Dewetting Behavior of Aqueous Cationic SurfactantSolutions on Liquid Films

Abia B. Afsar-Siddiqui, Paul F. Luckham, and Omar K. Matar*

Department of Chemical Engineering & Chemical Technology,Imperial College London, SW7 2AZ, U.K.

Received March 10, 2004. In Final Form: June 14, 2004

Previous experimental work has shown that the spreading of a drop of aqueous anionic surfactantsolution on a liquid film supported by a negatively charged solid substrate may give rise to a fingeringinstability (Afsar-Siddiqui, A. B.; Luckham P, F.; Matar, O. K. Langmuir 2003, 19, 703-708). However,upon deposition of a cationic surfactant on a similarly charged support, the surfactant will adsorb ontothe solid-liquid interface rendering it hydrophobic. Water is then expelled from the hydrophobic regions,causing film rupture and dewetting. In this paper, experimental results are presented showing how thesurfactant concentration and film thickness affect the dewetting behavior of aqueous dodecyltrimethyl-ammonium bromide solutions. At low surfactant concentrations and large film thicknesses, the film rupturesat a point from which dewetting proceeds. At higher concentrations and smaller film thicknesses, theruptured region is annular in shape and fluid moves away from this region. At still higher concentrationsand smaller film thicknesses, the deposited surfactant forms a cap at the point of deposition that neitherspreads nor retracts. This variation in dewetting mode is explained by considering the relative Marangoniand bulk diffusion time scales as well as the mode of assembly of the surfactant adsorbed on the solidsurface.

1. IntroductionThe presence of surface tension gradients across a thin

liquid film of uniform height induces shear stresses at theair-liquid interface. These stresses distribute the liquidfrom areas of low surface tension to areas of high surfacetension and, in doing so, also deform the interface resultingin height variations. This so-called Marangoni flow canbe generated by the presence of nonuniformly distributedsurface active material on a liquid film or by temperaturegradients along it.1 The disturbances in film height thatresult from Marangoni stresses may, in some cases, beso severe as to lead to film rupture and subsequent de-wetting. This is undesirable in, for example, gravureprinting and photofinishing applications where a uniformfinish is often required.2 Marangoni drying, however, relieson Marangoni stresses, created by alcohol vapor acrossthe surface of a wet substrate to dewet the area of contact.3,4

This is an effective means of drying integrated circuitsand liquid crystal displays. Thus it is important tounderstand the conditions that give rise to rupture anddewetting.

Rupture of a thin film is driven by the presence of vander Waals forces, a component of intermolecular forcesthat becomes significant for film thicknesses of order 1000Å5,6 or less. Intermolecular forces can be characterized byan interaction potential of the film, Φ(H), where H denotesthe local film thickness. This represents the long-range(van der Waals) and short-range (Born repulsion) inter-

actions between the air-liquid and solid-liquid interfacesand is defined as the free energy required in bringingthese two interfaces together from infinity to a distanceH. If Φ(H) is positive for all values of H, then the film issaid to be stable. If the second derivative of Φ(H), Φ′′(H),is negative, then the film is unstable and will spontane-ously thin. This mechanism is termed spinodal dewetting,due to its similarity to spinodal decomposition of mixtures,which occurs when the second derivative of the free energywith respect to composition becomes negative.7-10 A filmthat is unstable for small film thicknesses but stable atlarger thicknesses is termed metastable. Spinodally stablefilms can dewet only through the nucleation of a hole inthe film. This may be as a result of heterogeneities on thesolid-liquid interface, such as defects11,12 or chemicalpatterning,13,14 or at the air-liquid interface, such as dustparticles15 or surface active agents.16,17 Such disturbancescause local changes in the chemical potential giving riseto flows away from regions of high chemical potential.The resulting thinning of the film may be so severe thatlong-range intermolecular forces become significant. Thesefurther thin the film to a microscopically thin equilibriumthickness that is governed by the long-range attractiveforces and the short-range repulsive interactions betweenthe air-liquid and solid-liquid interfaces, thus setting

* To whom correspondence may be addressed. E-mail: o.Matar@imperial.ac.uk.

(1) Edwards, D. A.; Brenner, H.; Wasan, D. T. Interfacial TransportProcesses and Rheology; Butterworth-Heinemann: New York, 1991.

(2) Schwartz, L. W.; R. R. V.; Eley R. R.; Petrash, S. J. Colloid InterfaceSci. 2001, 234, 363.

(3) Leenars, A. F. M.; Wuethorst, J. A. M.; van Oekel, J. J. Langmuir1990, 6, 1701

(4) O’Brien, S. B. G. M. J. Fluid Mech. 1993, 254, 649.(5) Ruckenstein, E.; Jain, R. K. J. Chem Soc., Faraday Trans. 1974,

70, 132.(6) Israelachvili, J. N. Intermolecular and Surface Forces; Academic:

London, 1985

(7) Cahn, J. W.; Hillard, J. E. J. Chem. Phys. 1957, 28, 258.(8) Brochard-Wyart, F.; Daillant, J. Can. J. Phys. 1990, 68, 1084.(9) Mitlin, V. S. J. Colloid Interface Sci. 1993, 156, 491.(10) De Gennes, P. G.; Brochard-Wyart, F.; Quere D. Capillarity and

Wetting Phenomena: Drops, Bubbles, Pearls, Waves; Springer-Verlag,New York, 2003; Chapters 7 and 10.

(11) Jacobs, K.; Herminghaus, S.; Mecke, K. R. Langmuir 1998, 14,965.

(12) Strange, T. D.; Evans, D. E.; Hendrickson, W. A. Langmuir 1997,13, 4459.

(13) Kargupta, K.; Konnur, R.; Sharma, A. Langmuir 2001, 17, 1294.(14) Konnur, R.; Kargupta, K.; Sharma, A. Phys. Rev. Lett. 2000, 84,

931.(15) Kheshgi, H. S.; Scriven, L. Chem. Eng. Sci. 1983, 38, 525.(16) Kheshgi, H. S.; Scriven, L. Chem. Eng. Sci. 1991, 46, 519.(17) Warner, M. R. E.; Craster, R. V.; Matar, O. K. Phys. Fluids

2002, 14, 11.

7575Langmuir 2004, 20, 7575-7582

10.1021/la040041z CCC: $27.50 © 2004 American Chemical SocietyPublished on Web 08/06/2004

up the conditions that give rise to rupture and theformation of a dry patch or hole. This hole precedes thedewetting process and occurs through the transport ofmaterial away from the hole.

Kheshgi and Scriven16 demonstrated the dewetting ofa spinodally stable film by placing a 5 µL drop of methanolon a 0.4 mm glycerine/water film on a horizontal glassplate. Fluid is drawn away from the deposited region byMarangoni stresses and forms an elevated rim furtherdownstream; 4 s later rupture occurs. Rupture creates adry patch which grows as liquid moves away from it andaccumulates into a rim, the width and height of whichincrease with time as shown schematically in Figure 1.The native film ahead of the rim is motionless as verifiedby the use of talc particles.18 The width of the rim, lD, ispredicted to grow with time as t1/2.19 The radius of thedewetting hole, RD, is predicted to grow linearly with timeover long times. However, thin film experiments showthat the dewetting exponent can range from 0.7 to 0.9 forthe duration of hole growth.20

Dewetting through the retraction of the film edge istermed “autophobing”. In a spreading situation, thecontact angle will generally decrease as spreadingprogresses until a final contact angle is achieved. However,in the case of autophobing, the contact angle first decreasesand then increases as the solution first spreads and thenretracts. A surfactant solution may autophobe when thesurfactant headgroup and surface are mutually attractive.The attraction between the surfactant headgroup and thesurface causes the headgroup to adsorb onto the substrateleaving the hydrophobic tails exposed. This lowers thesurface energy so that the solution can no longer wet thenow hydrophobic surface and thus retracts.21 This wasdemonstrated by Frank and Garoff 22-24 using the cationicsurfactant cetyltrimethylammonium bromide [CTAB].Drops of 0.1cmc aqueous CTAB solution, at concentrationsbelow 0.45 times the critical micelle concentration (cmc),exhibit autophobing behavior on uncoated horizontalsilicon oxide substrates. There is an increase in the contactangle of 5-10° over a period of 30 s with the final contactangle being 25°. Above a concentration of 0.45cmc, dropsdo not spread or retract but maintain their initial contactangle, as previously observed by Marmur and Lelah25

using a range of cationic surfactants on glass slides.Woodward and Schwartz26 found a systematic depen-

dence of the dewetting mechanism of a surfactant solutionon the solution concentration and free energy of thesurface. Monolayers of octadecylphosphonic acid (OPA)

were deposited on a freshly cleaved mica surface at varioussolution concentrations and for varying immersion times.On removal of the mica disks from solution, three distinctbehaviors were noted: the surface emerged wet andremained wet; the surface emerged wet and then rupturedto form a dry patch which continued to grow (dewetting);or the sample emerged dry (autophobing). It was foundthat the behavior tends from wetting through dewettingto autophobing as the solution concentration increasesand as the hydrophobicity of the surface increases.

The aim of this experimental study is to systematicallyexplore the dependence of the dewetting mode on sur-factant concentration and film thickness. Drops of aqueousdodecyltrimethylammonium bromide (DTAB) solution,over a range of concentrations above and below the cmc,have been deposited on water films up to 200 µm inthickness. It has been found that across the studiedparameter range there are three types of dewettingbehavior. These tend from hole formation, through toautophobing, and finally cap formation, in which thedeposited surfactant forms a cap that neither spreads norretracts, with increasing surfactant concentration anddecreasing film thickness, but simply remains throughoutthe experiment. This behavior may be explained byconsidering the relative Marangoni and bulk diffusiontime scales as well as the mode of assembly of thesurfactant adsorbed on the solid surface.

2. Experimental Details2.1. Materials. The liquid substrate was ultrapure water

obtained from a Barnstead NANOpure II filter system with aresistively of less than 18 MΩ cm and a surface tension of 72.2( 0.5 mN/m at 25 °C. This was also used to clean the glasswareand syringe as well as to make up the surfactant solutions. ThesurfactantwasDTAB(dodecyltrimethylammoniumbromide,MW308.3, 99+%, Aldrich) which is a soluble cationic surfactant witha cmc of 1.4 × 10-2 M.27 The surfactant was made up to thedesired concentration using ultrapure water. The surface tensionof the surfactant solutions was determined using a platinumWilhelmy plate suspended from a Kruss microbalance, at aconstant temperature of 25 °C.

2.2. Visualization Technique. The experiments were per-formed in a circular glass Petri dish 15 cm in diameter with anoptically flat bottom. This was positioned in a four-pointadjustable level stage. The setup was illuminated from aboveusing a fiber optic lamp and the image projected onto a tracingpaper screen placed beneath the glass dish. A Pulnix CCDprogressive scan camera (model TM6710) was used to record theimages at a rate of 120 frames/s via a mirror (camera position1). This entire system rested on an antivibration table to isolatethe system from vibrations greater than 1 Hz. A schematic dia-gram of the setup is shown in Figure 2. On occasions, the CCDcamera was positioned above the Petri dish to observe, at a glanc-ing angle, the spreading of the liquid drop (camera position 2).

2.3. Experimental Procedure. To ensure complete wetta-bility, the Petri dishes were soaked for at least 12 h in a 2% RBS50 surfactant solution (Chemical Concentrates Ltd.) before beingthoroughly rinsed with ultrapure water. They were then left inan ultrasonic bath for at least 1 h to ensure that all traces ofdetergent were removed from the glass surface. The outside ofthe Petri dish was then dried, as were the inside walls, leavinga continuous water film over the base only. The exact height ofthis liquid film was calculated from the weight of the water layer.The error in the film thickness was estimated to be no more than4%, based on film evaporation between weighing and the startof experimentation and the curvature of the Petri dish where theside walls are fused to the base. The glassware used to make upthe surfactant solutions was also cleaned in the same way as thePetri dishes. Solutions were used within 24 h to avoid anydecrease in surface activity.28

(18) Redon, C.; Brochard-Wyart, F.; Rondelez, F. Phys. Rev. Lett.1991, 66, 715.

(19) Brochard-Wyart, F.; Martin, P.; Redon, C. Langmuir 1993, 9,3682.

(20) Ghatak, A.; Khanna, R.; Sharma, A. J. Colloid Interface Sci.1999, 212, 483.

(21) Birch, W. R.; Knewtson, M. A.; Garoff, S.; Suter, R. M.; Satija,S. Langmuir 1995, 11, 48.

(22) Frank, B.; Garoff, S. Langmuir 1995, 11, 87.(23) Frank, B.; Garoff, S. Langmuir 1995, 11, 4333.(24) Frank, B.; Garoff, S. Colloids Surf., A 1996, 116, 31.(25) Marmur, A.; Lelah, M. D. Chem. Eng. Commun. 1981, 13, 133.(26) Woodward, J. T.; Schwartz, D. K. Langmuir 1997, 13, 6873.

(27) Mukerjee, P.; Mysels, K. J. National Bureau of Standards, USDepartment of Commerce, Washington, DC, 1970.

Figure 1. A schematic side view of a ruptured film resting ona solid substrate. Downstream of the dewetted region, the fluidforms an elevated rim. lD denotes the width of the elevated rimand RD denotes the radius of the dewetting hole.

7576 Langmuir, Vol. 20, No. 18, 2004 Afsar-Siddiqui et al.

A 20 µL precision Hamilton syringe was used for drop delivery.This was flushed several times with the surfactant solution tobe used before the desired amount was drawn, in this case avolume of 9 µL. The Petri dish with the desired film thicknesswas placed on the adjustable stage. A drop of surfactant wasfirst released from the syringe such that it hung from the tip ofthe needle. The apex of the drop was then contacted with thewater film and was drawn across the water surface by Marangonistresses. This method proved to be the least disturbing to thewater surface.

The spreading was followed for about 20 s after deposition,and the images were analyzed using commercially availablesoftware. Each spreading run was repeated at least three timesto ensure good reproducibility with complete cleaning of all theequipment prior to each run.

3. Results

The deposition of DTAB over a range of concentrationson different film thicknesses gives rise to three distincttypes of behavior. At very low concentrations, there isstable spreading for several seconds, after which a holeforms and dewetting ensues (hole formation). At inter-mediate concentrations and on thick films, the spreadsurfactant solution retracts to a cap at the point ofdeposition (autophobing). At high concentrations and onthinner films, the spread surfactant does not retract butremains as a flattened cap in the center (cap formation).Table 1 summarizes the behavior of DTAB solutions overa range of surfactant concentrations and film thickness.

(a) Hole Formation. This type of behavior is observedat very low surfactant concentrations (0.04cmc) on filmthicknesses ranging from 25 to 100 µm. Figure 3 showsthe typical formation and growth of a dewetting hole foraqueous DTAB solutions at low concentration.

Upon surfactant deposition, the surfactant spreads ina stable and circular manner for approximately 8 s at allfilm thicknesses (Figure 3A). The spreading exponentassociated with the leading edge for this time is about

0.26. This is close to the theoretically predicted value of0.25 for surfactant spreading under the influence ofMarangoni forces, thus suggesting that this part of thespreading is Marangoni-driven.29 From a random pointwithin the spreading front (Figure 3B), a rupture site isformed and liquid is pushed outward leaving an apparentlydry region in its path (Figure 3C). No thinning of the filmis apparent prior to the formation of the rupture site;however, the asymmetric nature of the hole reflects theheight variations in the film. On the 25 µm water film, thedewetting continues for several seconds while on thethicker 50 and 100 µm water films, dewetting is completewithin 1 s and the expelled liquid is pushed out towardthe leading edge. Both the surfactant leading edge andthe dewetting front continue to advance but now with anexponent of about 0.5 (Figure 3D). The spreading rate ofthe surfactant leading edge is about 1 mm/s throughoutfor all film thicknesses.

On a thicker 200 µm film there is little appreciablespreading (Figure 4A), before the drop retracts (Figure4B) and subsequently sinks into the water film within afew seconds.

The Bond number, Bo, is defined as Bo ) (FgHo2)/S,

where F is the liquid film density, g is the gravitationalacceleration, Ho is the initial film thickness, and S is thespreading pressure, and it gives an indication of thestrength of gravitational forces relative to Marangonistresses. The Bond number at the leading edge of a 200µm Marangoni-driven film at this low concentration isclose to 1. It appears, therefore, that gravitational effectsat this film thickness and surfactant concentration aresufficiently significant to overwhelm Marangoni forcesand cause flow reversal.29

(b) Autophobing. At higher surfactant concentrations,but still below the cmc, the spreading surfactant dropadvances some distance under the influence of Marangoniforces before it retracts (autophobes). The extent of thespreading and retraction are functions of the surfactantconcentration and the film thickness. The autophobingsurfactant solution may retract completely either to arounded cap close to the point of surfactant deposition orto a more flattened cap with a smaller contact angle, againclose to the point of initial deposition. Figure 5 shows theautophobing behavior typically seen on a thin film atintermediate surfactant concentrations, where the sur-factant droplet retracts to a rounded cap soon aftersurfactant deposition.

Upon deposition, a spreading front is visible ahead ofa cap of surfactant that has formed at the point ofsurfactant deposition (Figure 5A). The white ring im-mediately downstream of the spreading front is indicativeof film thinning just ahead of the leading edge. This mostlikely occurs when the advancing elevated edge encountersthe undisturbed film and experiences a sudden decreasein velocity, thus causing a thinning further ahead. Thefaint front ahead of the spreading front visible throughoutthe spreading is thought to be a shock wave arising fromthe deposition of the surfactant on the surface in thisparticular experiment. Almost immediately after deposi-tion (0.3 s), the cap has retracted (Figure 5B), leaving adewetted region between the cap and the surfactant front.The front continues to advance radially with time (Figure5C) and develops crenulations at later times (Figure 5D).The region between the shock wave and the spreadingfront is not dewetted. A similar sequence of events occurswith 0.8cmc (11.2 mM) on a 50 µm water film (not shown)on a similar time scale.

(28) Burcik, E. J.; Vaughn, C. R. J. Colloid Interface Sci. 1951, 6,522.

(29) Gaver, D. P.; Grotberg, J. B. J. Fluid Mech. 1990, 213, 127;1992, 235, 399.

Figure 2. Schematic diagram of the experimental setup.

Table 1. Variation of the Spreading Behavior of AqueousDTAB Solutions with Initial Surfactant Concentration

and Initial Film Thickness

initial film thicknessinitial surfactantconcn (mM) 25 µm 50 µm 100 µm 200 µm

0.56 (0.04cmc) hole formation no spreading5.6 (0.4cmc) autophobing

11.2 (0.8cmc) cap formation autophobing16.8 (1.2cmc) cap formation autophobing22.4 (1.6cmc) cap formation autophobing56 (4cmc) cap formation

Dewetting Behavior of Surfactant Solutions Langmuir, Vol. 20, No. 18, 2004 7577

On thicker 50 and 100 µm films, the surfactant advancesfurther before retracting as illustrated in Figures 6 and 7.

In Figures 6 and 7, upon surfactant deposition, thesurfactant leading edge is seen as a dark ring with a shockwave ahead. Just behind the leading edge is a faint lighterfront that is thought to correspond to the region of thefilm thinned by Marangoni forces (Figure 6A). The thinnedfilm then ruptures isolating the flattened cap of spreadingsurfactant from the elevated front ahead (Figures 6B and7A). There is an additional light ring further upstreamvisible in Figure 7A, but it is unclear what this correspondsto. The cap quickly retracts leaving behind a dewettedregion between itself and the front (Figure 6C). The regionbetween the shock wave and the dewetting front is notdewet. After several seconds, distortions appear in thedewetting front (Figures 6D and 7B). Similar behavior isalso observed on deposition of 0.4 and 0.8cmc solutions on100 and 200 µm water films. At these high film thicknesses,the shock wave is seen to travel much faster.

At concentrations just above the cmc (1.2 and 1.6cmc)on 100 and 200 µm films, the film thins and ruptures butthe subsequent retraction is not to a rounded cap at thepoint of deposition but to a flattened cap with a smallercontact angle (Figure 8).

Tables 2 and 3 show the time and radius, respectively,at which the dewetting becomes apparent for each of theautophobing cases. These tables show that the time andradius at which the dewetting becomes apparent increaseswith increasing film thickness and decreasing surfactantconcentration.

(c) Cap Formation. Over certain parameter ranges,the majority of the deposited surfactant remains as aflattened cap at the point of deposition, a behavior thatwe have termed “cap formation”. The cap neither spreadsnor retracts for the duration of the spreading, remaining

at a constant radius typically around 10 mm for all thestudied experimental conditions. Further downstream,the elevated front continues to advance slowly downstreamfor the duration of the spreading; its speed appears todepend on the surfactant concentration.

Figure 9 illustrates typical cap formation behaviorobserved on thin (25 µm) films at a surfactant concentra-tion of 0.8cmc, the lowest surfactant concentration atwhich such behavior is seen.

In the autophobing case, Marangoni-driven spreadingwas observed for some fractions of a second before thethinned region of the film ruptured. In the case of capformation, a dewetted region forms between the cap andthe front almost immediately (Figure 9A). The cap remainsat a radius of approximately 10 mm throughout theduration of the spreading and appears to be distorted atthe edges, while the dewetting front advances slowlydownstream (Figure 9B). A shock wave is visible just aheadof the dewetting front even at late times. Similar spreadingbehavior is seen when depositing 1.2cmc surfactantsolution on a 25 µm thick film.

Figure 3. A 9 µL drop of 0.04cmc (0.56 mM) DTAB solution spreading on a 50 µm thick water film: (A) 1 s after deposition whenspreading is stable; (B) 8 s after deposition, a rupture site has formed and liquid is moving away from this region; (C) 8.5 s afterdeposition, dewetting continues; (D) 12 s after deposition, the dewetting front begins to merge with the advancing front at thesurfactant leading edge. Schematic inferred height profiles (through the vertical dashed lines in parts b and c) are given beloweach image.

Table 2. Variation in Approximate Onset Time (seconds)for Autophobing with Surfactant Concentration andWater Film Thickness for Aqueous DTAB Solutions

initial film thicknessinitial surfactantconcn (mM) 25 µm 50 µm 100 µm 200 µm

5.6 (0.4cmc) 0.125 0.16 0.3 0.4211.2 (0.8cmc) 0.12 0.2 0.2416.8 (1.2cmc) 0.12 0.1622.4 (1.6cmc) 0.04 0.08

Figure 4. A 9 µL drop of 0.04cmc (0.56 mM) DTAB solutionspreading on a 200 µm thick water film: (A) ∼1 s and (B) ∼2s after deposition.

Table 3. Variation in Approximate Onset Radius(millimeters) for Autophobing with SurfactantConcentration and Water Film Thickness for

DTAB Solutions

initial film thicknessinitial surfactantconcn (mM) 25 µm 50 µm 100 µm 200 µm

5.6 (0.4cmc) 7 8 15 1811.2 (0.8cmc) 6 12 1616.8 (1.2cmc) 11 1522.4 (1.6cmc) 10 14

7578 Langmuir, Vol. 20, No. 18, 2004 Afsar-Siddiqui et al.

Figure 10 shows the behavior of a 1.6cmc solution ona 25 µm water film. As the concentration is increased ona given film thickness, the rate of advance of the dewettingfront decreases. Figure 10 shows that even after severalseconds the dewetting front has not moved appreciablyand the cap has remained at a radius of about 5 mmthroughout the spreading. A shock wave is visible justahead of the dewetting front even at late times. Similarspreading behavior is seen when depositing 4cmc sur-factant solution on a 25 µm thick film.

Figures 11 and 12 show the typical dewetting behaviorof DTAB solutions above the cmc on thicker (50 and 100µm) films. Behavior similar to that shown in Figures 11and 12 is also observed when spreading 1.6 and 4cmcsolutions on 50, 100, and 200 µm films. On the thickestfilms, the shock wave travels much faster and is, therefore,not visible. There appear to be slight crenulations in thedewetting front and also at the edge of the cap.

4. DiscussionSince there is a surface tension gradient across the liquid

film upon surfactant deposition, the initial stages of

spreading are likely to be Marangoni driven. This isaccompanied by surfactant diffusion through the bulk andadsorption onto the solid-liquid interface because of theattractive interaction between the surfactant headgroupand the substrate.

Ionic surfactantadsorption toachargedsurface typicallyfollows three regimes, with increasing surfactant con-centration:30,31

(i) The surfactant monomers adsorb with their polarmoiety in contact with the surface in response to Coulombicattraction. The counterions in the diffuse double layerjust outside the surface are exchanged for surfactant ionswith the same charge. The hydrophobic tails can eitherlie flat on the surface or align perpendicular to the surface.For alkyl chains with greater than nine carbon atoms(as in this case), the tails are more likely to alignhorizontally.32

(30) Koopal, L. K.; Goloub, T. In Surfactant Adsorption and SurfaceSolubilization; Sharma, R., Ed.; American Chemical Society: Wash-ington, DC, 1995.

(31) Jonsson, B.; Lindman, B.; Holmberg, K.; Kronberg, B.Surfactantsand Polymers in Solution; John Wiley and Sons: Chichester, 1999.

Figure 5. A 9 µL drop of 0.4cmc (5.6 mM) DTAB solution spreading on a 25 µm thick water film: (A) at 0.08 s, a spreading frontand a cap can be seen; (B) at 0.3 s, the cap has retracted; (C) at 5 s, the dewetting front continues to grow; (D) at 17 s, crenulationsare evident in the dewetting front. The outermost faint front visible throughout the spreading is thought to be a shock wave.Schematic inferred side views are also given.

Figure 6. A 9 µL drop of 0.4cmc (5.6 mM) DTAB solution spreading on a 50 µm thick water film: (A) at 0.08 s, the spreadingfront of the surfactant is seen as a dark ring and a lighter front is just visible inside this; (B) at 0.2 s, the thin region of the filmhas ruptured isolating the spreading surfactant drop from the front ahead; (C) at 0.5 s, the surfactant drop has retracted into acap; (D) at 16 s, crenulations are evident in the dewetting front. The faint wave visible ahead of the leading edge is thought tobe a shock wave. Inferred side views are shown below each image.

Dewetting Behavior of Surfactant Solutions Langmuir, Vol. 20, No. 18, 2004 7579

(ii) The ion exchange leads to a higher surfactantconcentration close to the solid surface as compared tothe bulk. This induces a surface micellization process atthe solid-liquid interface at bulk concentrations belowthe bulk cmc. The analogous surface concentration isknown as the critical surface aggregation concentration(csac) and is typically of the order of one-tenth the valueof the cmc of the surfactant.33 These surface aggregateshave been termed hemimicelles.34 The tail groups start toalign perpendicularly to the surface, while the headgroupsare still in contact with the surface.

(iii) As the tail groups align perpendicularly to the solid-liquid interface, a hydrophobic surface is created and thehydrophobic chains of the surfactant adsorb onto this.The exact mode of assembly of the surfactant moleculesis dependent on both the interaction of the surfactantwith the surface and also the interaction between thehydrophobic moieties of the surfactant that gives rise tothe so-called hydrophobic effect.31 In the case of DTAB

adsorption on a glass surface, the surface-surfactantinteraction is stronger than the interaction between thehydrophobic chains and so a bilayer structure is formed.This occurs at bulk concentrations of around the cmc.

Surfactant adsorption giving rise to a hydrophobicsurface is followed by film rupture and formation of adewetted region from which liquid moves away. A typicaldewetting event is illustrated in Figure 13. Upon sur-factant deposition on a liquid support, Marangoni forceswill give rise to surface deformations. Surfactant willdiffuse more rapidly through the Marangoni thinnedregion to adsorb at the solid-liquid interface (Figure 13a).The assembly of the molecules is dependent on thesurfactant concentration. This region of the solid surfacethen becomes hydrophobic and water is repelled from thisarea, thus causing further thinning of the film to the extentthat long-range intermolecular forces become significant(Figure 13b). Since the disjoining pressure is negative,van der Waals forces act to further thin the film, whileshort range repulsive forces (Born repulsion) resistthinning. A dry patch or hole results with a microscopicallythin equilibrium dewetting thickness, te. Water movesaway from the hole via a dewetting front (Figure 13c).

The results show that there is a dependence of thedewetting behavior on both surfactant concentration andfilm thickness. With increasing surfactant concentration,the hydrophobicity of the solid surface increases up to thecmc and decreases thereafter as the surfactant headgroupsare exposed. In addition, the significance of Marangoniforces changes relative to bulk diffusion as the surfactantconcentration increases and there are two competingeffects. The Marangoni time scale can be defined as tm )(µR2/SHo), where µ is the film viscosity, R is the radialextent of spreading, S is the spreading pressure, and Hois the initial film thickness. The bulk diffusion time scalecan be defined as tDb ) (Ho

2/Db), where Db is the bulkdiffusivity. As the surfactant concentration increases, theMarangoni time scale is reduced in relation to the timetaken for the surfactant to diffuse through the bulk andadsorb onto the solid surface. However, an increasedconcentration provides a greater source of surfactant thatcan potentially adsorb at the solid-liquid interface andthe rate of surfactant adsorption is dependent on thesurfactant concentration. The higher the concentrationof the surfactant, the greater the rate of adsorption. Pagacet al.35 have shown that for CTAB adsorbing onto a silicasurface at 0.55cmc, it takes 900 min to reach equilibriumadsorption, at 0.89cmc, it takes 45 min, while at 11cmc,it takes only 2 min. It is assumed that DTAB adsorptionwill follow a similar trend. Hence increasing the surfactantconcentration will also progressively reduce the time scalefor adsorption onto the solid surface. Table 2 shows thatthe dewetting becomes apparent sooner at higher sur-factant concentrations, indicating that the latter effect isdominant.

Variations in the film thickness can also give rise tochanges in the significance of surfactant diffusion inrelation to Marangoni forces. As the film thicknessincreases, there is reduced viscous dissipation whichallows more rapid spreading. This reduces the Marangonitime scale in relation to the diffusion and adsorption time;a similar effect to increasing the surfactant concentration.However, more rapid flow will promote more severe filmthinning at the base of the drop, which will reduce thediffusion time of the surfactant through the bulk. Table2 reveals that, since dewetting becomes apparent morequickly on thinner films, the former effect dominates.

(32) Zajac, J.; Partyka, S. In Adsorption on New and ModifiedInorganic Sorbents Studies in Surface Science and Catalysis; Dabrowski,A., Tertykh, Y. A., Eds.; Elsevier Science: Amsterdam, 1996.

(33) Atkin, R.; Craig, V. S. J.; Biggs, S. Langmuir 2000, 16, 9374.(34) Gao, Y. Y.; Du, J. H.; Gu, T. R. J. Chem. Soc., Faraday Trans

1 1987, 83, 2671(35) Pagac, E. S.; Prieve, D. C.; Tilton, R. D. Langmuir 1998, 14,

2333.

Figure 7. A 9 µL drop of 0.4cmc (5.6 mM) DTAB solutionspreading on a 100 µm thick water film: (A) after 0.4 s, theleading edge of the surfactant is seen as a dark ring and alighter front is just visible inside this; (B) after 16 s, thesurfactant has retracted to a rounded cap in the center andcrenulations are evident in the growing outer dewetting front.Inferred side views are shown below the images.

Figure 8. A 9 µL drop of 1.6cmc (22.4 mM) DTAB solutionspreading on a 200 µm thick water film: (A) 0.5 s afterdeposition, the thinned region and elevated leading edge canbe seen; (B) 20 s after deposition, the spreading surfactant caphas retracted and the dewetting front has developed crenula-tions at late times. Inferred height profiles are given beneatheach image.

7580 Langmuir, Vol. 20, No. 18, 2004 Afsar-Siddiqui et al.

The results can now be explained in the light of theseconsiderations. Upon deposition of a 0.04cmc drop ofsurfactant, a surface tension gradient causes Marangoni

forces to spread the surfactant downstream. The spreadingexponent of 0.26 confirms that the flow at this stage isMarangoni driven. Because of the low bulk surfactantconcentration, the adsorption rate is low and it is likelythat, since the csac has not yet been reached, the surfactant

Figure 9. A 9 µL drop of 0.8cmc (11.2 mM) DTAB solution spreading on a 25 µm thick water film: (A) after 0.05 s, there is a frontand a flattened cap; (B) after 0.5 s; (C) after 19 s, the cap has remained at a constant radius throughout the spreading while theouter front has continued to grow. The outermost front is thought to be a shock wave. Inferred height profiles are given beneatheach image.

Figure 10. A 9 µL drop of 1.6cmc (22.4 mM) DTAB solution spreading on a 25 µm thick water film: (A) after 0.05 s, a cap of radius5 mm has formed, ahead of which is the dewetting front and a shock wave; (b) after 0.5 s, a cap of radius 5 mm has formed, aheadof which is the dewetting front and a shock wave; (C) after 8 s, the dewetting front has advanced only slightly. Inferred heightprofiles are also shown.

Figure 11. A 9 µL drop of 1.2cmc (16.8 mM) DTAB solutionspreading on a 50 µm thick water film: (A) after 0.5 s, thereis a cap ahead of which is a dewetting front and a shock wave;(B) after 16 s, the cap has remained at a constant radiusthroughout the spreading while the dewetting front hascontinued to grow. The outermost front is thought to be a shockwave.

Figure 12. A 9 µL drop of 4cmc (56 mM) DTAB solutionspreading on a 100 µm thick water film: (A) after 0.3 s, thereis a front and a flattened cap; (b) after 19 s, the cap has remainedat a constant radius throughout the spreading while the outerfront has continued to grow.

Dewetting Behavior of Surfactant Solutions Langmuir, Vol. 20, No. 18, 2004 7581

monomers adsorb onto the solid surface with their tailgroup lying horizontally on the surface. After severalseconds, there is sufficient adsorption at the solid-liquidinterface from the thinned region of the film to create ahole. The time taken for the hole to form is dependent onthe time taken for surfactant to diffuse through the bulkliquid film and adsorb onto the solid-liquid interface. Itis expected that this would occur sooner on an initiallythin film. However, the results show that hole formationoccurs at 8 s on all film thicknesses, the reasons for whichare not yet understood.

At higher surfactant concentrations, autophobing isseen. This case is similar to hole formation, but filmthinning results in the formation of a dry “ring” ratherthan a hole. This causes the fluid upstream of the dryregion to retract, while the downstream fluid moves awayfrom the dry patch via a dewetting front. The time offormation of the dry ring depends on the surfactantconcentration and film thickness. From Tables 2 and 3,it appears that the factors that promote film thinning(initially thin films and high surfactant concentrations)cause dewetting to become apparent sooner. At a con-centration of 0.4cmc on a 25 µm film and 0.8cmc on a 50µm film, the cap is seen to retract almost immediatelyupon surfactant deposition (Figure 5). This suggests that

the time scale for diffusion and adsorption is of the sameorder of magnitude as the Marangoni time scale, sincethe surfactant was not able to spread appreciably. At thesame concentrations on thicker films, the surfactant isseen to spread under the influence of Marangoni forcesbefore dewetting becomes apparent (Figures 6 and 7). Thisis because the ratio of the Marangoni time to the diffusivetime scale is reduced (i.e., it takes longer for surfactantto diffuse across a thick film). The relatively high contactangle, around 90°, of the retracted cap suggests that thedewetted region is highly hydrophobic and the surfactantis oriented with the hydrocarbon chains exposed. Thecrenulations in the dewetting front are thought to ariseas the front becomes unstable to small height andconcentration disturbances ahead. At concentrationsabove the cmc, the fluid retracts to a cap of smaller contactangle. At these concentrations, it is expected that theadsorbed surfactant will form a bilayer at the solid-liquidinterface, which presents a less hydrophobic surface tothe surfactant solution.

Over certain parameter ranges, the surfactant dropletdoes not appear to spread before retracting, but a flattenedcap forms upon surfactant deposition that neither spreadsnor retracts (cap formation). This indicates that the timescale for adsorption onto the solid surface is less than orof the same order of magnitude as the time scale forMarangoni flow. At this concentration, either a surfactantbilayer, or, more likely, adsorbed cylindrical micelles giventhe recent AFM images of Ducker et al.36 for DTABadsorbed to silica, will be present ahead of the cap ofsurfactant. This presents a less hydrophobic surface tothe surfactant than the hemimicelle formation at lowersurfactant concentrations and so the contact angle of thecap is smaller and the cap is more flattened (Figures 9-12).At 1.6cmc and 4cmc on a 25 µm thick film, the dewettingfront does not appear to advance downstream. This maybe due to the surfaces rapidly becoming hydrophilic onthese relatively thin water films. One would expect thesurface to be covered with adsorbed surfactant micelleswhich would be hydrophilic in nature, preventing furtherdewetting.

5. Conclusions

When the surfactant headgroup and substrate exhibita mutual attraction, then dewetting is seen to occur. Thedewetting patterns vary with surfactant concentrationand water film thickness. At low concentrations, dewettingproceeds through hole formation in the thinned region ofthe film. At higher concentrations, the thinned region ofthe film ruptures leaving a dewetted ring. Fluid upstreamof this retracts into a cap of surfactant at the point ofinitial deposition (autophobing). Downstream of the dryring liquid moves away from the hydrophobic region viaa dewetting front. At still higher concentrations, thedeposited surfactant forms a cap at the point of depositionthat neither spreads nor retracts (cap formation). Thisvariation of dewetting behavior can been explained byconsidering the relative Marangoni and bulk diffusiontime scales as well as the mode of assembly of thesurfactant adsorbed on the solid surface.

Acknowledgment. We are grateful to the EPSRC forfunding of this project.

LA040041Z

(36) Ducker, W. A.; Wanless, E. J. Langmuir 1996, 12, 5915.

Figure 13. Schematic representation of the events that leadto the formation of a dewetting hole: (a) Film thinning andsurfactant adsorption onto the solid surface preferentially fromthe Marangoni-thinned region as shown by the arrow. (b)Repulsion of water from the hydrophobic solid surface causingfurther film thinning. The arrows show the direction of waterflow. (c) Formation of a dry patch or hole following still furtherthinning by intermolecular forces. Water then moves away fromthe hole as shown by the arrows.

7582 Langmuir, Vol. 20, No. 18, 2004 Afsar-Siddiqui et al.