTO330 from Air Products and O2-plasma asthe precursors. The AP-LTO330 was kept at 25 ◦C and pulsedfor 200 ms and purged for 500 ms. The pulse time for O2plasma was 3 s with RF power at 100 W and purge time2 s. After ALD, the ALD SiO2 in planar areas was etchedfor 1 min in STS-AOE using the same process as before foroxide etching leaving only ALD SiO2 in the vertical sidewalls(Fig. 2m). Finally, all silicon was etched away with isotropicetching in STS-ASE in 15 min using 129-sccm SF6 flow and10-sccm O2 flow at 600-W coil power and 8-W platen powercreating doubly re-entrant structures consisting of only SiO2(Fig. 2n).

B. Surface Characterization

The contact angle measurements, drop height experimentsand the Laplace breakthrough pressure experiments were per-formed using a goniometer (Theta, Biolin Scientific) in an ISO6 cleanroom at 21 ± 0.5 ◦C temperature and 45 ± 5 % relativehumidity. Prior to all experiments, the surfaces were cleanedby 1 min O2 plasma treatment in Oxford Instruments PRS900to ensure a clean silica surface chemistry. De-ionized water(18 M�·cm) was obtained from the in-house water purificationsystem. Acetone and 2-propanol (isopropanol) were VLSIgrade (VLSI Puranal) from Honeywell. Hexadecane (99%),perfluorooctane (98%) and perfluorohexane (99%), were fromSigma-Aldrich.

C. Contact Angle Measurements

Advancing and receding contact angles were measured bythe sessile droplet needle method. A gauge 30 steel needle(0.311 mm outer diameter) was inserted into the droplet andthe volume of the droplet was increased or decreased at therate of 0.1 μl/s to measure the advancing and receding con-tact angles respectively. The measurements were carried outbetween droplet volumes 2 μl and 7 μl. For perfluorohexane,instead a gauge 22 steel needle (0.718 mm outer diameter) anda 0.2 μl/s drop rate were used for the advancing angle, whilethe receding angle was recorded from an evaporating droplet.All reported values are the averages and standard deviationsof three measurements.

D. Drop Height Experiments

Hexadecane and perfluorohexane droplets were dropped onthe surfaces from different heights and Cassie state or collapseinto a Wenzel state was observed. The height is calculatedfrom the tip of the needle to the surface. The volume ofthe droplet was increased with a rate of 0.5 μl/s until theforce balance between gravity and surface tensions led to thedroplet detaching. A gauge 30 and 22 steel needle was usedfor hexadecane and perfluorohexane respectively. The volumesof the droplets were measured from the images as an averageof 5 measurements to be 3.84 μl ± 0.25 μl for hexadecaneand 1.59 μl ± 0.04 μl for perfluorohexane. For each surface,the dropping height was increased using a z-stage until thethreshold for Wenzel collapse was found. The reported valuesare the averages of the highest measured height that still ledto the droplet staying in the Cassie state and the lowest heightthat caused a Wenzel state collapse. The reported ± valuesare half the difference between the highest Cassie and thelowest Wenzel height. Three experiments at each height foreach surface were performed. Most experiments had a clear3-0 result, but in some cases 2-1 (e.g. twice Cassie andonce Wenzel) were obtained, and these were counted for themajority.

E. Laplace Breakthrough Pressure Experiments

2-propanol and perfluorohexane droplets of 2–3 μl volumewere placed on a surface and their evaporation was recordedon the camera at 1 fps. The radius of the droplet in the last

4

Fig. 3. SEM micrographs of doubly re-entrant pillars: a) 10-μm pillar, b) 5-μm pillar with 20-μm pitch, c) large array of 5-μm pillars with 20-μm pitchand d) showing an edge of a 5-μm pillar array. Sample is coated with 30 nm of sputtered Pt for charging prevention. D means diameter of the pillar, p thepitch, and w and h is the width and the height of the overhang respectively. For our samples the D is either 5, 10 or 20 μm, p is 4 times D, h is 1 μm andw is 200 nm.

frame in which the droplet remained in the Cassie state beforethe Wenzel collapse was recorded. The reported values are theaverages and standard deviations of three measurements.

III. RESULTS AND DISCUSSION

A. Fabrication

Doubly re-entrant pillars were fabricated with a scalableprocess consisting of two lithography steps and no criticaltimed etching steps. The process consists of first etchingholes for the pillar stem through the device layer of an SOI-wafer. The buried oxide (BOX) layer in the SOI-wafer actsas an etch stop for the hole etching. The holes are thenfilled with conformal-deposited SiO2. The second lithography,which defines the 5, 10 and 20-μm diameter pillar caps,requires alignment with the pillar stems, but it is otherwisevery simple. Fig. 3a and b shows small misalignment in thealignment, which does not affect wetting performance. It isenough to align the pillar stem and cap so that a gap existsbetween the vertical overhang and pillar stem. Two etch stepsare performed using the same resist mask from the secondlithography step: oxide by reactive ion etching (RIE) and1 μm of silicon by deep reactive ion etching (DRIE). Thiscreates vertical walls into which vertical overhang materialis deposited by highly conformal ALD. The ALD SiO2 usedhere is a low-temperature process (120 ◦C) giving low stressSiO2 at high growth per cycle (1.77 Å/cycle). Anisotropicoxide etch without a mask is carried out to create sidewallspacers of SiO2. Oxide on horizontal surfaces is removed in

the process. Since the doubly re-entrant pillar is completelymade of SiO2, the final isotropic silicon etching is non-criticaland all the silicon can be easily etched away. Non-criticaletching is important since large etchable area processes tendto have poor uniformity. The etchable area is approximately90–95 % with superomniphobic surfaces.

Samples with three different pitches were fabricated for eachpillar cap diameter labeled as 5-μm S, M or L, 10-μm S,M or L and 20-μm S, M or L. The S pitch is 4 times thepillar cap diameter, M pitch is 5 times and L 6 times. Thesize of the pillar and the pitch affect both the solid fractionand the stability of the Cassie state, so it is important to havea fabrication process with robust control over both of these.Fig. 3 shows SEM micrographs of a 10-μm pillar (Fig 3a)and 5-μm S pillar array (Fig. 3b and c). These samples wereall fitted on a single 150-mm SOI-wafer in square arrays of 5,10, and 20 mm covering the entire wafer surface. Dividing thesurface into arrays prevented spreading of the liquids in caseof a loss of Cassie-Baxter state. Non-criticality of the finalisotropic etching, which is visual in Fig. 3d) allowed etchingaway the silicon completely even though lateral etch depthbetween 5 and 20 um pillar caps is 4 times larger. This is notpossible with the processes previously used to create doublyre-entrant pillars [17], [18] where small over etching in finalisotropic etch could be catastrophic.

The process that we have described can be further down-scaled. The first lithography could be efficiently done withan optical stepper or nanoimprint lithography (NIL) enablingsub-micron stem. The second lithography’s alignment becomes

5

slightly more challenging with smaller features. Here we haveused high-temperature TEOS CVD oxide process for the holefilling. With decreased dimensions, the hole filling could berealized with a low temperature ALD process decreasingthe temperature budget of the whole fabrication process to<120 ◦C. An ALD process for the sidewall spacer also enableseasy downscaling of the spacer thickness. With downscaling,the processing time would also decrease as etch and depositionsteps would become shorter. The materials could also bechanged. Instead of the BOX, another etch stop material couldbe used. Silicon on the device layer could be changed toanother material that can be etched both anisotropically andisotropically with good selectivity to the etch stop layer andthe doubly re-entrant pillar material. The doubly re-entrantpillar can be made from practically any material that can bedeposited with ALD. It could also be possible to realize sur-faces on polymer substrates using similar fabrication strategy.In addition to pillars, the same process can be used to makedoubly re-entrant cavities shown in [18] or gradient lines usedfor droplet manipulation [23].

B. Omniphobicity Characterization

In order to validate that the omniphobic performance of thesurfaces is at least as good as the previously reported, we per-formed droplet contact angle and stability characterization.

The obtained surfaces were able to repel water, hexadecane,acetone, 2-propanol and perfluorohexane as shown in Fig. 4aand Table II. The surface chemistry in all of these caseswas freshly plasma-cleaned SiO2. All the test liquids wouldcompletely wet a planar SiO2 surface, so the doubly re-entrant geometry is purely responsible for the stability of theCassie state. The droplets were in a non-sticky Cassie statecharacterized by high contact angles (Fig. 4b) and low slidingangles. Fig. 4c–d show a 2-μl perfluorohexane droplet slidingon the 5-μm M surface with a tilting angle of 6◦.

Table II shows the results of the advancing and recedingcontact angle measurements. For the range of solid fractionspresent on our surfaces, the Cassie contact angles variedbetween 144◦ and 161◦. Since the inherent contact angle is∼0◦ for all liquids, we do not expect the surface tensionof the liquid to affect the contact angle on the omniphobicsurface. Nevertheless, we note that there is a downward trendin both the advancing and receding contact angles as thesurface tension becomes smaller.

The advancing and the receding contact angles in Table IIget lower as the solid fraction of the samples increases (the10-μm M sample with hexadecane is the only exception)as is predicted by the Cassie equation (2), although withinthe studied solid fraction range (2.7–9.6 %) the effect onthe contact angles is quite modest. It is worth recognizingthat there is significant uncertainty in the measured valuesof very high contact angles [24], especially on low surfacetension liquids [25], [26]. To achieve similar solid frac-tions using the doubly re-entrant cavity approach presentedby Domingues et al. [18], the spacing between cavity wallswould need to be approximately ten times larger than our pillarspacing. This can cause a problem when the repelled droplets

TABLE II

CONTACT ANGLES

(diameter 100 μm–1 mm in our case) are of similar size asthe cavities to achieve the same ratio. It is also known that thewavelength of roughness should be small enough, possiblyup to 3 orders of magnitude smaller, compared to the dropletsize [27].

1) The Effect of Scaling on the Stability of the Cassie State:The size of the pillar and the spacing between the pillars bothhad a clear effect on the stability of the Cassie state. Thestability for doubly re-entrant pillars is characterized by thebreakthrough pressure (�Pmax) given by

�Pmax = γπ D�

p2 − π D�2/4[17], (3)

where p is the pitch and D� is the diameter of a pillar minustwice the width of the vertical part overhang (w), which inour case is close to the diameter of the pillar since the widthof the overhang is only 200 nm. The internal pressure of aspherical droplet is given by the Laplace pressure:

PLaplace = 2γ

R, (4)

where R is the radius of the droplet.Two types of experiments were performed to characterize

the effect of the size of the pillar and the spacing on thebreakthrough pressure. The Laplace pressure test is shownin Fig. 5a–e and the drop height test is shown in Fig. 5f–g.Both tests showed that scaling down the pillar size is beneficialas big differences were observed between 20-μm pillars,10-μm pillars and 5-μm pillars. Fig. 4a shows an evaporatingperfluorohexane droplet on the 5-μm S surface just before theLaplace pressure becomes too high for the surface to sustain,after which it collapsed to the Wenzel state as shown in Fig. 5b.

6

Fig. 4. Omniphobic surfaces. a) Digiphotos of water, hexadecane, acetone, 2-propanol and perfluorohexane droplets. Water, acetone and perfluorohexane areon the 5-μm S surface. Hexadecane and 2-propanol are on the 5-μm M surface. b) The advancing contact angle of perfluorohexane on the 10-μm M surface.d) A 2-μl perfluorohexane droplet sliding with 6◦ sliding angle on the 5-μm M surface. The white arrows point to the same position in both images. Thescalebar (bottom right) is 2 mm in all figures.

Fig. 5. Effect of the pillar size to the stability of the Cassie state of low surface tension liquids. a) An evaporating perfluorohexane droplet on the surface justbefore the Wenzel collapse on a 5-μm S surface. b) The same droplet just after Wenzel collapse. c) An evaporating perfluorohexane droplet on the surfacejust before the Wenzel collapse on a 10-μm S surface. d) An evaporating 2-propanol droplet on the surface just before the Wenzel collapse on a 5-μm Msurface. e) An evaporating 2-propanol droplet on the surface just before the Wenzel collapse on a 10-μm M surface. f) A perfluorohexane droplet (1.6 μl)dropped on the 5-μm S surface from the height of 3.3 mm. g) A hexadecane droplet (3.8 μl) dropped on the 5-μm M surface from the height of 8 mm. Thescale bar (bottom right) is 500 μm in all figures.

See also supplementary Video S1. In this case, the diameterof the droplet was 199 μm at the point of Wenzel collapse,corresponding to a Laplace pressure of 240 Pa. In contrast,Fig. 5c shows the perfluorohexane droplet just before collapseon the 10-μm S surface, which has the pillar diameter andthe pitch doubled. The diameter of the droplet just before thecollapse was 629 μm, corresponding to a Laplace pressureof 76 Pa. Fig. 5d–e show a corresponding difference betweenthe size of a 2-propanol droplet just before collapse on the5-μm M and 10-μm M surfaces. The results of the Laplacepressure stability test are summarized in Table III.

It is clear from the results that making the pillars smallerincreased the maximum Laplace pressure that the surfacecould tolerate before the droplet collapsed into Wenzelstate. In fact, the perfluorohexane droplets on our 5-μmS surfaces (200 μm diameter) are smaller than the per-fluorohexane (FC-72) droplets (300 μm diameter) reportedby Liu and Kim [17] and the smallest (highest Laplacepressure) perfluorohexane droplets reported to date. Thefour pairs of surfaces that could be directly comparedwere 5-μm S/M to 10-μm S/M with perfluorohexane and5-μm M/L to 10-μm M/L with 2-propanol, for which the

7

TABLE III

RESULTS OF THE LAPLACE PRESSURE STABILITY MEASUREMENTS

increase in maximum pressure varied between 3.2 fold and6.2 fold.

Theoretically, approximately two-fold increase would beexpected for scaling between 10 μm and 5 μm (approx-imately, since the width of the overhang was not varied).One partial explanation is that any additional pressure, e.g.from vibrations, would tend to increase the ratio above two.This is supported by the fact that the experimentally observedWenzel collapses mostly occurred earlier than expected fromthe theoretical predictions given by (3). The 20-μm samplesdid not reliably sustain droplets in the Cassie state and weretherefore excluded from the quantitative analysis. Finally,the case with 2-propanol on the 5-μm S samples was unclear,with seemingly no clear point of collapse into Wenzel state(see supplementary Fig. S1).

The results from the drop height experiments are summa-rized in Table IV. Fig. 5f shows a perfluorohexane dropletdropped on the 5-μm S surface from the height of 3.3 mmand remaining in the Cassie state. Fig. 5g shows a hexadecanedroplet dropped on the 5-μm M surface from the heightof 8 mm bouncing off, before settling in the Cassie state. Forperfluorohexane, the clearest result is that the 5-μm S samplesustained drops from clearly greater heights than the othersamples. Compared to its counterpart, the 10-μm S sample,the maximum drop heights were 3.8 mm ± 0.4 mm and2.7 mm ± 0.3 mm, respectively. The 10-μm L and all the20-μm samples did not remain in Cassie state even whendropped from 2 mm, at which point the droplet starts to touchthe surface before detaching from the pipette. The beneficialeffect of scaling on the stability of the Cassie sate is veryclear from the hexadecane results. The maximum drop heightsincrease monotonously when downscaling from the 20-μmto 10-μm and to 5-μm. For example, for the M samples,the maximum drop heights were 3.8 mm ± 0.8 mm, 7.2 mm± 0.8 mm and 8.8 mm ± 0.8 mm for the 20-μm, 10-μm and5-μm sized samples, respectively.

One additional observation that we made was that the fluori-nated oil perfluorooctane was seemingly more difficult to repelthan perfluorohexane, although it has a higher surface tensionof ∼14 mN/m at 20 ◦C [30] compared to the 11.91 mN/m ofperfluorohexane [12]. It was possible to obtain a Cassie statewith perfluorooctane as well, but it commonly underwent an

TABLE IV

RESULTS OF THE DROP HEIGHT EXPERIMENT

unexplained transition into Wenzel state after short residencetime on the surface, especially when the droplet was stationary(rolling droplets were more stable). A possible explanationis that the much higher vapor pressure of perfluorohexane(23 kPa at 20 ◦C) [31] compared to perfluorooctane (3 kPa at20 ◦C) [31] helps to stabilize the Cassie state due to pressurebuildup under the droplet.

IV. CONCLUSION

We have presented a scalable process for fabricating doublyre-entrant silicon oxide microstructures in a process that doesnot require a critically timed etching step. The process utilizesstandard microfabrication steps and can produce surfacesin the wafer scale, 150-mm wafers in this study. Further,we showed how scaling down the pillar dimensions leads toan increase in the stability of the Cassie state, as predictedby theory. There are two principal ways to increase thestability of a given liquid on an omniphobic surface of agiven overall geometry. First, it is possible to attempt toincrease the upward surface tension force by decreasing thepitch. However, that approach has a clear drawback in that thesolid fraction of the surface will then increase, which at somepoint renders the droplets immobile, which is contrary to thepurpose of omniphobic surfaces. The second option, whichwas investigated in this work, is to downscale the geometry.

8

Contrary to the first approach, there is no increase of the solidfraction, and thus no drawback for reducing the performance.The main drawback of downscaling is that the processingbecomes more difficult, which can be addressed by havinga microfabrication process that is easy to scale down, such asthe one presented in this work.

ACKNOWLEDGMENT

Micronova provided facilities for the research.

REFERENCES

[1] J. Bico, U. Thiele, and D. Quéré, “Wetting of textured surfaces,” ColloidsSurf. A, Physicochem. Eng. Aspects, vol. 206, nos. 1–3, pp. 41–46,Jul. 2002.

[2] G. B. Sigal, M. Mrksich, and G. M. Whitesides, “Effect of surfacewettability on the adsorption of proteins and detergents,” J. Amer. Chem.Soc., vol. 120, no. 14, pp. 3464–3473, Apr. 1998.

[3] K. Rykaczewski et al., “How nanorough is rough enough to makea surface superhydrophobic during water condensation?” Soft Matter,vol. 8, no. 33, pp. 8786–8794, Mar. 2012.

[4] H. Liu, Y. Wang, J. Huang, Z. Chen, G. Chen, and Y. Lai, “Bioin-spired surfaces with superamphiphobic properties: Concepts, synthesis,and applications,” Adv. Funct. Mater., vol. 28, no. 19, Mar. 2018,Art. no. 1707415.

[5] A. Lafuma and D. Quéré, “Superhydrophobic states,” Nature Mater.,vol. 2, no. 7, pp. 457–460, Jun. 2003.

[6] N. J. Shirtcliffe, G. McHale, M. I. Newton, G. Chabrol, and C. C. Perry,“Dual-scale roughness produces unusually water-repellent surfaces,”Adv. Mater., vol. 16, no. 21, pp. 1929–1932, Nov. 2004.

[7] V. Jokinen, L. Sainiemi, and S. Franssila, “Complex droplets onchemically modified silicon nanograss,” Adv. Mater., vol. 20, no. 18,pp. 3453–3456, Sep. 2008.

[8] A. B. D. Cassie and S. Baxter, “Wettability of porous surfaces,” Trans.Faraday Soc., vol. 40, no. 5, pp. 546–551, Jun. 1944.

[9] S. Herminghaus, “Roughness-induced non-wetting,” Europhys. Lett.,vol. 52, no. 2, pp. 165–170, Oct. 2000.

[10] L. Cao, H.-H. Hu, and D. Gao, “Design and fabrication of micro-texturesfor inducing a superhydrophobic behavior on hydrophilic materials,”Langmuir, vol. 23, no. 8, pp. 4310–4314, Mar. 2007.

[11] A. Tuteja et al., “Designing superoleophobic surfaces,” Science, vol. 318,no. 5856, pp. 1618–1622, Dec. 2007.

[12] J. J. Jasper, “The surface tension of pure liquid compounds,” J. Phys.Chem. Reference Data, vol. 1, no. 4, p. 841, Oct. 1972.

[13] A. R. Bielinski et al., “Rational design of hyperbranched nanowiresystems for tunable superomniphobic surfaces enabled by atomic layerdeposition,” ACS Nano, vol. 11, no. 1, pp. 478–489, Jan. 2017.

[14] S. Pan, A. K. Kota, J. M. Mabry, and A. Tuteja, “Superomniphobic Sur-faces for Effective Chemical Shielding,” J. Amer. Chem. Soc., vol. 135,no. 2, pp. 578–581, Dec. 2013.

[15] R. Campos et al., “Superoleophobic surfaces through control of sprayed-on stochastic topography,” Langmuir, vol. 28, no. 25, pp. 9834–9841,May 2012.

[16] A. Tuteja, W. Choi, J. M. Mabry, G. H. McKinley, and R. E. Cohen,“Robust omniphobic surfaces,” Proc. Nat. Acad. Sci. USA, vol. 105,no. 47, pp. 18200–18205, Nov. 2008.

[17] T. Liu and C.-J. Kim, “Turning a surface superrepellent even to com-pletely wetting liquids,” Science, vol. 346, no. 6213, pp. 1096–1100,Nov. 2014.

[18] E. M. Domingues, S. Arunachalam, and H. Mishra, “Doubly reentrantcavities prevent catastrophic wetting transitions on intrinsically wettingsurfaces,” ACS Appl. Mater. Interfaces, vol. 9, no. 25, pp. 21532–21538,Jun. 2017.

[19] E. M. Domingues, S. Arunachalam, J. Nauruzbayeva, and H. Mishra,“Biomimetic coating-free surfaces for long-term entrapment of air underwetting liquids,” Nature Commun., vol. 9, Sep. 2018, Art. no. 3606.

[20] X. Liu et al., “3D printing of bioinspired liquid superrepellent struc-tures,” Adv. Mater., vol. 30, no. 22, May 2018, Art. no. 1800103.

[21] Z. Dong et al., “Superoleophobic slippery lubricant-infused surfaces:Combining two extremes in the same surface,” Adv. Mater., vol. 30,no. 44, Aug. 2018, Art. no. 1803890.

[22] W. Xiong et al., “Simultaneous additive and subtractive three-dimensional nanofabrication using integrated two-photon polymerizationand multiphoton ablation,” Light Sci. Appl., vol. 1, p. e6, Apr. 2012.

[23] J. Li, Q. H. Qin, A. Shah, R. H. A. Ras, X. Tian, and V. Jokinen, “Oildroplet self-transportation on oleophobic surfaces,” Sci. Adv., vol. 2,no. 6, Jun. 2016, Art. no. e1600148.

[24] V. Liimatainen et al., “Mapping microscale wetting variations on bio-logical and synthetic water-repellent surfaces,” Nature Commun., vol. 8,Nov. 2017, Art. no. 1798.

[25] W. S. Y. Wong and A. Tricoli, “Cassie-levitated droplets for distortion-free low-energy solid–liquid interactions,” ACS Appl. Mater. Interfaces,vol. 10, no. 16, pp. 13999–14007, Apr. 2018.

[26] K. Liu, M. Vuckovac, M. Latikka, T. Huhtamäki, and R. H. A. Ras,“Improving surface-wetting characterization,” Science, vol. 363,no. 6432, pp. 1147–1148, Mar. 2019.

[27] A. Marmur and E. Bittoun, “When Wenzel and Cassie are right:Reconciling local and global considerations,” Langmuir, vol. 25, no. 3,pp. 1277–1281, Jan. 2009.

[28] G. Vazquez, E. Alvarez, and J. M. Navaza, “Surface tension of alcoholwater + water from 20 to 50 ◦C,” J. Chem. Eng. Data, vol. 40, no. 3,pp. 611–614, May 1995.

[29] J. Livingston, R. Morgan, and A. J. Scarlett, Jr., “The properties of mixedliquids. IV. The law of mixtures. II,” J. Amer. Chem. Soc., vol. 39, no. 11,pp. 2275–2293, Nov. 1917.

[30] M. Tsige and G. S. Grest, “Surface tension and surface orienta-tion of perfluorinated alkanes,” J. Phys. Chem. C, vol. 112, no. 13,pp. 5029–5035, Mar. 2008.

[31] P. N. Bengesai, W. M. Nelson, P. Naidoo, and D. Ramjuger-nath, “Phase equilibria for perfluoroethane + (n-perfluorohexaneor n-perfluorooctane) binary systems: Measurement and modeling,”J. Chem. Eng. Data, vol. 61, no. 9, pp. 3363–3370, Aug. 2016.

Ville Rontu received the M.Sc. degree in materi-als science from Aalto University, Espoo, Finland,in 2014, where he is currently pursuing the Ph.D.degree in materials science with the Department ofChemistry and Materials Science, School of Chem-ical Engineering.

He is currently a Process Development Engineerwith Aalto Nanofab. His research interests involveatomic layer deposition, thin films, micro- and nano-fabrications, and MEMS.

Ville Jokinen received the Ph.D. degree from theUniversity of Helsinki, Helsinki, Finland, in 2011,with a focus on surface engineering for microfluidicsand mass spectrometry under docent M. Baumann.He is currently a University Lecturer with theDepartment of Chemistry and Materials Science,School of Chemical Engineering, Aalto University.His research interests include surface modification,superhydrophobicity, the chip–cell interface, andbiological applications of microfluidics.

Sami Franssila received the B.Sc. and M.Sc.degrees in physics from the University of Helsinki,Helsinki, Finland, and the Ph.D. degree from theHelsinki University of Technology in 1995. HisPh.D. thesis was on plasma etching. He worked atVTT and imec on CMOS and MEMS technologiesbefore joining Aalto University, Helsinki, in 1998.He is currently a Professor of materials science withAalto University. He has authored or coauthoredover 160 peer-reviewed journal articles and the text-book Introduction to Microfabrication (John Wiley,

2010). His research deals with fabrication, materials, and surface treatmentsfor fluidic, bio, and thermal microdevices.