Journal of Colloid and Interface Science 392 (2013) 369–375

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Thermodynamics of hydrophobic interaction between silica surfaces coatedwith octadecyltrichlorosilane

Zuoli Li, Roe-Hoan Yoon ⇑Center for Advanced Separation Technologies, Virginia Tech, Blacksburg, VA 24060, United States

a r t i c l e i n f o

Article history:Received 30 April 2012Accepted 13 August 2012Available online 11 October 2012

Keywords:Hydrophobic forceExcess thermodynamic propertiesDerjaguin approximationStructural forceAtomic force microscope (AFM)Octadecyltrichlorosilane (OTS)

0021-9797/$ - see front matter � 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.jcis.2012.08.078

⇑ Corresponding author. Address: 146 Holden Hall,States. Fax: +1 540 231 3948.

E-mail address: ryoon@vt.edu (R.-H. Yoon).

a b s t r a c t

Surface force measurements conducted with thiolated gold surfaces showed previously that hydrophobicinteraction entails a decrease in excess film entropy, suggesting that hydrophobic force originates fromchanges in the structure of the medium (water) confined between hydrophobic surfaces [1]. As a fol-low-up work, surface force measurements have been conducted in the present work using an atomicforce microscope (AFM) with silica surfaces coated with octadecyltrichlorosilane (OTS) at temperaturesin the range of 10–40 �C. A thermodynamic analysis of the results show that both the excess film entropy(DSf) and excess film enthalpy (DHf) decrease with decreasing thickness of the water films between thehydrophobic surfaces. It has been found also that |DHf| > |TDSf|, which represents a necessary conditionfor the excess free energy change (DGf) to be negative and hence the hydrophobic interaction be attrac-tive. Thus, the results obtained with both the thiolated and silylated surfaces show that hydrophobicforces originate from the structural changes in the medium. It is believed that the water molecules inthe thin liquid films (TLFs) of water form clusters as a means to reduce the free energy when they cannotform H-bonds to neighboring hydrophobic surfaces.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

Hydrophobic particles placed in water are attracted to eachother much more readily than predicted by the DLVO theory dueto the presence of the hydrophobic force [2,3], which is longer ran-ged and stronger than the van der Waals force. Israelachvili andPashley [4] reported the first direct measurement of hydrophobicforce using the surface force apparatus (SFA). Many investigators[5–10] confirmed its existence and discussed possible origins,while others suggested that the hydrophobic force is an artifactdue to bubbles or cavities [11–13]. Still others suggested thathydrophobic forces, particularly those of longer range, may origi-nate from mechanisms that are unrelated to surface hydrophobic-ity [14,15].

In an effort to better understand the origin(s) of hydrophobicforce, one of us determined the thermodynamic functions of thehydrophobic interactions by conducting surface force measure-ments at several different temperatures [1]. The results showedthat macroscopic hydrophobic interactions entail decreases in boththe excess entropy (Sf) and the excess enthalpy (Hf) of the thin li-quid films (TLFs) of water confined between hydrophobic surfaces.Here, the term ‘‘excess’’ refers to the thermodynamic indicator of

ll rights reserved.

Blacksburg, VA 24060, United

the thin film in question relative to that of the infinitely thick film(or bulk water). It was found also that the changes in excess filmenthalpy (DHf) are slightly larger than the absolute temperature(T) times the changes in excess film entropy (DSf). Based on theseresults, it was suggested that macroscopic hydrophobic interac-tions may involve building structures of water in the vicinity ofhydrophobic surfaces. In effect, DHf represents the energy gainedin building the structures, while TDSf represents the thermody-namic cost of building the structure. In this previous work, the sur-face force measurements were conducted using an atomic forcemicroscope (AFM) with gold-coated silica surfaces hydrophobizedwith n-ethanethiol (C2–SH) and n-hexadecane thiol (C16–SH).

The thermodynamic studies described above suggested thathydrophobic force is a structural force, a term first used by Derja-guin and Kusakov [16] to describe the repulsive hydration forcepresent in wetting films. In general, the term structural force refersto the non-DLVO force created when two boundary layers, whosestructures are different from that of the bulk solution, overlap[17]. The non-DLVO forces observed between two hydrophilic sur-faces are referred to as positive structural force, while those ob-served between hydrophobic surfaces are referred to as negativestructural force [18]. Eriksson et al. [19] suggested that the long-range attractive forces observed between hydrophobic surfacesmay be due to the surface-induced changes in water structure.

In the present work, we have conducted AFM force measure-ments using silica surfaces hydrophobized with octadecyltrichlo-

370 Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375

rosilane (OTS) at temperatures in the range of 10–40 �C. The resultshave been converted to the excess Gibbs free energies (DGf) of theTLFs using the Derjaguin approximation [20], which were thenused to determine DSf and DHf in the same manner as describedpreviously [1].

Silylated silica may provide a more stable and robust hydropho-bic surface than thiolated gold, as the latter is susceptible to oxida-tion. It has been shown that upon exposure to air under ambientconditions, self-assembled monolayers (SAMs) of alkane thiolatesoxidize to alkane sulfonates [21,22]. In the presence of halide,the thiol groups oxidize to disulfide and subsequently to sulfonate,with the substrate (Au) being oxidized to Au3+ species [23]. Stabil-ity of hydrophobic surfaces using thiolated gold may thus becomean issue particularly when conducting force measurements at ele-vated temperatures. On the other hand, OTS-coated silica surfacesare stable at temperatures up to 175 �C [24].

Surface force data obtained with silylated surfaces and reportedin the literature are controversial. Some investigators showedstrong hydrophobic forces [25,26], while others showed capillaryforces related to bubbles or cavities [11,27,28]. The experimentaldata obtained in the present work showed the presence of long-range hydrophobic forces with no evidences for bubbles or cavita-tion causing them. Thermodynamic analysis of the results led tothe same conclusion as obtained previously using thiolated goldsurfaces: hydrophobic forces originate from changes in the waterstructure [1].

2. Experimental

2.1. Materials

All experiments were conducted using ultrapure water obtainedusing a Direct-Q3 water purification system from Millipore. Thewater had a resistivity of 18.2 MO�cm and <10 ppb of total organiccarbon. Silica spheres (Whitehouse Scientific) of �10 lm radiusand fused-quartz plates (Technical Glass Products Inc.) were usedas macroscopic surfaces for surface force measurements. Thespheres and plates were hydrophobized by immersion in octade-cyltrichlorosilane (OTS, 95% purity, Alfa Aesar)-in-toluene (99.9%purity, Fisher Chemical, dried with a molecular sieve) solutions.The hydrophobized surfaces were rinsed with chloroform (99.9%purity, Fisher Chemical), acetone (99.9 + % HPLC grade, Aldrich),and ultrapure water sequentially. Silicon wafers (Sumco, Oregon)were used for imaging the OTS-coated surfaces. Sulfuric acid(H2SO4, 98% purity, VMR international) and hydrogen peroxide(H2O2, 29.0–32.0% purity, Alfa Aesar) were used to clean the mac-roscopic surfaces before hydrophobization with OTS.

2.2. Sample preparation

The macroscopic surfaces used in the present work werecleaned in piranha solutions (H2O2/H2SO4, 3:7 by volume) at�80 �C for 1 h to remove organic contaminants. The plates werethen rinsed thoroughly with ultrapure water in an ultrasonic bathfor 10 min and blow-dried in a pure nitrogen stream.

The cleaned surfaces were hydrophobized by immersing themin freshly prepared OTS-in-toluene solutions. Special care was ta-ken to remove water from the solvent. The hydrophobicity of thesilylated surfaces, as measured by water contact angles, was con-trolled by varying the immersion time in a 5 � 10�5 or3 � 10�3 M OTS-in-toluene solutions contained in TFPE beakers.The excess OTS present on silylated surfaces was removed by sub-jecting the plates and spheres to ultrasonic vibration sequentiallyin chloroform, acetone, and pure water for a few minutes. The sily-

lated surfaces free of excess OTS were then blow-dried with purenitrogen gas and stored.

2.3. AFM imaging

A Nanoscope V atomic force microscope (AFM), Veeco Instru-ments Inc., was used to capture the images of the silylated surfacesunder contact mode in air. A silicon nitride AFM cantilever (Nanoworld Innovative Technologies) with a spring constant of�0.48 N/m was used.

2.4. Surface force measurement

The Nanoscope V AFM was also used to measure the surfaceforces acting between silylated silica sphere and silica plate inultrapure water. The AFM was equipped with a temperature con-trol system, which allowed the measurement of surface forces atdifferent temperatures. The measurement was conducted in therange of 10–40 �C in the same manner as described previously[1]. For a given series of measurements, a silylated silica spherewas glued onto the tip of a cantilever (AI silicon probe, Budget Sen-sors) with a spring constant (k) of �7.4 N/m using a polymer resin(EPON-1004F, Shell Chemical Co.). The resonant frequency tech-nique [29] was used to accurately determine the spring constantfor each cantilever.

2.5. Contact angle measurement

Droplets of ultrapure water were placed on the silylated fused-silica and silicon wafer surfaces to determine the contact anglesusing the sessile drop technique. A Rame-Hart goniometer wasused for the measurements. Advancing and receding angles weremeasured by changing the volume of the ultrapure water dropsby means of a Microliter syringe. For a given surface, five differentmeasurements were conducted and averaged.

3. Results and discussion

3.1. OTS monolayer on silica

The AFM images of the fused-silica plates used in the presentwork showed grains of �100 nm in dimension, and the distinctboundaries between them made it difficult to see the surfactantmolecules adsorbed on the surface [30]. Further, some of theimages showed lines representing scratches created during polish-ing. Therefore, we used silicon wafers for imaging OTS-coated sur-faces. Silicon wafers readily oxidize to form silicon dioxide [31];therefore, the information derived using the wafer may be relevantto the surface force data obtained using the silylated silica surfaces.

Fig. 1 shows the water contact angles of ultrapure water placedon the surfaces of silicon wafers treated with OTS-in-toluene solu-tions at 5 � 10�5 and 3 � 10�3 M. Table 1 shows the data obtainedat the lower concentration. As anticipated, both the advancing (ha)and receding angles (hr) increased with contact time, and the kinet-ics of adsorption was much faster at the higher concentration. Alsoshown is the contact angle hysteresis at the two different concen-trations. In both cases, the hysteresis decreased with increasingcontact time, indicating that as the adsorption density of OTS in-creased, the adsorption layers became smoother.

Fig. 2 shows the AFM images of the silicon wafers contactedwith 5 � 10�5 M OTS-in-toluene solutions for different time peri-ods to obtain equilibrium contact angles (he) of 30�, 80�, 93� and101�. Untreated surface exhibited an equilibrium contact angle of4�. The images were captured in air with a scan area of 1 � 1 lmand a vertical color scale of 3 nm. The vertical profiles shown under

0

40

80

120

0 200 400 600 800 1000

0 20 40 60 80 1000

40

80

120

θh

θr

(a)5×10-5 M OTSθa

θh

θr

θa

3×10-3 M OTSCon

tact

ang

le (°

)

Immersion time (min)

(b)

Fig. 1. Effects of immersion time on the water contact angles of silicon wafer in (a)5 � 10�5 M and (b) 3 � 10�3 M OTS-in-toluene solutions. ha and hr representadvancing and receding angles. The contact angle hysteresis (hh) decreases withtime indicating formation of smoother OTS monolayers. OTS adsorbs much faster atthe higher concentration.

Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375 371

the images represent the changes in height along the horizontallines drawn in the images. The bare silicon surface with he = 4�showed RMS roughness of 0.11 nm, demonstrating that the siliconwafer surface was smooth (Fig. 2a).

Table 1Water contact angles and surface roughness of the silicon wafer surfaces treated in a 5 �

Contact time (min) Contact angle (�)

he ha hr

20 30 35 25200 80 85 67.5380 93 98 88720 101 104 95

Fig. 2. Contact mode 1 � 1 lm AFM images of the silicon wafers of different equilibriumobtained by varying the immersion times in a 5 � 10�5 M OTS solution. Height profilesadsorbs in patches, which coalesce to obtain a smooth monolayer.

After a short contact time of 20 min, he became 30� (Fig. 2b) andsmall spots of �20 lm diameter appeared. These spots indicatedthat OTS adsorbs in clusters due to lateral interactions. As he in-creased to 80� (Fig. 2c), the clusters became more abundant, larger(40–100 nm), and taller as shown in the vertical profiles. It appearsthat as he increased to 93� (Fig. 2d) and then to 101� (Fig. 2e) withincreasing contact time, the clusters coalesced and became larger,eventually creating a smooth monolayer.

The series of AFM images presented in Fig. 2 shows that themonolayer coating has a thickness of approximately 2.5 nm, aswas also reported by other investigators [32,33]. This value is closeto the length of 18 carbon alkane chain, indicating that the OTScoatings obtained in the present work were monomolecular layers.It is believed that the key to forming a smooth monolayer was toensure that the solvent (toluene) was completely dehydrated be-fore use. The adsorption layers formed in the presence of waterproduced much thicker (�200 nm) layers [34], possibly indicatingthe formation of polymeric silanes.

3.2. Surface forces and contact angle

In Fig. 3, surface forces (F) measured between silylated silicaplates and spheres normalized by the radii (R) of the spheres areplotted vs. thickness (h) of the TLF of the water formed betweenthe two macroscopic surfaces. In each experiment, a pair of silicaplate and sphere was hydrophobized in an OTS solution(5 � 10�5 M in toluene) simultaneously, so that the two surfaceshave identical contact angles. Fig. 3 shows the results obtainedwith silica surfaces of he = 50.0�, 78�, and 92.5�. The contact angleswere varied by controlling the immersion time in the OTS solution.As shown, all of the force curves exhibited net attractive forcesstronger than the van der Waals force with a Hamaker constant(A131) of 0.8 � 10�20 J [35]. Note here that the force curve obtainedat he = 92.5� was smooth, showing no evidence for bubbles or vaporcavities affecting the measurement. Cavitation is thermodynami-

10�5 M OTS solution at different contact times.

Contact angle hysteresis (�) Surface roughness (nm)

10.0 0.1817.5 0.4610.0 0.22

9.0 0.11

contact angles. (a) Bare surface with he = 4�, (b–e) surfaces with different he values asgiven below each image were obtained along the white line across the image. OTS

0 10 20 30 40-2

-1

0

F/R

(mN

/m)

h (nm)

50.0°

78.0°

θe=92.5°vdWforce

Fig. 3. Surface forces (F) normalized by the radius (R) of spheres vs. the closestseparation distance (h). The surfaces were treated in a 5 � 10�5 M OTS solution fordifferent times to obtain different contact angles. Solid lines represent the extendedDLVO plots with w1 = �40 mV and A131 = 0.8 � 10�20 J. The double-layer and vander Waals forces are negligibly small particularly at longer separations. Thus, themeasured forces practically represent the hydrophobic forces, which are shown toincrease with he. The arrows represent the separations at which two surfacesjumped into contact.

0 10 20 30 40-2.0

-1.5

-1.0

-0.5

0.0

0.5θe= 92.5°

10 20 30

F/R

(mN

/m)

h (nm)

t (°C)

vdWforce

Fig. 4. Surface forces (F) measured at different temperatures in the range of 10–30 �C and normalized by the radius of sphere (R) vs. separations (h). The silicasurfaces were treated in a 5 � 10�5 M OTS solution to obtain he = 92.5�. Themeasured forces became stronger and longer-ranged with decreasing temperatureof the medium.

0 10 20 30 40-2.0

-1.5

-1.0

-0.5

0.0

0.5

10 20 30 40

F/R

(mN

/m)

h (nm)

θe= 78.0°

t (°C)

vdW force

Fig. 5. Surface forces (F) measured at different temperatures in the range of 10–40 �C and normalized by the radius of sphere (R) vs. separations (h). The silicasurfaces were treated in a 5 � 10�5 M OTS solution to obtain he = 78�. The measuredforces became stronger and longer-ranged with decreasing temperature of themedium.

372 Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375

cally possible at contact angles above 90�. If bubbles and/or cavi-ties were the causes for the non-DLVO attractive forces [11], theforces measured at he < 90� should be equal to the van der Waalsforce. The experimental results were contrary as shown in Fig. 3.Further, the attractive forces increased with increasing contact an-gle, indicating that the non-DLVO forces shown in Fig. 3 are relatedto the hydrophobicity of the surfaces used in the measurements.

The arrow on each curve represents the distance at which asphere jumps into contact with a plate due to the presence of theattractive hydrophobic force in the TLF confined between the mac-roscopic hydrophobic surfaces. The solid lines in Fig. 3 representthe surface forces calculated using the extended DLVO theory,which includes contributions from the van der Waals force, electri-cal double-layer force, and the hydrophobic force. The hydrophobicforce is represented by the single-exponential form,

FR¼ �C exp � h

D

� �ð1Þ

where C and D are fitting parameters. The double-layer force wascalculated using the surface potential of �40 mV for the OTS-coatedsilica surfaces, which was assumed to be the same as the f-potentialmeasured in the present work using the Zetasizer Nano-ZS,Malvern. We found that the f-potentials did not change signifi-cantly with the OTS concentrations used in the present work forhydrophobization. It was found that contributions from van derWaals and double-layer forces were minimal as compared to thosefrom hydrophobic forces.

3.3. Effect of temperature on hydrophobic force

Figs. 4 and 5 show the surface forces measured with OTS-coatedsilica surfaces in ultrapure water at temperatures in the range of10–40 �C. The objective of these measurements was to determinethe thermodynamic functions of the macroscopic hydrophobicinteractions, which may help better understand the origin(s) ofthe hydrophobic force. It has been shown previously that hydro-phobic interactions entail decrease in the excess entropy of theTLFs, which lead to a conclusion that the water structure becomesmore ordered in thin films than in thick films.

Parker et al. [11] reported the measurement of attractive forcesbetween silica surfaces hydrophobized with (tridecafluoro-1,1,2,2,-tetrahydroxyoctyl)dimethylchorosilane at 41 �C and room temper-ature (22 �C). They found that the forces became stronger at thehigher temperature, which was contrary to what was observed inthe present work. The fluorinated surfaces exhibited advancingand receding contact angles of 110� and 90�, respectively, and thesurfaces showed no crystallinity. The force curve obtained at theroom temperature showed significant steps, which became stron-ger at the higher temperature. Therefore, the authors attributedthe increase in attractive force with temperature to the growth ofbubble size, which should promote the bridging mechanism.

In the present work, two sets of force measurements were con-ducted, one with OTS-coated silica surfaces of he = 92.5� andhe = 78.0�. If the bridging mechanism played a role, the result ob-tained at the higher contact angle would have been similar to whathas been reported by Parker et al. [11]. The results were diametri-cally opposite. First, the force curves obtained at he = 92.5� weresmooth, as shown in Fig. 4, providing no evidence for bridging

Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375 373

nano-bubbles affecting the force measurement. Second, the mea-sured forces decreased with increasing temperature. Further, thehydrophobic forces increased with increasing surface hydropho-bicity. It appears, therefore, that the long-range attractions ob-served in the present work were due to hydrophobic interactionsrather than to bridging bubbles.

The results obtained with surfaces with he = 78.0� are presentedin Fig. 5. Again, the force curves were smooth without visible steps,and yet, the forces were substantially stronger than the van derWaals force with a Hamaker constant of 0.8 � 10�20 J. It may beof interest to note here that the surface force measurements beganat 10 �C and then stepped up to 20 �C, 30 �C, and 40 �C. After thelast set of measurements, the temperature was brought down to20 �C and the measurements were repeated. The results werereproducible indicating that the OTS-coated surface was stable inthe range of temperatures where the measurements were con-ducted. It has actually been reported that OTS-coated surfacesare stable up to 175 �C as measured by water contact angle [24].

Note that Fig. 4 shows the data obtained at temperatures of 10,20, and 30 �C only. At 40 �C, we saw some bubbles nucleating onthe OTS-coated silica plate with he = 92.5�, and the results werenot very reproducible. Nevertheless, the force curves were smoothand showed no discontinuities or steps. In view of the difficultieswith reproducibility, we decided not to include the results ob-tained at 40 �C in the data set presented in Fig. 4. We did not havethis problem with the OTS-coated silica surfaces with he = 78.0� asshown in Fig. 5. Obviously, bubbles can more readily nucleate onthe more hydrophobic surface, and the surface force measure-ments become irreproducible in the presence of bubbles.

The solid lines of Figs. 4 and 5 represent the extended DLVO the-ory, with the hydrophobic forces represented by Eq. (1). The C andD parameters used to fit the experimental data are plotted in Fig. 6.As shown, both of these parameters decrease linearly with increas-ing temperatures (T), indicating that hydrophobic forces decreasewith increasing temperature. The slopes of the C vs. T plots ob-tained at he = 92.5� and 78.0� are about the same. The slopes ofthe D vs. T plots are also about the same, although the slope ob-tained at he = 92.5� tends to be slightly steeper than that obtainedat he = 78.0�. The results presented in Fig. 6 are similar to those re-ported by Wang et al. [1] with gold surfaces thiolated with C2SHand C16SH. These results are consistent with those reported byother investigators. Israelachvili and Pashley [36] showed thatthe hydrophobic forces measured between mica surfaces coatedwith CTAB by self-assembly decreased as the temperature was in-creased from 21 to 50 �C. Tsao et al. [37] also showed that theattractive forces measured between mica surfaces hydrophobizedby L–B deposited double-chain cationic surfactants decreased asthe temperature was raised from 25 to 50 �C.

7

8

9

10

10 20 30 40

92.5°

78.0°

C (m

N/m

)

t (°C)

(a)

Fig. 6. (a) C vs. t and (b) D vs. t plots. The C and D parameters of the hydrophobic forceFigs. 4 and 5 to the extended DLVO theory with w1 = �40 mV and A131 = 0.8 � 10�20 J.

Using the Derjaguin approximation [20], one can relate the sur-face force (F/R) measured at a separation distance h to the Gibbsfree energy change per unit area (DGf) as follows:

FR¼ 2pDGf ð2Þ

where

DGf ¼ �Z h

1PðhÞdh ð3Þ

As shown in Eq. (3), the free energy change is the same in magni-tude as the work done in thinning an infinitely thick plane-parallelfilm to a finite thickness h against the disjoining pressure P(h),which in turn is a function of film thickness. For a water film con-fined between two hydrophobic surfaces, the disjoining pressureshould include contributions from the hydrophobic force as dis-cussed above.

The partial differential of DGf with respect to T at a constantpressure and film thickness allows us to calculate the change inthe excess film entropy for moving a plane-parallel film from aninfinite thickness to an h as follows:

@DGf

@T

!p;h

¼ @ðF=2pRÞ@T

� �p;h¼ �DSf ð4Þ

in which DSf is the change in excess entropy per unit area of the li-quid film confined between two hydrophobic surfaces.

From Eqs. (1), (2), and (4), one obtains

DSf ¼ � F2pR

d ln CdT

þ hD

d ln DdT

� �ð5Þ

which can be used to calculate the changes in excess film entropy(DSf) at different film thicknesses using the surface force data, suchas F/R, D, and the temperature coefficients for the C and D parame-ters given in Fig. 6.

Fig. 7 gives the values of DSf obtained in this manner from thesurface forces measured with OTS-coated silica surfaces withhe = 78.0� and 92.5�. The entropy change becomes increasingly neg-ative with decreasing h, indicating that the entropy (Sf) of the TLFconfined between two hydrophobic surfaces decreases withdecreasing film thickness. As has been suggested previously [1],the low entropy may be due to structuring of the water moleculesin the vicinity of hydrophobic surfaces. The vicinal water moleculesmust be of high free energy state as they cannot form H-bonds tothe hydrophobic surface. This is akin to the case for the non-H-bonded water molecules at the air/water and hydrocarbon–waterinterfaces being responsible for high interfacial tensions [38]. Oneway to expend the excess free energy would be to form structures,

7

8

9

10

D (n

m)

10 20 30 40t (°C)

78.0°

92.5°

(b)

represented by Eq. (1) were obtained by fitting the experimental data presented in

-4

-3

-2

-1

0

0 10 20 30 40

-4

-3

-2

-1

0

4030

20

ΔSf × 1

06 (JK

-1m

-2)

10°C

10°C

θe=78.0°

h (nm)

θe=92.5°

2030

Fig. 7. Changes in excess film entropy (DSf) in the thin films of water between OTS-coated silica surfaces with contact angles of 78.0� and 92.5� at different mediumtemperatures.

-1.0

-0.5

0.0

0 10 20 30 40-1.5

-1.0

-0.5

0.0

θe= 78.0°

10°C203040

ΔH

f (mJm

-2)

h (nm)

10°C2030

θe= 92.5°

Fig. 8. Changes in excess film enthalpy (DHf) in the thin films of water betweenOTS-coated silica surfaces with contact angles of 78.0� and 92.5� at differentmedium temperatures.

-1.5

-1.0

-0.5

0.0

0 10 20 30 40

-1.5

-1.0

-0.5

0.0

Enge

rgy

(mJ/

m2 )

θe= 78°

h (nm)

ΔGf

ΔGf

TΔSf

TΔSf

ΔHf

ΔHf

20°C

20°C

θe= 92.5°

Fig. 9. Changes in the excess thermodynamic properties for the thin liquid filmsbetween OTS-coated silica surfaces with he = 78.0� and 92.5� at 20 �C.

374 Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375

in which the ‘‘frustrated’’ vicinal water molecules form strongerH-bonds with neighbors by taking advantage of the cooperativebonding mechanism first recognized by Frank and Wen [39] andused to explain the mechanism of hydrophobic hydration. It hasbeen shown that the O–O distance between water molecules de-creases with increasing number of water molecules in a structure(or cluster) [40]. The structure can be coplanar or concave (orclathrate-like) depending upon the temperature, pressure, anddissolved species [41,42]. It may be reasonable to assume thatthe clusters become larger and more abundant when water is con-fined in a TLF formed between very hydrophobic surfaces. Recentspectroscopic studies conducted in bulk water showed that watermolecules can form low-density liquid (LDL) species more readilythan previously thought to be possible [43–46]. These clustersmay be collectively referred to as ‘‘partial clathrates.’’ Clathratesare found in supercooled water or at high pressures and lowertemperatures particularly in the presence of dissolved hydrophobicsolutes [41,42].

Fig. 7 shows also that DSf becomes more negative with increas-ing hydrophobicity and decreasing temperature. The former maybe due to the likelihood that the number of non-H-bonded watermolecules increases with increasing surface hydrophobicity, whilethe latter can be attributed to the low kinetic energies of the watermolecules at a low temperature and hence increased propensity toform clusters.

One can readily obtain DGf from the surface force data using theDerjaguin approximation (Eq. (2)). From the values of DGf and DSf,the excess film enthalpy per unit area (DHf) can be determined asfollows:

DHf ¼ DGf þ TDSf ð6Þ

Fig. 8 shows the changes in DHf determined in this manner. Asshown, the enthalpy change associated with forming H-bondedstructures becomes more negative with decreasing h and T. Thisfinding suggests that both the number and size of the clusters in-crease with decreasing film thickness. Further, the strength and lifetime of the clusters may also increase as film becomes thinner.

Fig. 8 shows also that the enthalpy change becomes more negativeas the confining surfaces become more hydrophobic and the tem-perature is lowered. As has been the case with the entropy change,the former may be related to the increased number density of non-H-bonded water molecules, while the latter to decreased kineticenergies of water molecules.

The values of DGf, TDSf, and DHf determined in the manner de-scribed above are presented in Fig. 9. These values represent thechanges in the thermodynamic functions involved in the macro-

Z. Li, R.-H. Yoon / Journal of Colloid and Interface Science 392 (2013) 369–375 375

scopic hydrophobic interactions between the OTS-coated silicasurfaces with he = 78.0� and 92.5 �C at 20 �C. Note that at both con-tact angles, |DHf| > |TDSf|, which is a necessary condition for thehydrophobic force to be attractive when both enthalpy and entro-py changes are negative. The enthalpy term represents the free en-ergy gained in building the structures, while the entropy (TDSf)term representing the thermodynamic cost for building structures.As a result of the enthalpy–entropy tradeoff, the hydrophobicforces in general are weak and are difficult to be measured. Asnoted by Eriksson et al. [19], the excess free energy per moleculerelated to the long-range attractions is in the range of 10�5–10�3 kT, while the energy of a hydrogen bond is about 7 kT. Notealso that DGf is more negative at the higher contact angle, whichcan be attributed to more negative DHf.

The condition that |DHf| be larger than |TDSf| for macroscopichydrophobic interactions is opposite to what is generally knownfor molecular-scale hydrophobic interactions or hydrophobic ef-fect. For the latter, it is necessary that |DHf| < |TDSf| for two hydro-phobic molecules to attract each other as in self-assembly. It isgenerally accepted that the inequality arises from the entropy in-crease associated with the release of the water molecules sur-rounding hydrophobic moieties of high curvatures. The basicdifference between the macroscopic- and molecular-scale hydro-phobic interactions is due to the differences in curvatures of thesurfaces involved, as discussed by Chandler [42].

The authors of this communication showed recently that long-range attractions are also observed in short-chain alcohols such asmethanol, ethanol, and butanol – all well-known H-bonding liq-uids [47]. It has been suggested that the origin of the long-rangeattractions observed in these short-chain alcohols is the same asthe hydrophobic forces observed in water, that is, structure forma-tion. It was found also that the long-range attractions becomestronger as the confining surfaces become more hydrophobic.The same has been observed in the present work with water asH-bonding liquid. It may be suggested, therefore, that hydrophobicforce is the consequence of an H-bonding liquid striving to mini-mize free energy when it cannot form H-bonds to confining sur-faces. Conversely, a key to increasing hydrophobic force is toeliminate H-bond sites from the confining surfaces.

There have been long-standing controversies concerning theorigin of the hydrophobic force since Israelachvili and Pashley [4]published a seminal paper in 1982. It is hoped that the results pre-sented in this communication have provided a basis for the hydro-phobic force be recognized as a thermodynamic force rather thanan artifact due to bubbles or cavities.

Acknowledgments

The authors acknowledge the financial support from the USDepartment of Energy (DE-FC26-05NT42457) and Professor JanChrister Eriksson for helpful discussions and encouragement.

References

[1] J. Wang, R.-H. Yoon, J.C. Eriksson, J. Colloid Interface Sci. 364 (2011) 257.[2] Z. Xu, R.-H. Yoon, J. Colloid Interface Sci. 132 (1989) 532.[3] Z. Xu, R.-H. Yoon, J. Colloid Interface Sci. 134 (1990) 427.[4] J. Israelachvili, R. Pashley, Nature 300 (1982) 341.[5] H.K. Christenson, P.M. Claesson, Adv. Colloid Interface Sci. 91 (2001) 391.[6] E.E. Meyer, K.J. Rosenberg, J. Israelachvili, Proc. Natl. Acad. Sci. USA 103 (2006)

15739.[7] R.H. Yoon, S.A. Ravishankar, J. Colloid Interface Sci. 166 (1994) 215.[8] Y.I. Rabinovich, D.A. Guzonas, R.H. Yoon, J. Colloid Interface Sci. 155 (1993)

221.[9] Y.I. Rabinovich, D.A. Guzonas, R.H. Yoon, Langmuir 9 (1993) 1168.

[10] R.H. Yoon, S.A. Ravishankar, J. Colloid Interface Sci. 179 (1996) 391.[11] J.L. Parker, P.M. Claesson, P. Attard, J. Phys. Chem. 98 (1994) 8468.[12] J.W.G. Tyrrell, P. Attard, Phys. Rev. Lett. 87 (2001) 176104.[13] M. Sakamoto, Y. Kanda, M. Miyahara, K. Higashitani, Langmuir 18 (2002) 5713.[14] E.E. Meyer, Q. Lin, J.N. Israelachvili, Langmuir 21 (2005) 256.[15] M.U. Hammer, T.H. Anderson, A. Chaimovich, M.S. Shell, J. Israelachvili,

Faraday Discuss. 146 (2010) 299.[16] B.V. Derjaguin, M.M. Kusakov, Izv. Akad. Nauk SSSR, Ser. Khim. 5 (1936) 741.[17] B.V. Derjaguin, N.V. Churaev, Langmuir 3 (1987) 607.[18] N.V. Churaev, B.V. Derjaguin, J. Colloid Interface Sci. 103 (1985) 542.[19] J.C. Eriksson, S. Ljunggren, P.M. Claesson, J. Chem. Soc. Faraday Trans. II (85)

(1989) 163.[20] B.V. Derjaguin, Kolloid Zeits. 69 (1934) 155.[21] Y. Li, J. Huang, R.T. McIver, J.C. Hemminger, J. Am. Chem. Soc. 114 (1992) 2428.[22] M.H. Schoenfisch, J.E. Pemberton, J. Am. Chem. Soc. 120 (1998) 4502.[23] M. Dasog, R.W.J. Scott, Langmuir 23 (2007) 3381.[24] W.R. Ashurst, C. Yau, C. Carraro, R. Maboudian, M.T. Dugger, J.

Microelectromech. Syst. 10 (2001) 41.[25] Y.I. Rabinovich, R.H. Yoon, Langmuir 10 (1994) 1903.[26] R.H. Yoon, D.H. Flinn, Y.I. Rabinovich, J. Colloid Interface Sci. 185 (1997) 363.[27] N. Ishida, N. Kinoshita, M. Miyahara, K. Higashitani, J. Colloid Interface Sci. 216

(1999) 387.[28] N. Ishida, M. Sakamoto, M. Miyahara, K. Higashitani, Langmuir 16 (2000) 5681.[29] J.P. Cleveland, S. Manne, D. Bocek, P.K. Hansma, Rev. Sci. Instrum. 64 (1993)

403.[30] D.H. Flinn, D.A. Guzonas, R.H. Yoon, Colloid Surf. A 87 (1994) 163.[31] A.H. Carim, M.M. Dovek, C.F. Quate, R. Sinclair, C. Vorst, Science 237 (1987)

630.[32] N. Tillman, A. Ulman, J.S. Schildkraut, T.L. Penner, J. Am. Chem. Soc. 110 (1988)

6136.[33] M. Mao, J.H. Zhang, R.H. Yoon, W.A. Ducker, Langmuir 20 (2004) 4310.[34] M.J. Wang, K.M. Liechti, Q. Wang, J.M. White, Langmuir 21 (2005) 1848.[35] J.H. Zhang, R.H. Yoon, J.C. Eriksson, Colloid Surf. A 300 (2007) 335.[36] J.N. Israelachvili, R.M. Pashley, J. Colloid Interface Sci. 98 (1984) 500.[37] Y.H. Tsao, S.X. Yang, D.F. Evans, H. Wennerstrom, Langmuir 7 (1991) 3154.[38] Q. Du, E. Freysz, Y.R. Shen, Science 264 (1994) 826.[39] H.S. Frank, W.-Y. Wen, Discuss. Faraday Soc. 24 (1957) 133.[40] K. Liu, J.D. Cruzan, R.J. Saykally, Science 271 (1996) 929.[41] F.H. Stillinger, Science 209 (1980) 451.[42] D. Chandler, Nature 437 (2005) 640.[43] F. Mallamace, C. Branca, M. Broccio, C. Corsaro, C.Y. Mou, S.H. Chen, Proc. Natl.

Acad. Sci. USA 104 (2007) 18387.[44] F. Mallamace, C. Corsaro, M. Broccio, C. Branca, N. Gonzalez-Segredo, J.

Spooren, S.H. Chen, H.E. Stanley, Proc. Natl. Acad. Sci. USA 105 (2008) 12725.[45] F. Mallamace, Proc. Natl. Acad. Sci. USA 106 (2009) 15097.[46] O. Mishima, H.E. Stanley, Nature 396 (1998) 329.[47] J. Wang, Z. Li, R.-H. Yoon, J.C. Eriksson, J. Colloid Interface Sci. (2012).