Published: November 16, 2011

r 2011 American Chemical Society 14726 dx.doi.org/10.1021/la2041168 | Langmuir 2011, 27, 14726–14731

LETTER

pubs.acs.org/Langmuir

Carbon Nanotube-Based Robust Steamphobic SurfacesIla Badge, Sunny Sethi, and Ali Dhinojwala*

Department of Polymer Science, The University of Akron, Akron, Ohio 44325-3909, United States

bS Supporting Information

’ INTRODUCTION

Surfaces that remain dry under a flux of steam are desiredfor many applications, such as the blades of steam turbines andthe surfaces of heat exchangers. Different techniques are em-ployed to create water-repellent surfaces, commonly referredto as superhydrophobic surfaces. They are defined as surfaces onwhich water forms a contact angle of greater than 150� andare formed by a suitable combination of surface chemistry androughness.1�3 However, the ability of a surface to remain dryby letting water droplets roll off depends on the contact anglehysteresis (CAH) defined as the difference between advancingand receding water contact angles on a surface. The major factorinfluencing the CAH on a superhydrophobic surface is the super-hydrophobic state of the water droplets.4,5A droplet of water onsuch a surface may be in a wetting state (Wenzel state) with watercompletely filling the roughness asperities on the surface6 or anonwetting state (Cassie�Baxter state) with the water dropletsupported on the air�solid composite surface.7 In the Cassie�Baxter state, water droplets have lower CAH and easily roll off thesurface as compared to those in the Wenzel state.

Compared to the water repellency of a superhydrophobicsurface where a water droplet is deposited on the surface, inthe case of the condensation of water vapor on such a surface theformation of a water droplet follows a nucleation and growthmechanism where nucleation may be initiated within the pores.This would result in the formation of water droplets in theWenzel state.8 Therefore, a superhydrophobic surface that hasa low contact angle hysteresis with liquid water may not showsimilar behavior in the case of steam condensation. For the surfaceto remain dry under condensation, it is required that the waterdroplets undergo a transition from the Wenzel to the Cassie�Baxter regime. Depending on if this transition is favorablethermodynamically, a typical superhydrophobic surface may or

may not act as an anticondensation surface. Few approachesproposed and tested in the past to create surfaces that canundergo such a transition include the design of surfaces withdifferential intrinsic wettability in the form of either a continuousgradient9 or patches of hydrophilic�hydrophobic regions on thesurface.10 A similar vertical surface energy gradient was proposedto explain the ability of a lotus leaf to remain dry under dewformation.11 The systematic design of roughness features on thesurface such that the nanoscale roughness is present as either partof the structural hierarchy or the surface being nanoporous byitself has also been shown as a design approach to making asurface remain dry under condensation.12�14 However, the exactmechanism behind the wetting behavior of a structured surfaceunder condensation is still not understood completely. All of thecondensation studies reported pertain to the low temperatureand pressure of the condensing water vapor, typically carried outin the range of 0�5 �C. In the case of steam condensation, themechanism of droplet formation is similar to low-temperaturecondensation. However, the high-temperature, high-pressureenvironment of condensing steam to which the surface is ex-posed may exceed the mechanical and hydrostatic stability limitsof the nonwetting state of the surface. Thus, the stability limi-tations make it more difficult to design a steamphobic surfaceeven though control over the surface energy and geometry iseasily achieved with a wide range of materials.

This work creates unique coatings for stainless steel thatremain dry under prolonged exposure to steam and maintain theirsuperhydrophobicity. The coatings are formed using a meshlikecarbon nanotube (CNT) structure. The structure provides the

Received: October 20, 2011Revised: November 14, 2011

ABSTRACT: The wetting behavior of a surface under steam condensation depends on itsintrinsic wettability and micrometer or nanoscale surface roughness. A typical superhydropho-bic surface may not be suitable as a steamphobic surface because of the nucleation and growth ofwater inside the valleys and thus the failure to form an air�liquid�solid composite interface.Here, we present the results of steam condensation on chemically modified nanostructuredcarbon nanotube (CNT) mats. We used a plasma-enhanced chemical vapor deposition(PECVD) process to modify the intrinsic wettability of nanostructured CNT mats. Thecombination of low surface energy achieved by PECVD and the nanoroughness of the surfaceprovides a mechanism to retain the superhydrophobicity of the CNT mats under steamcondensation. The ability to withstand steam temperature and pressure for as long as 10 himplies the remarkably improved stability of the superhydrophobic state of the surface. Thethermodynamic calculations carried out using a unit cell model clearly explain the steamphobicwetting behavior of the surface.

14727 dx.doi.org/10.1021/la2041168 |Langmuir 2011, 27, 14726–14731

Langmuir LETTER

desired surface roughness and porosity. The chemistry of CNTstructures is modified using plasma-enhanced chemical vapordeposition (PECVD), more commonly known as the plasmapolymerization process.

’EXPERIMENTAL SECTION

Meshlike CNTs are grown on acid-treated stainless steel substratesusing a floating catalyst chemical vapor deposition (CVD) process.The CVD process is carried out in a tubular reactor. The stainlesssteel (SS 304) plates are acid treated by etching in 9 M H2SO4 for10 min, followed by washing with deionized water. Xylene was usedas the carbon source, and ferrocene was used as the catalyst. The acid-treated stainless steel plates are heated to 750 �C inside the furnaceunder an argon�hydrogen (85:15 v/v) flow atmosphere. The vaporsof the solution of 1 g of ferrocene in 100 mL of xylene are sublimed at190 �C and introduced into the furnace at a rate of 0.11 mL/min.Further details of the procedure for CNT growth are describedelsewhere.15

The meshlike structure is coated using rf-PECVD (radio frequencyplasma-enhanced chemical vapor deposition). The precursor chosenfor PECVD is 1H,1H,2H-perfluoro-1-dodecene (C10F21�CHdCH2)(97% pure, purchased from Matrix Scientific), which is a representativeof the class of long-chain perfluoro alkyl monomers. The plasma-deposited film of this monomer has been reported to have a low surfaceenergy value of 7.5 mN m�1, making the surface hydrophobic as well asoleophobic.16 The PECVD experiments are carried out in an inductivelycoupled, cylindrical vacuum chamber. The coupling is done through animpedance-matching network that connects the coil wound around thechamber to the radio frequency generator operated at 13.56 MHz. Thereactor is connected to the rough pump via a pressure gauge and a liquidnitrogen cold trap. The glass tube containing the liquid precursor isconnected to the inlet of the chamber, and the control valve is used toadjust and monitor the vapor pressure. For the precursor chosen, thedeposition is carried out at a vapor pressure of 200 mTorr. The chamberis operated in continuous mode at 35 W input power. A typical cycleconsisted of three steps of 5 min each and followed as precursor vapo-rization, plasma ignition, and precursor vaporization. Clean aluminumplates and silicon wafers are used as flat reference substrates.Scanning electron microscopy (SEM) and transmission electron

microscopy (TEM) are used to characterize the CNT mesh. A JEOLJSM-7401F field emission scanning electron microscope was used toacquire SEM images. Images were acquired under a vacuum of 10�6

Torr and a potential difference of 5 kV. No sputter coating was requiredfor the current samples. TEM images were acquired using a JEOLJEM 1230 electron microscope at a potential difference of 120 kV. Thediameter distribution of uncoated and plasma-coated CNTs is obtainedand compared by calculating the diameters of about 130 tubes in eachcase.IR spectroscopy and XPS are used for chemical composition analysis.

The FTS 3000 Excalibur Series DIGILAB IR spectrometer is used intransmission mode. A standard IR liquid cell is used to obtain the IRspectrum of the liquid precursor. The plasma-enhanced films are depo-sited on the compression-molded KBr discs formed with IR-grade KBrpowder (purchased from Sigma-Aldrich), and the spectra are acquired intransmission mode. Sixty-four scans are collected for IR spectra of boththe background and the sample. The XPS spectra are collected using aPHI Quantum 2000 scanning XPS microprobe instrument that uses AlKα radiation. The survey spectra are acquired to detect the presence ofelements on the surface. The survey scans are obtained over the entirerange of binding energies (i.e., 0�1000 eVwith a pass energy of 117.5 eVand a step size of 0.5 eV). The survey scans are run for 5 min. The high-resolution spectra for the C 1s and F 1s regions are obtained over a 20 eVbinding energy width (i.e., 280�300 eV for C 1s and 680�700 eV for

the F 1s region). The high-resolution spectra in the C 1s region areacquired at an 11.75 eV pass energy and a 0.05 eV step size. The F 1sregion spectra are obtained at a 58.7 eV pass energy and a 0.125 eV stepsize. The high-resolution scans in the C 1s and F 1s regions are run for10 min. These spectra are also used for chemical quantification.

The wetting behavior of plasma-coated (PCNT) and uncoated(CNT) mesh is studied by contact angle analysis. CNT and PCNT sur-faces are exposed to steam and analyzed visually. Immediately afterexposure to steam, contact angles are measured on the surfaces at roomtemperature. The contact angle measurements are carried out usingRam�e-Hart Instruments Advanced Goniometer 500 F1 with DropImage Advanced software. The contact angle is measured with an 8�10 μL droplet of deionized water. The contact angle hysteresis ismeasured by tilting the base until the drop rolls off the surface. Thesteam condensation experiments are carried out by exposing the sampleto a continuous stream of condensing steam. Steam is generated in asteam chamber composed of stainless steel and heated using resistiveheating. The temperature of the chamber is monitored using a J-thermo-couple and kept constant for all of the measurements. The chamber isfed with water at a continuous rate. As the water enters the heatedchamber, it boils. This generated steam generated is carried through astainless steel pipe to the sample block in which the sample plates areheld horizontally. The entire assembly is thermally insulated tominimizethe loss of heat to the atmosphere. The sample holder block is providedwith a thermocouple, and the temperature inside the holder is mon-itored continuously. The samples are exposed to condensing steam atatmospheric pressure and a temperature of 96 ( 2� throughout theexperiment. Immediately after steam exposure, the water contact angle ismeasured and plotted as a function of the steam exposure time.

The CNT mats grown on stainless steel plates show nonalignedmeshlike morphology, as can be seen in the SEM image (Figure 1a). In3D space, themorphology of the layer of CNTs formed can be visualizedas a stack of many of such entangled meshes as shown schematically inFigure 4a. Calculations for fs used in the Cassie�Baxter model are basedon a rectangular unit cell shown in Figure 4b. Figure 4b represents thesingle topmost layer of a meshlike structure. In the Cassie�Baxter state,

Figure 1. Electron microscope images of the CNT mesh structure.(a) SEM image of CNTs. (b, c) TEM images of uncoated and plasma-coatedCNTs, respectively. (d) Diameter distribution of uncoated (CNT)and plasma-coated CNTs (PCNT).

14728 dx.doi.org/10.1021/la2041168 |Langmuir 2011, 27, 14726–14731

Langmuir LETTER

the water droplet is assumed to be in contact with the projection of thissingle layer and not the entire thickness of the CNT mesh. The averagelength t and the average width h of a rectangular cavity enclosed by fournanotubes are determined from SEM images using ImageJ software. Ineach SEM image, about 20 rectangular unit cells are chosen at differentpositions, an image is rotated as required to be able to draw a rectanglethat fits the given unit cell, and the values of t and h are calculated. Thesame procedure is repeated for about 8�10 SEM images to calculateaverage values and standard deviations.

’RESULTS AND DISCUSSION

The SEM image in Figure 1a shows the structure of as-grownCNTs on stainless steel (SS) plates. The morphology of CNTmats can be described as an entangled nanoporous structure.This morphology is retained even after the deposition of theplasma coating. The PCNT structures are not easily distinguish-able from CNT structures, as can be seen by comparing the SEMimages (Figure 1a and Supporting Information). The thicknessof the plasma-coated films on CNT is estimated by the analysis ofthe distribution of diameters of CNTs. TEM is used to image theCNT structure (as shown in Figure 1b,c), and Image J software isused to analyze the tube diameters. Figure 1d compares theoverall size distribution of CNTs and PCNTs. It can be seen thatthe minimum diameter range for CNTs is 5�10 nm, whereasfor PCNTs it is 10�15 nm. The peak in the diameter distributionof CNTs lies in the 15�20 nm range. In the case of PCNTs,it is seen to be in the 20�25-nm-diameter range. From theseobservations, the thickness of the plasma-deposited coating canbe approximated to be about 5�10 nm, which is not easily detec-table with the resolution limits of the imaging instruments usedand given the nonuniformity of the tube diameters. Thus, the sizedistribution analysis gives a good estimate of the coating thick-ness, and the thin layer of the plasma coating serves the purposeof surface chemistry modification without significantly affectingthe nanostructure of the CNTs. A thin layer of an organic coatingis also desirable for preventing any significant effects on heat-transfer coefficients.

To confirm the presence of plasma-coated films on nanos-tructures further, the IR spectra of the 1H,1H,2H-perfluoro-1-dodecenemonomer and the coating formed by plasma-enhanceddeposition are compared (Figure 2a). The IR spectrum of theplasma-deposited coating shows that there is a loss of character-istic alkene C�H bending vibrations in the 1000�650 cm�1

region. This implies the activation of the double bond of themonomer in the plasma state. The characteristic perfluoroalkylchain stretching bands are observed in the 1110�1280 cm�1

(CF2) and 1110�1340 cm�1 (CF3) regions in both the mono-mer and the plasma-deposited coating. The broadening ofthe peaks in this region for the plasma-deposited coating canbe attributed to the lower structural retention of the monomerarising from the fragmentation, rearrangement, and cross-linkingreactions that take place in the plasma-enhanced depositionprocess.16�18

The quantitative chemical analysis of the plasma-depositedcoating is carried out with XPS. Figure 2b shows the C 1s spectraof uncoated CNT mats and 1H,1H,2H-perfluoro-1-dodeceneplasma-coated CNT mats. The C 1s spectra of the same plasmacoating on the aluminum and silicon surfaces that are used asreference flat substrates are also shown in Figure 2b. The spectraare fitted to a Gauss�Lorentz function. The main peak in thecase of uncoatedCNTs, which shows up at around 284 eV, can be

assigned to the graphitic C�C bond in a carbon nanotube. Thespectra for coated aluminum and silicon surfaces look almostidentical and do not show any peak at 284 eV. All four peaksobserved for the coating on aluminum and silicon can be assignedto the characteristic binding-energy range of C�Fn bonds.

16,17,19

The percent retention of CF2 (291 eV) and CF3 (293 eV) at thesurfaces can be estimated by integrating the area under the curve.This is particularly significant because the presence of thesefunctional groups on the surface is responsible for making theCNT surface hydrophobic (i.e., modifying its intrinsic wettabi-lity).20 In the case of PCNTs, an additional peak at around284 eV is assigned to the C�C bond of CNT. The presence ofthis peak along with other C�Fn peaks implies that the thicknessof the coating on CNT should be less than the typical analysisdepth of about 10 nm for XPS. Therefore, the thickness of thecoating that is estimated is consistent with the results obtainedwith the size distribution analysis of the tubes. The comparablevalues obtained from the functionality retention and F/C atomicratios of plasma-coated aluminum, silicon, and CNTs are con-sistent with the fact that PECVD was carried out under identicalconditions for the deposition of the coating on all of the surfaces.The F 1s spectra obtained for all of the samples show a singlepeak at about 688 eV that can be assigned to the C�F bond(corresponding data in Supporting Information). XPS analysisthus confirms the presence of a thin layer of a plasma-depositedcoating on the carbon nanotube surface.

CNTmats show a water contact angle of 151( 3� and a CAHof 5�. The water contact angle on the PCNT surface is 156( 2�,and the CAH is 2�. Thus, there is no significant difference in the

Figure 2. (a) IR spectra of the monomer and the coating formed byPECVD. (b) XPS C 1s spectra of uncoated (CNT) and plasma-coated(PCNT) CNTs and plasma-coated Al and Si.

14729 dx.doi.org/10.1021/la2041168 |Langmuir 2011, 27, 14726–14731

Langmuir LETTER

wetting behavior of CNT and PCNT in the water contact angletest, and both surfaces can be characterized as superhydrophobic.Figure 3b,d shows optical images of water droplets on CNT andPCNT surfaces.

Under steam condensation, however, it is observed that thetwo surfaces are drastically different in their wetting properties.The experiments consist of exposing the CNT and PCNTsurfaces (held horizontally) to the condensing steam flux and,following steam exposure, measuring the water contact angleby depositing a water droplet on the surface. Upon exposureto steam, it is seen that the condensate forms as small sphericaldroplets on the PCNT surface, as shown in Figure 3a. Thenonuniform coverage of these droplets on the surface may implythe removal of larger droplets by coalescence.12,21 The dropletsize continues to shrink because of evaporation, and eventuallyno droplets are seen on the surface. When the PCNT surface ismeasured for its water contact angle, it is seen that the surface issuperhydrophobic even after it is exposed to steam (Figure 3e).However, in the case of an uncoated CNT mat, within the time-scale of minutes, a thin film of condensed water appears to beformed upon steam exposure. Even though the film is not clearlyvisible, the complete wetting of the deposited droplet followingsteam exposure on this surface implies the presence of a thinwater film. Thus, the uncoated CNT surface loses its superhydro-phobicity upon exposure to steam (Figure 3c). We also observethat the superhydrophobicity of the CNT surface is restored afterit is dried off completely. Figure 3f shows the contact anglemeasured on PCNTmats after prolonged exposure to steam as afunction of exposure time. Interestingly, it is observed that thesuperhydrophobicity of the PCNT surface is retained even aftersteam exposure of as long as 10 h, whereas the CNT surface losesits superhydrophobicity when exposed to steam for only a fewminutes.

Differences in the wetting behavior of PCNT and CNT matsunder steam condensation can be explained on the basis ofthe modification of the surface chemistry and thus the intrinsicwettability (θY) of CNTs. The intrinsic contact angle of thesurface of a single CNT has been measured to be about 80�.22After plasma coating, we estimate the intrinsic contact angle ofthe PCNT surface to be 110� from the contact angles measuredfor plasma-coated flat plates. We propose here a model to predictwhether the Cassie�Baxter or the Wenzel states are thermo-dynamically favored for CNT or PCNT surfaces. The free energyof the given wetting state as a function of the corresponding

apparent contact angle (θ) and intrinsic wettability (θY) can beexpressed in its dimensionless expression form.23 For a Wenzelstate, the free energy can be written as

G�W ¼ ½FðθWÞ��2=3½2� 2 cos θW � r sin2 θW cos θY�ð1Þ

where θW is the Wenzel contact angle that is defined as follows:6

cos θW ¼ r cos θY ð2ÞHere, r is the surface roughness defined as the ratio of the actualsurface area to the projected surface area (r > 1). F(θ) is definedas

FðθÞ ¼ ð2� 3 cos θ þ cos3 θÞ ð3ÞSimilarly, the expression for the free energy of the Cassie�Baxterstate can be written as

G�CB ¼ ½FðθCBÞ��2=3½2� 2 cos θCB � sin2 θCBð� 1 þ f s þ f s cos θYÞ�ð4Þ

Here, θCB is the Cassie�Baxter contact angle defined as7

cos θCB ¼ � 1 þ f sð1 þ cos θYÞ ð5Þfs is the fraction of solid in contact with water when the droplet issupported on an air�solid composite surface. The free -energydifference between the Cassie�Baxter and Wenzel states can bedefined as ΔG* = G*CB� G*W. We use a model rectangular unitcell as shown in Figure 4b to estimate fs. Using SEM and imageanalysis, we calculate fs = 0.204( 0.05. On the basis of the valuesof fs, the model predicts the θCB value to be 140 ( 5� for CNTsurface and 150 ( 4� for PCNT surface. It can be seen that theθCB values calculated using the Cassie�Baxter model are closeto the static contact angle measured experimentally. Because0�e θWe 180�, G*W and thusΔG* can be defined only for 1ere 2.92. The upper limit on r is imposed byθY = 110� for PCNT.In the range of rwhere the real θW value is defined, it can be seen(Figure 4c) thatΔG* for a CNT surface is an increasing functionof r, implying that the Wenzel state becomes more favorable asthe surface roughness increases for θY = 80�. In the case of thePCNT surface with θY = 110�, ΔG* decreases as r increases andcrosses the ΔG* = 0 line for higher values of r. The negativevalues of ΔG* imply that the Cassie�Baxter state becomes the

Figure 3. (a) Condensed steam droplets on the PCNT surface after steam exposure for 1 h. Contact angles measured on (b) CNT (uncoated) and (d)PCNT (plasma-coated) surfaces. Following steam condensation, the water contact angles weremeasured on (c) CNT and (e) PCNT surfaces. (f)Watercontact angle measured on PCNTs as a function of the steam exposure time.

14730 dx.doi.org/10.1021/la2041168 |Langmuir 2011, 27, 14726–14731

Langmuir LETTER

thermodynamically more favorable state as the surface roughnessis increased. For these CNT and PCNT surfaces, we expect r tobe high, most likely greater than 3. Thus, for the nanoporousmeshlike PCNT surface the Cassie�Baxter state with a highcontact angle is more favored whereas for the CNT surface theWenzel state with a low contact angle is more favored thermo-dynamically. This implies that even though the Cassie�Baxterwetting state is possible for both CNT and PCNT surfaces(on the basis of estimated values of fs), thermodynamically itrepresents only a metastable state for the CNT surface. Aftersteam condensation, when a droplet in the more stable Wenzelstate is formed on this surface, its failure to transition to theCassie�Baxter state results in the loss of superhydrophobicity.However, even if a condensate forms aWenzel state droplet uponexposure to steam on the PCNT surface, the transition from theWenzel to the Cassie�Baxter state is favored in this case. Thisexplains the retention of the superhydrophobicity of a PCNTsurface even after prolonged exposure to steam.

Steam condensation involves the nucleation and coalescenceof the condensed droplets. From the classical theory of thenucleation and growth of a condensate, it is known that nuclea-tion rates are dependent on the intrinsic wettability (θY) of thesurface. It is estimated that the free-energy barrier that is required

to be overcome for the condensation of water vapor is higher andcondensation rates are lower on a hydrophobic surface comparedto those on a hydrophilic surface.9,10 The hydrophobic surfacesdelay the nucleation but do not completely eliminate this pro-cess. During coalescence, if the Wenzel state is preferred thenthe surfaces will exhibit lower contact angles than the metastableCassie�Baxter state and this is the case for the CNT sampleswithout the fluorinated coatings. However, if the Cassie�Baxterstate is a true thermodynamically stable state, then the coales-cence droplets will be expelled outward and the superhydro-phobic Cassie�Baxter state will be preferred. Therefore, fluori-nated coatings on CNT result in lower nucleation rates as well asa stable Cassie�Baxter state to keep the surfaces dry and stableduring steam condensation.

It can thus be concluded here that two surfaces that do notdiffer much structurally and are characterized as superhydropho-bic surfaces exhibit extreme steamphobic behavior. The combi-nation of the nanoscale roughness of the CNT mats along withthe surface energy contrast achieved by the deposition of a low-surface-energy plasma-deposited coating results in the formationof a highly robust steamphobic surface. The biggest advantage ofthe PECVD process used here for chemical modification is thatit is a completely dry process. Unlike wet processes, in which thecapillary forces tend to collapse the nanoporosity, the PECVDprocess enables the deposition of a very thin layer of a film onsuch a surface without disturbing its structural integrity. The useof stainless steel to form the steamphobic coatings is an addedadvantage because it is a widely used industrial material. Therobust steamphobic surfaces are of particular interest in design-ing heat exchangers. They are ideal for dropwise steam conden-sation and can be expected to improve the exchanger efficiencyremarkably because of the fast droplet removal and low adhesionof condensed steam to the surfaces. The robustness of thestructure can also be taken advantage of in antifog, antiice, andantidew coatings.

’ASSOCIATED CONTENT

bS Supporting Information. Additional data are providedfor the following systems: SEM data for PCNT; XPS data forfluorinated coatings on CNT and other model substrates; andTGA data for CNT and PCNT surfaces. This material is availablefree of charge via the Internet at http://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: ali4@uakron.edu.

’ACKNOWLEDGMENT

We acknowledge financial support from NSF-DMR-0512156(A.D.), the Nanoscale Interdisciplinary Research Teams grantfrom NSF-0609077 (A.D.), and the Goodyear Tire and RubberCompany for an industrial fellowship (I.B.).

’REFERENCES

(1) McHale, G.; Shirtcliffe, N. J.; Newton, M. I. Analyst 2004,129, 284.

(2) Qu�er�e, D. Rep. Prog. Phys. 2005, 68, 2495.(3) Wang, S.; Jiang, L. Adv. Mater. 2007, 19, 3423.

Figure 4. (a) Schematic showing the CNTmesh structure. It can be vi-sualized as multiple meshes stacked on top of each other. (b) Schematicof the rectangular unit cell model. It has inner side dimensions of (t� h)and outer dimensions of (t + d)� (h + d), where t and h are the averagelength andwidth, respectively, of a rectangular cavity between nanotubesin a mesh structure and d is the average diameter of the CNT. (c) ΔG*calculated as a function of r for θY = 80� (CNT) and θY = 110� (PCNT);1e re 2.9. The calculation ofΔG* (G*CB�G*W) depends on the valueof fs. We calculated the value of fs to be 0.204 ( 0.05 using therectangular cell model. At every point, G*CB for the correspondingfs value was calculated, and the difference between G*CB and G*W wasplotted as a function of r. The range of fs determines the lower and upperlimits for ΔG* in the cases of both CNT and PCNT as shown byhighlighted regions in the graph. Therefore, for any given r, the range ofΔG* can be estimated from the model calculations here.

14731 dx.doi.org/10.1021/la2041168 |Langmuir 2011, 27, 14726–14731

Langmuir LETTER

(4) McHale, G.; Shirtcliffe, N. J.; Newton, M. I. Langmuir 2004,20, 10146.(5) Lafuma, A.; Qu�er�e, D. Nat. Mater. 2003, 2, 457.(6) Wenzel, R. N. Ind. Eng. Chem. 1936, 28, 988.(7) Cassie, A. B. D.; Baxter, S. Trans. Faraday Soc. 1944, 40, 546.(8) Narhe, R. D.; Beysens, D. A. Phys. Rev. Lett. 2004, 93, 076103.(9) Daniel, S.; Chaudhury, M. K.; Chen, J. C. Science 2001, 291, 633.(10) Varanasi, K. K.; Hsu, M.; Bhate, N.; Yang, W.; Deng, T. Appl.

Phys. Lett. 2009, 95, 094101.(11) Zheng, Y.; Han, D.; Zhai, J.; Jiang, L. Appl. Phys. Lett. 2008,

92, 084106.(12) Dorrer, C.; R€uhe, J. Adv. Mater. 2008, 20, 159.(13) Lau, K. K. S.; Bico, J.; Teo, K. B. K.; Chhowalla, M.;

Amaratunga, G. A. J.; Milne, W. I.; McKinley, G. H.; Gleason, K. K.Nano Lett. 2003, 3, 1701.(14) Chen, C.-H.; Cai, Q.; Tsai, C.; Chen, C.-L.; Xiong, G.; Yu, Y.;

Ren, Z. Appl. Phys. Lett. 2007, 90, 173108.(15) Sethi, S.; Dhinojwala, A. Langmuir 2009, 25, 4311.(16) Laguardia, L.; Ricci, D.; Vassallo, E.; Cremona, A.; Mesto, E.;

Grezzi, F.; Dellera, F. Macromol. Symp. 2007, 247, 295.(17) Coulson, S. R.; Woodward, I. S.; Badyal, J. P. S. Langmuir 2000,

16, 6287.(18) He, P.; Shi, D.; Lian, J.; Wang, L.M.; Ewing, R. C.; vanOoij, W.;

Li, W. Z.; Ren, Z. F. Appl. Phys. Lett. 2005, 86, 043107.(19) Coulson, S. R.; Woodward, I. S.; Badyal, J. P. S. Chem. Mater.

2000, 12, 2031.(20) Hare, E. F.; Shafrin, E. G.; Zisman, W. A. J. Phys. Chem. 1954,

58, 236.(21) Dietz, C.; Rykaczewski, K.; Fedorov, A. G.; Joshi, Y. Appl. Phys.

Lett. 2010, 97, 033104.(22) Barber, A. H.; Cohen, S. R.; Wagner, H. D. Phys. Rev. Lett. 2004,

92, 186103.(23) Marmur, A. Langmuir 2003, 19, 8343.