Physical Properties of Nanobubbles on Hydrophobic Surfaces in

We thank Chiara Neto for assistance with sample preparation and Mika Kohonen for macroscopic contact measurements. Tim Senden, Drew R. Evans, and ...
0 downloads 0 Views 662KB Size
Langmuir 2006, 22, 5025-5035

5025

Physical Properties of Nanobubbles on Hydrophobic Surfaces in Water and Aqueous Solutions Xue H. Zhang, Nobuo Maeda, and Vincent S. J. Craig* Department of Applied Mathematics, Research School of Physical Sciences and Engineering, Australian National UniVersity, Canberra ACT 0200, Australia ReceiVed January 18, 2006. In Final Form: March 22, 2006 In recent years there has been an accumulation of evidence for the existence of nanobubbles on hydrophobic surfaces in water, despite predictions that such small bubbles should rapidly dissolve because of the high internal pressure associated with the interfacial curvature and the resulting increase in gas solubility. Nanobubbles are of interest among surface scientists because of their potential importance in the long-range hydrophobic attraction, microfluidics, and adsorption at hydrophobic surfaces. Here we employ recently developed techniques designed to induce nanobubbles, coupled with high-resolution tapping-mode atomic force microscopy (TM-AFM) to measure some of the physical properties of nanobubbles in a reliable and repeatable manner. We have reproduced the earlier findings reported by Hu and co-workers. We have also studied the effect of a wide range of solutes on the stability and morphology of these deliberately formed nanobubbles, including monovalent and multivalent salts, cationic, anionic, and nonionic surfactants, as well as solution pH. The measured physical properties of these nanobubbles are in broad agreement with those of macroscopic bubbles, with one notable exception: the contact angle. The nanobubble contact angle (measured through the denser aqueous phase) was found to be much larger than the macroscopic contact angle on the same substrate. The larger contact angle results in a larger radius of curvature and a commensurate decrease in the Laplace pressure. These findings provide further evidence that nanobubbles can be formed in water under some conditions. Once formed, these nanobubbles remain on hydrophobic surfaces for hours, and this apparent stability still remains a well-recognized mystery. The implications for sample preparation in surface science and in surface chemistry are discussed.

Introduction Recent reports of stable nanobubbles on hydrophobic surfaces in water suggest that they may play a significant role in a number of phenomena.1 For example, nanobubbles were invoked to explain the long-range hydrophobic attractive force.2,3 The presence of nanobubbles will influence the boundary condition of fluid flow at a solid surface4-6 and the adsorption and immobilization of biomolecules.7 In engineering, nanobubbles have been used in the design of microdevices 8 and also as the template in the manufacture of nanostructures.9 Even though nanobubbles have attracted a lot of recent attention, the existence of nanobubbles as a stable and spontaneously forming entity has not been unequivocally established and is still disputed in the literature for a number of reasons. First of all, the total free energy of a system (or free energy per unit area) in which water is in contact with a hydrophobic surface always increases upon the formation of a gas layer or nanobubbles unless the surface is extremely rough. In addition, bubbles of nanometer size should dissolve and disappear rapidly because of the very high Laplace pressure inside the bubble.10,11 Despite * Corresponding author. Tel: +61-2-6125-3359. Fax: +61-2-6125-0732. E-mail: [email protected]. (1) Ball, P. Nature 2003, 423, 25. (2) Parker, J.; Claesson, P.; Attard, P. J. Phys. Chem. 1994, 98, 8468. (3) Attard, P. AdV. Colloid Interface Sci. 2003, 104, 75. (4) Tretheway, D. C.; Meinhart, C. D. Phys. Fluids 2004, 16, 1509. (5) Lauga, E.; Brenner, M. P. Phys. ReV. E: Stat., Nonlinear, Soft Matter Phys. 2004, 70. (6) de Gennes, P. G. Langmuir 2002, 18, 3413. (7) Wu, Z.; Zhang, X.; Zhang, X.; Gang, L.; Sun, J.; Zhang, Y.; Li, M.; Hu, J. Surf. Interface Anal. 2005, 37, 797. (8) Paxton, W. F.; Kistler, K. C.; Olmeda, C. C.; Sen, A.; St. Angelo, S. K.; Cao, Y. Y.; Mallouk, T. E.; Lammert, P. E.; Crespi, V. H. J. Am. Chem. Soc. 2004, 126, 13424. (9) Fan, Y. W.; Wang, R. Z. AdV. Mater. 2005, 17, 2384. (10) Eriksson, J. C.; Ljunggren, S. Colloids Surf., A 1999, 159, 159.

these thermodynamic objections, some experimental studies have supported the existence of nanobubbles at the solid/water interface, such as atomic force microscopy (AFM),8,12-18 cryofixation scanning electron microscopy (SEM),19 and neutron reflectivity.20 In contrast, other experimental results obtained using neutron reflectivity,21 electrochemical quartz crystal microbalance (eQCM),22 and ellipsometry 23 found no support for the existence of nanobubbles. The recent development of techniques for the controlled induction of nanobubbles permits some of the physical properties of nanobubbles to be investigated in a reliable and repeatable manner. Hu and co-workers24-29 found that long-lived nanobub(11) Ljunggren, S.; Eriksson, J. C. Colloids Surf., A 1997, 130, 151. (12) Agrawal, A.; Park, J.; Ryu, D.; Hammond, P.; Russell, T.; McKinley, G. Nano Lett. 2005, 5, 1751. (13) Holmberg, M.; Kuhle, A.; Garnaes, J.; Morch, K.; Boisen, A. Langmuir 2003, 19, 10510. (14) Ishida, N.; Inoue, T.; Miyahara, M.; Higashitani, K. Langmuir 2000, 16, 6377. (15) Otsuka, I.; Yaoita, M.; Higano, M.; Nagashima, S. Surf. ReV. Lett. 2003, 10, 337. (16) Simonsen, A.; Hansen, P.; Klosgen, B. J. Colloid Interface Sci. 2004, 273, 291. (17) Tyrrell, J.; Attard, P. Phys. ReV. Lett. 2001, 8717. (18) Yang, J. W.; Duan, J. M.; Fornasiero, D.; Ralston, J. J. Phys. Chem. B 2003, 107, 6139. (19) Switkes, M.; Ruberti, J. Appl. Phys. Lett. 2004, 84, 4759. (20) Steitz, R.; Gutberlet, T.; Hauss, T.; Klosgen, B.; Krastev, R.; Schemmel, S.; Simonsen, A.; Findenegg, G. Langmuir 2003, 19, 2409. (21) Doshi, D.; Watkins, E.; Israelachvili, J.; Majewski, J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 9458. (22) Tsionsky, V.; Kaverin, A.; Daikhin, L.; Katz, G.; Gileadi, E. Phys. Chem. Chem. Phys. 2005, 7, 1830. (23) Mao, M.; Zhang, J. H.; Yoon, R. H.; Ducker, W. A. Langmuir 2004, 20, 1843. (24) Lou, S.; Ouyang, Z.; Zhang, Y.; Li, X.; Hu, J.; Li, M.; Yang, F. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.sProcess., Meas., Phenom. 2000, 18, 2573. (25) Lou, S.; Gao, J.; Xiao, X.; Li, X.; Li, G.; Zhang, Y.; Li, M.; Sun, J.; Hu, J. Chin. Phys. 2001, 10, S108.

10.1021/la0601814 CCC: $33.50 © 2006 American Chemical Society Published on Web 04/26/2006

5026 Langmuir, Vol. 22, No. 11, 2006

bles can be formed when a substrate that is first in contact with ethanol is subsequently exposed to water. Once formed, these nanobubbles remain on hydrophobic surfaces for hours. Given that the lack of reproducibility and repeatability of studies on nanobubbles in the past has led to confusion in the relevant literature, it is imperative to study the physical properties of these induced nanobubbles because they can be used to test whether the substance constituting these nanobubbles is indeed gas. The surprising stability of nanobubbles suggests that the nanobubble/solution interface may be contaminated with a “skin” that inhibits both the dissolution of gas and the coalescence between nanobubbles. Because of the very low surface area of nanobubbles and the difficulties associated with ensuring that solutions and materials are completely free of contaminants, it is impossible to completely remove all sources of contamination. Rather, here we introduce surface-active materials that should overwhelm the influence of any contaminant to probe the possible influence of a nanobubble skin. We also add electrolytes, which should alter the properties of all contaminants, particularly charged species. Induced nanobubbles can be used to investigate the formation, morphology, and apparent stability of the nanobubbles and reveal the conditions under which nanobubbles are formed. This will allow researchers to employ sample preparation techniques designed to produce or avoid the presence of nanobubbles at a surface with confidence. This is important in a number of areas. For example, the great variation reported in force measurements between hydrophobic surfaces can be attributed to the lack of experimental control of nanobubbles at the interface.30 This manuscript reports studies of the properties of induced nanobubbles in a variety of solutions to understand the parameters that influence the stability and morphology of nanobubbles. Materials and Methods Substrates. 1. Octadecyltrichlorosilane (OTS) Silicon. We followed the method described by Wang et al.31 First, silicon wafers (monitor wafer, MEMC, St. Peters, MO) were cut into 20 mm wide strips and cleaned using a freshly prepared hot piranha solution of 7:3 (v/v) H2SO4 (96%) and H2O2 (30%) for 30 min, then rinsed with a large quantity of Milli-Q water and distilled ethanol. Wafers were dried under a stream of nitrogen gas immediately before deposition of OTS. Deposition was conducted in a laminar flow cabinet in a clean room (∼24 °C, relative humidity (RH): ∼47%), using 0.25 mL of anhydrous OTS (90+% purity, Aldrich) dissolved in 50 mL of bicyclohexyl (g99% purity, Sigma-Aldrich). The clean silicon substrates were immersed in the OTS solution for 24 h, within a sealed container. Upon removal, they were quickly rinsed with anhydrous chloroform (g99+% pure, Sigma-Aldrich) and sonicated for 15 min each in chloroform, toluene, and ethanol, respectively. After drying under a stream of nitrogen gas, the substrates were kept in a clean room for at least 24 h. Before use, they were sonicated in acetone and ethanol for 5 min each and dried under a stream of nitrogen gas. They were further cleaned under a stream of CO2 snow for at least 5 s to remove any dust particles.32 Ice on the substrate was rinsed away with ethanol, and the ethanol was removed under a stream of nitrogen gas. A tapping-mode AFM (TM-AFM) image of an OTS silicon substrate thus prepared is shown in Figure 1a. (26) Lou, S.; Gao, J.; Xiao, X.; Li, X.; Li, G.; Zhang, Y.; Li, M.; Sun, J.; Li, X.; Hu, J. Mater. Charact. 2002, 48, 211. (27) Zhang, X.; Zhang, X.; Lou, S.; Zhang, Z.; Sun, J.; Hu, J. Langmuir 2004, 20, 3813. (28) Zhang, X.; Li, G.; Wu, Z.; Zhang, X.; Hu, J. Chin. Phys. 2005, 14, 1774. (29) Zhang, X.; Wu, Z.; Zhang, X.; Li, G.; Hu, J. Int. J. Nanosci. 2005, 4, 399. (30) Christenson, H. K.; Claesson, P. M. AdV. Colloid Interface Sci. 2001, 91, 391. (31) Wang, M. J.; Liechti, K. M.; Wang, Q.; White, J. M. Langmuir 2005, 21, 1848. (32) Chow, B. Y.; Mosley, D. W.; Jacobson, J. M. Langmuir 2005, 21, 4782.

Zhang et al.

Figure 1. TM-AFM images of OTS silicon (a) and HOPG (b) in air. The root-mean-square (RMS) roughness is 0.25 and 0.7 nm, respectively. The scan size for both is 5 × 5 µm, and the data scale is 5 nm in panel a and 10 nm in panel b. The freshly cleaved HOPG surface consists of atomic steps and steps of several or dozens of atomic layers. 2. Highly Ordered Pyrolytic Graphite (HOPG). HOPG (ZYB quality from MikroMasch, Madrid, Spain) was freshly cleaved immediately before each experiment by peeling off the outermost layers with adhesive tape. A typical image immediately after cleavage is shown in Figure 1b. Measurement Methods. 1. TM-AFM Imaging. A multimode Nanoscope IIIa scanning probe microscope (SPM, Digital Instruments, Veeco Metrology Group, Plainview, NY) with a fluid cell was used to image substrates and nanobubbles, and an MFP-3D atomic force microscope (Asylum Research, Santa Barbara, CA) was used to calibrate the spring constant of the cantilever. Noncontact silicon cantilevers (Ultrasharp Cantilever & Gratings, Silicon-MDT Ltd., Moscow, Russia) with a typical spring constant of 7.5 N/m were used for imaging the substrate in tapping mode (TM) in air. “V”-shaped NP cantilevers (Digital Instruments, Veeco Metrology Group, Plainview, NY) with nominal spring constants of 0.32 or 0.58 N/m were used for TM images in liquid (resonant frequency 6-12 kHz in water). The cantilevers were cleaned by immersion in predistilled ethanol and water for half an hour before being treated with water vapor plasma using a custom-built plasma reactor (30 W, 20 s). During imaging, the height images of nanobubbles were recorded at different set point ratios, rsp ) A/A0, where A0 is the free amplitude of the cantilever, and A is the set point amplitude used during imaging. 2. Force-Volume. Because it is extremely challenging to image nanobubbles using contact mode, we did not attempt to locate the AFM tip on nanobubbles using contact mode to obtain the deflectionpiezo displacement curve. Rather, after confirming the presence of nanobubbles on a substrate by TM-AFM imaging, we utilized the force-volume function to map the interaction between the surface and the AFM tip for the entire area. The main parameters for this mode were set as follows: trig mode: relative; threshold deflection: 150 nm; tip velocity: 5 µm/s; sample per line: 64; number of samples: 64; force per line: 64. The contrast between the nanobubbles and the substrate was clear from a force-volume image, and we could extract the force curve on the nanobubbles as well as on the bare substrate. 3. Charge-Coupled DeVice (CCD) Camera. Our AFM instrumentation included a long focal-length optical microscope and a CCD which allowed recording of (low-resolution) optical images of the substrate area. Typical magnification of the optical microscope was about 250×. 4. Contact-Angle Goniometry. We measured the macroscopic contact angle of water and surfactant solutions on HOPG and OTS silicon substrates that were prepared using procedures identical to those used for TM-AFM imaging. An optical contact-angle goniometer with automatic dispenser (CAM 200, KSV Instruments Ltd.) was used to record dynamic contact angles. Sample liquid was expelled from a precleaned syringe at a constant and precalibrated rate of 0.2 µL/s onto a given substrate. The profile of the advancing droplet was recorded using a video camera. Formation of Nanobubbles. Nanobubbles were deliberately produced by displacing (predistilled) ethanol with water (purified

Nanobubbles on Hydrophobic Surfaces in Water

Figure 2. TM-AFM images of OTS silicon in water before (a) and after (b) ethanol-water exchange. Irregularly shaped (noncircular three-phase contact line) nanobubbles (c) can be occasionally detected on OTS silicon substrates. Scan size: 8 × 8 µm; data scale: 20 nm. by a Mill-Q system, Millipore Corporation, Boston, MA), as described previously.24,27 The displacement of ethanol was achieved by slowly injecting a sufficient amount of water (∼10 mL) using a precleaned syringe from one of the two ports of the AFM fluid cell and letting the ethanol flow out of the cell from the other port (we refer to this procedure as “ethanol-water exchange”). The detailed procedure involved first injecting water into the fluid cell and then capturing a TM-AFM image. After confirming that there was no contamination on the surface, ethanol injection replaced the water, and another TM-AFM image was captured. Finally we injected more water to replace the ethanol in the fluid cell (ethanol-water exchange). This resulted in the production of nanobubbles. To measure the physical properties of nanobubbles in an aqueous solution other than water, nanobubbles were produced in water, which was subsequently replaced with the solution of interest. We allowed approximately 20-30 min of equilibration time before the surface was imaged by TM-AFM. Calculation of Nanoscopic Contact Angle from TM-AFM Images of Nanobubbles. The bubbles were nanoscopic in height but microscopic laterally; nonetheless, we follow convention in this manuscript and refer to them as nanobubbles and refer to the contact angle as being nanoscopic. The nanobubble contact angle is defined as the angle subtended by the liquid phase (i.e., the denser phase) adjacent to a nanobubble. A cross-section or a profile of a nanobubble deduced from an AFM image can be fitted by a spherical cap corrected for the local curvature of the AFM tip (radius 20 nm), apart from a small deviation near the substrate. The contact angle of a nanobubble was inferred from this and the position of the substrate. The deviation at the edges of the nanobubbles is attributed to the interaction between the tip and the substrate.

Results Formation of Nanobubbles in Water. Prior to ethanolwater exchange, nanobubbles were not detected by TM-AFM on freshly cleaved HOPG surfaces, as reported previously,29 nor on the OTS silicon surfaces (Figure 2a), which is contrary to some previous reports.14,16,17,20 This may in part be due to different surface preparation procedures.33 However, nanobubbles were reproducibly detected by TM-AFM after the exchange of ethanol with water (Figure 2b,c) on both substrates. We note that OTS silicon has not been used as the hydrophobic substrate by others who use the ethanol-water exchange as a means to produce nanobubbles. The formation of nanobubbles was more efficient when warm ethanol and water (∼45 °C) were used, than when the solvents were used at room temperature (∼23 °C), as reported by Zhang et al.27 The apparent height of these nanobubbles measured from TM-AFM images varied from a few nanometers to several hundreds of nanometers. The lateral dimensions varied from several hundreds of nanometers to a few microns. The three-phase line of the nanobubbles is usually circular, although, on several occasions, we could also detect nanobubbles that had an irregularly shaped (noncircular) three-phase line on the OTS silicon substrates, as shown in Figure 2c. In contrast, such (33) Evans, D. R.; Craig, V. S. J.; Senden, T. J. Physica A 2004, 339, 101.

Langmuir, Vol. 22, No. 11, 2006 5027

irregularly shaped (noncircular) three-phase lines were not detected on the HOPG substrates. In addition to the nanobubbles that could only be detected by TM-AFM, larger (radius: ∼10 micrometers) bubbles were often observed through the CCD (Supporting Information, Figure 1). These micrometer-sized bubbles were observed on the OTS silicon substrates, but not on the HOPG substrates. We observed that some of these micrometer-sized bubbles exhibited noncircular three-phase contact lines that remained unchanged on the substrate for many hours. The spatial distribution of the micrometer bubbles was not uniform over the substrate. There was a clear correlation between the spatial distribution of the nanobubbles and that of the micrometer-sized bubbles. That is, a larger population of nanobubbles was detected using TM-AFM when the cantilever was placed in the vicinity of the micrometer-sized bubbles. This observation precludes meaningful assessment of the average surface number density of nanobubbles produced, so we are unable to compare our measurements with other reports of nanobubble coverage. Morphology of Nanobubbles in Water. 1. Contact Angle of Nanobubbles Inferred from TM-AFM Images. We calculated the contact angle of approximately 100 nanobubbles on HOPG and approximately 40 nanobubbles on OTS silicon. The nanoscopic contact angle obtained using the procedure described in the Materials and Methods section is 168 ( 9° on OTS silicon and 164 ( 6° on HOPG. These values are much greater than the macroscopic advancing contact angles of water droplets on these substrates, as measured using the contact-angle goniometer: 108 ( 5° on OTS silicon and 72 ( 11° on HOPG. 2. TM-AFM Imaging at Different Set Point Ratios. One possible source of the large discrepancy between the macroscopic and nanoscopic contact angles is that the profile of the nanobubbles deduced from TM-AFM images may be altered because of possible deformations of the nanobubbles during imaging. A compression of the nanobubble could lead to an underestimation of the nanobubble height and an overestimation of the contact angle. To examine this point, we imaged nanobubbles using different imaging set point amplitudes. The set point is used to control the imaging force between the tip and the surface, a lower set point amplitude is indicative of a larger imaging force. TM-AFM images of nanobubbles using different set point amplitudes (Asp) are shown in Figure 3 and Supporting Information Figure 2. The set point amplitude ratio, rsp ) Asp/A0 was varied from 0.926 to 0.741. The first point we note is that the TM-AFM images of nanobubbles were surprisingly insensitive to rsp. The images are almost indistinguishable from each other, even though the set point was varied from 0.926 to 0.741. Consequently, the cross-sectional profile of the nanobubbles (Figure 3c) shows a very small variation with respect to rsp in the range studied. The average height of the nanobubble is 26 ( 1 nm, the lateral size is 591 ( 15 nm, the radius of curvature is 1497 ( 106 nm, and the nanoscopic contact angle is 169 ( 1° over this range of rsp (see Supporting Information, Table 1). A typical amplitude versus piezo displacement curve obtained on a nanobubble is shown in Figure 4. The amplitude of the cantilever decreases sharply over a very narrow separation (piezo displacement) range. Therefore a broad selection of set point ratios will result in very similar feature heights (i.e., z-distance values) of the nanobubbles. The underlying physical reason for this behavior is not clear, and we therefore studied force curves (without oscillation) on nanobubbles to obtain further insight. 3. Reconstruction of Nanobubble Profiles from Force CurVes. To obtain deflection versus piezo displacement curves on nanobubbles, force-volume imaging was applied to the area in

5028 Langmuir, Vol. 22, No. 11, 2006

Figure 3. TM-AFM images of nanobubbles imaged using different set point amplitude ratios. Set point ratio, rsp: (a) 0.926; (b) 0.741. Scan size: 1.5 × 1.5 µm; data scale: 75 nm. Spring constant of cantilever: 0.079 N/m; scan rate: 1.57 Hz; drive frequency: 6.58 kHz. (c) Cross-sectional profiles of the nanobubble on OTS silicon imaged using set point ratios between 0.926 and 0.741. The solid line is a fit to the data using a spherical cap.

Figure 4. Typical amplitude-z-piezo displacement curve obtained on a nanobubble in water. The z-piezo displacement describes the vertical movement of the cantilever tip away from the substrate. Note that the amplitude changes dramatically over a small separation, indicating that changes to the set point amplitude (the feedback setting) will not result in significant changes in the z-piezo displacement. This is consistent with the images presented in Figure 3a,b, Supporting Information Figure 2, and profiles presented in Figure 3c.

which the presence of nanobubbles had been detected by TMAFM. Figure 5 shows different slices of a typical force-volume image at different piezo displacements when the AFM tip is first moved toward and then away from the surface. The contrast between the nanobubbles and the substrate becomes very sharp at a well-defined distance. The deflection versus piezo displacement curves at different locations could be extracted from the force-volume data, as shown in Figure 6a,b. A jump-in occurred during the approach (Figure 6a,c) both on nanobubbles and on OTS silicon substrates. After the jump-in, the extension curve on nanobubbles initially exhibited a linear regime with a slope much smaller than that on the substrate. With further approach, the slope of the force curve increased abruptly to a value identical to that obtained on bare substrate. This indicated that the tip had penetrated the nanobubble and was now in contact with the substrate. The interaction between a spherical particle and bubbles has been studied in different systems.34-36 The bubble-water interface is negatively

Zhang et al.

charged,36-38 and the silicon nitride probe we used was also reported to have a weak negative charge in water.39 We note that we see a very weak repulsive force before the jump-in in some force curves on nanobubbles, which is difficult to see on the scale used in Figure 6. The cantilever apparently penetrates the nanobubbles once the tip has touched the nanobubble surface. We interpret the position where the tip first touches the nanobubble surface as the top of the nanobubble. After the jump-in, the nanobubble apparently shows a finite resistance to further compression until the AFM tip reaches the underlying (hard) substrate. This is evidence that the tip penetrates the nanobubble and continues to do so until it reaches the underlying substrate, as indicated by the position where the compliance coincides with that on the substrate. We calculate the height of the nanobubble from the distance at which the force deviates from the baseline in the force versus piezo displacement plots. A few interesting features appear in the retracting branch of the force curves taken on nanobubbles in Figure 6b,d. The retraction curve initially appears to be almost identical to that obtained on unadorned OTS silicon substrates down to a small repulsive force. Then the AFM tip detaches from the substrate at a small but positive force; that is, there is a repulsive force acting on the cantilever. One possible explanation is that the nanobubble may become highly deformed as it is penetrated by the tip, presumably because of the pinning of the three-phase line around the AFM tip. Given that the AFM tip is weakly hydrophilic, we expect the receding contact angle of water in air to be small. This would result in a large deformation of the nanobubble around the AFM tip. Such deformation remains until the AFM tip is being retracted from the substrate. Given that no measurable adhesion force (jump-out) was observed as the AFM tip detached from the substrate, the restoring force arising from the deformed nanobubble is significant. We analyzed all extending branches of the force curves taken along a section of a number of nanobubbles and, from these, reconstructed profiles of the nanobubbles, as shown in Figure 7. A similar method has been used to reconstruct the profile of other soft materials.40 For nanobubbles with approximately the same lateral size, the height deduced from TM-AFM imaging is less than half of that deduced from the force curves. This result, taken together with the earlier result that the morphology of nanobubbles deduced from TM-AFM imaging was not sensitive to the set point amplitude, suggests that nanobubbles are deformed during TM-AFM imaging. Nevertheless, the deformation of nanobubbles during TM-AFM imaging is too small to account for the vast discrepancy between the nanoscopic and macroscopic contact angles, as the nanoscopic contact angle of water deduced from the TM-AFM height image was 175° and that deduced from the force curves was 168°. Effects of Cationic, Anionic, and Nonionic Surfactants on the Morphology and Stability of Nanobubbles. To obtain further insight into the physical properties of nanobubbles, we examined nanobubble morphology and stability upon the addition of a surfactant to the aqueous phase after the formation of nanobubbles. Cationic (cetyltrimethylammonium bromide (34) Ducker, W.; Xu, Z.; Israelachvili, J. Langmuir 1994, 10, 3279. (35) Nguyen, A. V.; Evans, G. M.; Nalaskowski, J.; Miller, J. D. Exp. Therm. Fluid Sci. 2004, 28, 387. (36) Preuss, M.; Butt, H. J. Langmuir 1998, 14, 3164. (37) Usui, S.; Sasaki, H.; Matsukawa, H. J. Colloid Interface Sci. 1981, 81, 80. (38) Abdel-Fattah, A. I.; El-Genk, M. S. AdV. Colloid Interface Sci. 1998, 78, 237. (39) Miyatani, T.; Okamoto, S.; Rosa, A.; Marti, O.; Fujihira, M. Appl. Phys. A: Mater. Sci. Process. 1998, 66, S349. (40) Connell, S. D. A.; Allen, S.; Roberts, C. J.; Davies, J.; Davies, M. C.; Tendler, S. J. B.; Williams, P. M. Langmuir 2002, 18, 1719.

Nanobubbles on Hydrophobic Surfaces in Water

Langmuir, Vol. 22, No. 11, 2006 5029

Figure 5. A sequence of slices of force-volume data as the tip is moved through a complete force-curve cycle. Extend: panels a-d; Retract: panels e-i. The piezo displacement for each image is indicated in Figure 6. For all images, scan size: 3.2 × 3.2 µm; z display scale: 25 nm. Note that the difference between the images obtained at the same displacement upon extension and retraction indicates significant hysteresis in the force curve.

Figure 6. Typical deflection-z-piezo displacement curves (a,b) and corresponding force-separation curves (c,d) on nanobubbles and on bare OTS silicon substrate. The selection of different force-volume images in Figure 5 corresponds to the points labeled in the force curves of panels a and b. The spring constant of the cantilever for the force measurement was 0.04 N/m.

Figure 7. Profiles of the nanobubble deduced from force curves (O) and from a TM-AFM height image (]). The solid lines indicate fitted spherical caps used for determination of the contact angle. The spring constant of the cantilever for the force measurement was 0.04 N/m, and that for the tapping mode image was 0.08 N/m. The height of the nanobubble deduced from TM-AFM height images is smaller than that deduced from the force curves, whereas the lateral size of the nanobubbles is similar. This discrepancy in the nanobubble height resulted in a difference in the measured nanoscopic contact angle of 175° - 168° ) 7°.

(CTAB)), anionic (sodium dodecyl sulfate (SDS)) and nonionic (Tween 20) surfactants were investigated. 1. Effect of Surfactant Solution on the Morphology of Micrometer-Sized Bubbles. CCD images revealed that the boundary of micrometer-sized bubbles on an OTS silicon substrate immediately relaxed to a circular shape (from the irregular shapes sometimes observed in water) after the surfactant solution was injected. These bubbles remained stable for hours, allowing sufficient time for detailed experiments. Figure 8 shows a typical CCD picture of micrometer-sized bubbles in SDS (0.86 critical micelle concentration (CMC)). The other two surfactant solutions had a similar effect, also reducing the pinning around the perimeter of the nanobubbles. 2. Effect of Surfactant Solution on the Morphology and Stability of Nanobubbles and the Nanoscopic Contact Angle. First, we note that nanobubbles on OTS silicon were stable in CTAB,

Figure 8. Picture of micrometer-sized bubbles in 0.86 CMC SDS solution on OTS silicon observed from the CCD.

Figure 9. Typical TM-AFM images of nanobubbles on OTS silicon in (a) 0.5 CMC (5 × 10-4 M) CTAB solution, (b) 0.86 CMC (6.3 × 10-3 M) SDS solution, and (c) 0.5 CMC (4.5 × 10-5 M) Tween 20 solution. Scan size: (a,c) 10 × 10 µm, (b) 30 × 30 µm. Data scale for all images: 100 nm.

SDS, and Tween 20 surfactant solutions of different concentrations. From TM-AFM images, we found that the morphology of nanobubbles in surfactant solutions was different from that in water. As was the case for the micrometer-sized bubbles, the irregular shape of the three-phase contact line of nanobubbles on OTS silicon substrates in water became circular in surfactant solution, as shown in Figure 9. This result provides further evidence that the nanobubbles consist of gas, since the same behavior was observed for microscopic bubbles. No change in contact-line shape was observed for HOPG upon the addition of

5030 Langmuir, Vol. 22, No. 11, 2006

Zhang et al.

Figure 10. Typical TM-AFM height images of nanobubbles on HOPG in (a) water, (b) 0.86 CMC SDS solution, and (c) 0.5 CMC CTAB solution. Scan size: (a,b) 5 × 5 µm, (c) 10 × 10 µm. Data scale for all images: 100 nm.

surfactant. Note that the nanobubble three-phase line is already circular in water (Figure 10). Figure 11 summarizes the nanoscopic contact angle deduced from TM-AFM images of nanobubbles in water and surfactant solutions and the macroscopic counterpart measured using a contact-angle goniometer. The difference between the nanoscopic and the macroscopic contact angles on either surface for a given solution remains remarkably large, and, from this limited study, the influence of all the surfactants is qualitatively similar, despite the surfactants having very different hydrophilic headgroups. Figure 11c shows the radius of curvature in relation to the nanoscopic contact angle. This shows that the nanoscopic contact angle remains constant, within error, over the size range of nanobubbles observed, although we know the contact angle must fall to the macroscopic value as the size of the bubbles increase. The bubble size range where this occurs (2-100 µm) was not accessible in our experiments. 3. TM-AFM Imaging Nanobubbles in Surfactant Solution at Different rsp Values. We systematically collected images of nanobubbles in Tween 20 solution using different set point amplitude ratios, rsp, to investigate the effect of deformation arising from TM-AFM imaging, as we did for water. We first note that, unlike in water, the profile of the nanobubbles deduced from TM-AFM images is strongly dependent on rsp. Figure 12 shows a series of TM-AFM images of an area that was scanned in a 0.5 CMC Tween 20 solution using different rsp values. The apparent profile of nanobubbles in Tween 20 solution clearly became smaller as the set point was decreased. The profile of the nanobubble (Figure 13) could still be approximated by a spherical cap down to rsp ) 0.806. However, the height of the nanobubble was reduced to almost half, and the lateral dimension of the nanobubbles also diminished as rsp was decreased from rsp ) 0.968 to rsp ) 0.774. This result suggests severe deformation of nanobubbles during TM-AFM imaging. Interestingly, reduction of the set point resulted in a decrease in both the height and the radius of the spherical cap, leading to similar nanoscopic contact angles for all the profiles. The contact angle obtained by fitting spherical caps to the nanobubble profiles at different rsp values are summarized in Table 1. No hysteresis was observed in the nanobubble compression, as shown in Figure 13b and Table 1. This result shows that the gas inside the nanobubble was not removed when the nanobubble was severely deformed by the ATM tip. We note that, as has been the case for water, the difference between the nanoscopic and macroscopic contact angles cannot be attributed to the uncertainty in the profile of the nanobubbles deduced from the AFM measurements. A typical amplitude versus piezo displacement curve for a nanobubble in a 0.5 CMC Tween 20 solution is shown in Figure 14. The amplitude decreased slowly as the tip approached the surface. This is consistent with the result that the profile of a nanobubble changes with the set point amplitude, a markedly different feature from the result obtained in water (Figure 4).

Figure 11. Nanoscopic and macroscopic contact angles of water and surfactant solutions on HOPG (a) and on OTS silicon (b) substrates. The measurements of macroscopic contact angle were taken within 20 min of the HOPG being cleaved. Nanoscopic contact angles calculated from nanobubbles of different sizes in water on HOPG substrates are shown in panel c. The nanoscopic contact angle remains constant, within error, in the size of nanobubbles observed, although we know the contact angle must fall to the macroscopic value as the size of the bubbles increase.

4. Force CurVes on Nanobubbles in Surfactant Solution. As reported above for water, deflection-piezo displacement curves were obtained by force-volume mapping in surfactant solutions. Typical force curves on nanobubbles and on bare OTS silicon substrates in 0.5 CMC Tween 20 solution are shown in Figure 15. In stark contrast to the case in water reported above, no jump-in was observed on nanobubbles or on bare OTS silicon substrates during the approach. Instead, the interaction force was monotonically repulsive down to the compliance regime. The repulsive force on the approach, prior to the compliance regime, is longer in range on nanobubbles than on bare substrates. This long-range repulsive force on nanobubbles was observed

Nanobubbles on Hydrophobic Surfaces in Water

Langmuir, Vol. 22, No. 11, 2006 5031

Figure 12. AFM images of nanobubbles in a 0.5 CMC Tween 20 solution under different set point amplitude ratios. The set point was first decreased from panel a to panel g and then increased from panel h to panel j. Set point ratio: (a) 0.968, (b) 0.935, (c) 0.903, (d) 0.871, (e) 0.839, (f) 0.806, (g) 0.774, (h) 0.839, (i) 0.903, and (j) 0.935. Scan size for all images: 700 × 700 nm. Data scale for all images: 40 nm. The spring constant of the cantilever was 0.8 N/m, the scan rate was 3 Hz, and the drive frequency was 6.58 kHz. Table 1. Details from the Profiles of Nanobubbles in 0.5 CMC Tween 20 Solution Obtained under Different Set Point Amplitude Ratios set point amplitude ratio

radius of curvature (nm)

height of nanobubble (nm)

lateral size of nanobubble (nm)

nanoscopic contact angle (°)

0.968 0.935 0.903 0.871 0.839 0.806 0.774a 0.839b 0.903b 0.935b

280 260 230 205 195 130 75 195 210 245

35 30 27 25 22 19 17 23 26 28

315 303 301 304 298 309 306 309 304 302

151 152 152 152 154 150 147 152 151 152

a Nearly fit to a spherical cap. b Increased the set point amplitude ratio after the decrease of it.

Figure 13. Cross-section of the nanobubble shown in Figure 12, (0.5 CMC Tween 20 solution). Panel a shows data obtained while decreasing the set point, which corresponds to the images in Figure 12a-g. Panel b shows data obtained while subsequently increasing the set point (gray lines), which corresponds to the images in Figure 12h-j, and these are compared to the profiles obtained at the same set point in panel a shown as solid black lines. The value of the set point is indicated by the numbers in the panels.

for both charged (cationic, anionic) and uncharged (nonionic) surfactant species, and therefore cannot be attributed to electrostatic repulsive forces between the tip and the nanobubble. This repulsion is attributed to the deformation of the nanobubble. Given that there is no sign of jump-in during the approach, we can assume that the surface of the nanobubble was not punched through by the AFM tip. Indeed, Figure 12 suggests that the surface of the nanobubble in Tween 20 solution is highly deformable, and this results in resilience to being punctured by the tip of the cantilever. The observed repulsive force is attributed to the deformation of the nanobubble surface while the AFM tip is in contact with the nanobubble. It was reported that a fluid

Figure 14. Typical amplitude-z-piezo displacement curve on a nanobubble in 0.5 CMC Tween 20 solution, indicating that the amplitude reduces gradually as the tip sample separation is altered. This is indicative of a very soft interaction and is commensurate with the observation that the images obtained were highly dependent upon rsp, as shown in Figure 12.

interface with an interfacial tension γ essentially acts like an elastic spring with a spring constant of γ.41-43 If we assume that the cantilever deflection is balanced by the restoring force of the nanobubble surface, we can roughly estimate the interfacial energy of the bubble surface in Tween 20 solution from the extending (41) Chan, D.; Dagastine, R.; White, L. J. Colloid Interface Sci. 2001, 236, 141. (42) Attard, P.; Miklavcic, S. Langmuir 2003, 19, 2532. (43) Attard, P.; Miklavcic, S. Langmuir 2001, 17, 8217.

5032 Langmuir, Vol. 22, No. 11, 2006

Zhang et al.

Figure 15. The deflection-z-piezo displacement curves (a,b) and the corresponding force-separation curves (c,d) on nanobubbles and on OTS silicon substrates in 0.5 CMC Tween 20 solution. The spring constant of the cantilever for the force measurement was 0.387 N/m.

branch of the measured deflection versus z-piezo displacement curve in Figure 15a. To this end, first let us denote the spring constant of the nanobubble surface and the AFM cantilever as knanobubble and kcantilever, respectively, and the deflection of each as xnanobubble and xcantilever, respectively. Then the z-piezo displacement, z, is given by

z ) xnanobubble + xcantilever

(1)

and the balance of the force yields

F ) knanobubble xnanobubble ) kcantilever xcantilever

(2)

Now we are in a position to calculate the interfacial tension of the nanobubble surface in Tween 20 solution. Although the nanobubble compliance portion of the extending branch of Figure 15a is not completely linear, we can deduce that the z-piezo displacement of z ≈ 50 nm (from z ) 300 nm to z ) 250 nm) roughly corresponds to the AFM cantilever deflection of xcantilever ≈ 5 nm. On the basis of the spring constant of the AFM cantilever we used (0.387 N/m; independently calibrated by the thermal tune method), we can calculate the interfacial tension of the nanobubble surface in Tween 20 solution as

γ ≈ knanobubble ) (kcantilever xcantilever)/(z - xcantilever)

(3)

) 0.387 N/m × 5 nm/(50 nm - 5 nm) ) 43 mN/m. Our estimate compares well with the literature value of ∼40 mN/m at the same concentration.44,45 Unfortunately, we cannot apply the same treatment to determine the surface tension of nanobubbles in pure water because, in water, the tip penetrates the interface. 5. Lift-Mode Images in Tween 20 Solution. We made an attempt to obtain further insight into the influence of the interaction between the nanobubble surface and the AFM tip by utilizing “lift-mode imaging”. The lift-mode amplitude images of nanobubbles in 0.5 CMC Tween 20 solutions are shown in Figure 3 of the Supporting Information. Evaluation of the nanobubble height based on the lift-mode images shows that the contact angle of (44) Ruiz, C. C.; Molina-Bolivar, J. A.; Aguiar, J.; MacIsaac, G.; Moroze, S.; Palepu, R. Colloid Polym. Sci. 2003, 281, 531. (45) Miller, R.; Fainerman, V. B.; Leser, M. E.; Michel, A. Colloids Surf., A 2004, 233, 39.

Figure 16. Nanobubbles on an HOPG surface in 0.5 M NaCl. After the formation by ethanol-water exchange, nanobubbles did not disappear after water was replaced by 0.5 M NaCl solution. Similar results were obtained for all other salt, acid, and base solutions studied.

the nanobubbles is still much larger than that observed for macroscopic bubbles. Effects of Salts and pH on the Stability and Other Physical Properties of Nanobubbles. To obtain further insight into the physical properties of nanobubbles, we examined whether any change in the morphology or stability could be induced by the addition of monovalent or divalent salts or acids or bases. We used NaCl (1 M), Na2SO3 (0.5 M), Na3PO4 (0.1 M), CH3COONa (1 M), and Ca(NO3)2 (0.25 M) salt solutions. We also employed acidic solutions, H2SO4 (0.1 M) and HCl (0.1 M), and basic solutions, KOH (0.01 M) and NaOH (0.01 M). We limited the use of bases to moderate concentrations to avoid etching of the optical surfaces of the glass AFM fluid cell. We found that the nanobubbles, once formed, were insensitive to the addition of salts. They were also found to be insensitive to the pH of the aqueous phase. An example of a TM-AFM image in 0.5 M NaCl solution is shown in Figure 16. No difference was seen in the stability or morphology of the nanobubbles between water and any of the salt solutions investigated. Initial observations of TM-AFM images in acid and base solutions also showed no obvious difference from those obtained in water.

Discussion Before we discuss the specific details of our findings, it is important to acknowledge that the total free energy of a system (or free energy per unit area) in which water is in contact with a smooth hydrophobic surface always increases upon the

Nanobubbles on Hydrophobic Surfaces in Water

formation of a gas layer or the introduction of nanobubbles, regardless of the contact angle, unless the aqueous phase is supersaturated with the gas. This is due to (1) the large surface energy of water (about 72 mN/m) compared to the interfacial energy between water and a hydrophobic surface (typically 50 mN/m) and (2) the formation of a gas layer or nanobubbles introducing an additional interface (between the hydrophobic substrate and gas) to the system that has a relatively small, but nonetheless positive, surface energy. Thus, our finding that no nanobubbles were detected by TM-AFM prior to ethanol-water exchange (Figure 2a) is significant in itself because this finding is in accordance with the thermodynamic principle mentioned above and therefore adds extra credibility to the other findings we discuss below. Another important conclusion that can be drawn from the above finding is that the tapping action of an AFM tip during TM-AFM imaging did not nucleate or induce nanobubbles prior to the ethanol-water exchange. There has been concern that the tapping action of an AFM tip during TM-AFM imaging may somehow induce nucleation of nanobubbles.21 We note that micron-sized bubbles, as observed by optical microscopy (CCD), are formed after ethanol-water exchange on the same substrate without the use of AFM imaging, and also in areas far away from the AFM tip. We also note that the exchange of degassed ethanol and water resulted in the absence of nanobubbles.27,29 Moreover, our observations that the addition of surfactant results in a more deformable yet resilient nanobubble-aqueous phase interface and the relaxation of the pinning of the three-phase line at the perimeter of the nanobubbles are consistent with the behavior that one expects from macroscopic bubbles. Thus, for the remainder of the Discussion section, we assume that the nanobubbles we detected do indeed consist of gas, and they can be induced only under certain circumstances such as ethanolwater exchange. What, then, is causing the formation of nanobubbles during the ethanol-water exchange? We propose a mechanism below. Possible Mechanisms for the Formation of Nanobubbles during Ethanol-Water Exchange and the Implications for Sample Preparation. When water displaces ethanol, it will initially leave a thin layer of ethanol next to the substrate due to the large advancing contact angle of water on a hydrophobic surface and the small receding contact angle of ethanol. However, ethanol and water are miscible, so the ethanol will soon dissolve in the water. Air is an order of magnitude more soluble in ethanol than it is in water,46-48 so this will result in a local supersaturation of gas at the interface. Further, the difference in the solubility of gas between ethanol and water increases with temperature, and the formation of nanobubbles is more efficient when warm ethanol and water are used. We propose that this supersaturation is responsible for both the micron-sized bubbles observed by optical microscopy and the nanobubbles observed by TM-AFM. The implications of our findings for sample preparation procedures in surface science are important. The exchange of ethanol and water happens frequently in a number of experiments 49-51 in the field of surface science. It is necessary to totally dry substrates after ethanol cleaning, before subjecting the substrate to water, to avoid the formation of nanobubbles; that is, even (46) Fischer, K.; Wilken, M. J. Chem. Thermodyn. 2001, 33, 1285 (47) International Critical Tables of numerical data, physics, chemistry and technology; Washburn, E. W., Ed.; McGraw-Hill: New York, 1930. (48) CRC Handbook of Chemistry and Physics, 84th ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 2004. (49) Zhang, X. Y.; Zhu, Y. X.; Granick, S. Science 2002, 295, 663. (50) Bremond, N.; Arora, M.; Ohl, C.-D.; Lohse, D. J. Phys.: Condens. Matter 2005, 17, S3603. (51) Craig, V. S. J.; Neto, C.; Williams, D. R. M. Phys. ReV. Lett. 2001, 8705.

Langmuir, Vol. 22, No. 11, 2006 5033

though ethanol and water are mutually miscible, attempts to rinse ethanol off with water may not result in the same surface as rinsing after drying. We believe that the precise methods used to prepare the sample and introduce solvents, such as solvent exchange, solvent temperature, and solvent gas saturation, are responsible for the wide variety of results reported in the nanobubble literature.8,12-23 Future work in this area should adopt procedures that control these variables. Difference between OTS Silicon and HOPG Substrates. The nanobubbles on OTS silicon are generally larger than those on HOPG, which might be due to the higher hydrophobicity of OTS silicon. The advancing contact angle of water on OTS silicon is ∼108° compared to that on HOPG, which is ∼72°. We note that higher contact angles (∼85°) are often reported for HOPG; however, we found that these were obtained after storing the freshly cleaved samples for some minutes. In addition, the HOPG surface is expected to be more chemically homogeneous compared to the OTS silicon surface because the HOPG surface is cleaved from the pure layered crystal, whereas OTS is chemically deposited on silicon to form an imperfect layer. Indeed, we can occasionally see in AFM images some patches on the OTS silicon substrate. The irregular shape of nanobubbles on OTS silicon shows that the pinning force is crucial to the morphology of nanobubbles. The pinning force is apparently reduced in surfactant solution. We note that the three-phase line of nanobubbles in water on HOPG are always spherical, and there is no evidence of pinning. The CCD images show that micron-sized bubbles are concentrated along the path of the ethanol and water exchange streams. Interestingly, a large density of nanobubbles was observed in the vicinity of the micron-sized bubbles using TMAFM, suggesting a similar formation mechanism for both the nanobubbles and the micron-sized bubbles. A similar effect is expected whenever a flow of water displaces a solvent that is highly soluble in water, has a smaller contact angle than water on the substrate, and has a higher gas solubility than water. We also note that supersaturation of water with air can occur by other means such as shaking, pouring, or an increase in temperature. The Effect of Surfactants on Nanobubble Morphology. Nanobubbles were stable in all three surfactant solutions studied. In the presence of surfactant, the interface becomes more compliant, causing the AFM tip to deform the nanobubble without the tip penetrating the interface (until the tip and the substrate are in hard contact). This was observed in both TM imaging and force measurements. Further, the deformation is only present under load. The original profile is recovered when the nanobubble is uncompressed. This indicates that the gas inside the nanobubble did not dissolve away during this severe deformation. All of the surfactants were found to reduce the pinning of the nanobubble three-phase line, as is observed for larger bubbles on OTS silicon. These findings add credibility to our assumption that the detected nanobubbles do indeed consist of gas, as opposed to an obscure and unknown contaminant. From the TM-AFM height images, we have not detected any significant difference among the three surfactants within the concentration ranges studied. This observation, along with the very different behavior observed for nanobubbles in water, suggests that there is no evidence for the presence of a skin that considerably alters the property of the nanobubble. A contaminant skin would likely consist of partially oxidized surface-active organic material and therefore present a small negative charge. The presence of significant amounts of such material would be expected to lower the surface tension of the nanobubble and

5034 Langmuir, Vol. 22, No. 11, 2006

produce a behavior similar to that observed in the surfactant systems. Additionally, the surfactants employed might be expected to interact with contaminants in a different manner because of the differing headgroup charge. No such difference was observed. The Effect of Salts, Acids, and Bases on Nanobubble Morphology. It has been suggested that gas (air)-water interfaces are negatively charged.52 Then one may expect a change in the stability of nanobubbles upon the addition of electrolytes because they will screen electrostatic forces. In addition, some salts can influence the ease of coalescence of bubbles.53,54 Thus, we ought to be able to detect some difference in the stability or morphology of nanobubbles upon the addition of electrolytes if these are significant factors. Surprisingly, we did not see any significant effect upon the addition of any of the salts studied, nor did we see any measurable effect upon the addition of the two different acids or two different bases. This latter result is in stark contrast to the effect observed on the stability of oil-in-water emulsions stabilized by degassing in which addition of acid readily resulted in the destabilization of oil droplets.55,56 The lack of influence of a range of electrolyte solutions, including acids and bases, is further evidence that the presence of a contaminant skin is not responsible for the high nanoscopic contact angles observed. Our findings, taken together with the apparent lack of difference among nanobubbles in cationic, anionic, and nonionic surfactants, consistently suggest that electrostatic forces may be unimportant in determining the stability or morphology of nanobubbles after they are formed. We note, however, that we have not attempted to exchange ethanol directly with electrolyte solutions and therefore cannot rule out the possibility that these factors may influence the formation of nanobubbles. We will leave these themes for future study. The Large Discrepancy between the Microscopic and Macroscopic Contact Angles. The morphology of nanobubbles is very important because it is closely related to the stability of nanobubbles, which is still mysterious. We used various methods to confirm the value of the nanoscopic contact angle formed by nanobubbles. As the AFM technique is invasive, one has to be careful in employing it to determine the nanoscopic contact angle. The three different modes of AFM (TM height imaging, force measurement, and lift-mode imaging) yield somewhat different nanobubble heights. In addition, for a given nanobubble height, our procedure to fit a spherical cap to the measured profiles of nanobubbles usually leaves about 2-6 nm of uncertainty near the base of the spherical cap. Still, in all cases, the base radius of nanobubbles is 5-20 times larger than the height (i.e., the nanobubble has a very flat profile). Thus the large difference between macroscopic and nanoscopic contact angles cannot be discounted as an artifact of the measurement techniques. Our findings are in agreement with those reported by other authors.16,18 (Note that they defied convention and reported the angle through the less dense air phase, and therefore the value reported needs to be subtracted from 180° for comparison with the values we report.) One possibility that has to be addressed is that the nanobubbles are in a quasi-equilibrium state, and therefore the Young equation should apply. Thus, the difference in contact angle observed is due to either the presence of contaminants on the surface or an additional term in the Young equation that applies when the (52) Marinova, K. B.; Alargova, R. G.; Denkov, N. D.; Velev, O. D.; Petsev, D. N.; Ivanov, I. B.; Borwankar, R. P. Langmuir 1996, 12, 2045. (53) Craig, V. S. J.; Ninham, B. W.; Pashley, R. M. J. Phys. Chem. 1993, 97, 10192. (54) Ninham, B. W.; Yaminsky, V. V. Langmuir 1997, 13, 2097. (55) Pashley, R. M. J. Phys. Chem. 2003, 107, 1714. (56) Maeda, N.; Rosenberg, K.; Israelachvili, J.; Pashley, R. Langmuir 2004, 20, 3129.

Zhang et al.

curvature is significant. However, inspection of the Young equation suggests that contamination is unlikely to be responsible for the observed change in contact angle.

γsv ) γsl + γlv cos θ

(4)

Such a change could result from a reduction in γsv, an increase in γsl, or a decrease in γlv. A reduction in γsv is unlikely because this interfacial energy is already low for a hydrophobic surface. A large increase in γsl is also unlikely because most contaminants will result in a lowering of γsl. A decrease in γlv will result with the adsorption of contaminants; however, we have shown that the addition of surfactants that greatly reduce γlv has not resulted in a large change in contact angle. (We note that the three different surfactants are expected to adsorb to both the solid-liquid interface and the solid-vapor interface and that their influence on γsl - γsv should be comparatively small.) Given that it is unlikely that the observed contact angles are due to contamination, we now consider the second scenario. The Young equation is often modified to account for the influence of line tension.57 Indeed, Yang et al. proposed that the line tension is responsible for the difference between nanoscopic and macroscopic contact angles.18 We analyzed our data presented in Figure 11c in the same manner (not shown) and found that the uncertainty in the contact angle measurement precludes any sensible determination of line tension. Similarly, the conclusions of Yang et al. are only sustainable if the contact angle can be measured with a very high degree of accuracy (better than (2°). We do not think this is possible when using the TM-AFM imaging technique, particularly when the tip convolution has not been addressed. We note that one effect of the very large contact angle is a large increase in the radius of curvature and an accompanying decrease in Laplace pressure, and speculate that the decrease in the internal pressure of the bubble may be associated with the change in contact angle. However, we do not see a significant change in contact angle as a function of nanobubble curvature (see Figure 11c) as might be expected if this were the case. Resolution of the cause of the difference in contact angle for nanoscopic and macroscopic bubbles remains to be elucidated. We now consider the scenario that nanobubbles only exist in a nonequilibrium state. Possible Mechanisms for the Long-Term Stability of Nanobubbles. As we noted at the beginning of the Discussion section, the presence of any form of air layer (including nanobubbles) between a hydrophobic substrate and the aqueous phase is thermodynamically unfavorable, yet the nanobubbles we observe are stable for hours. We therefore look to kinetics to explain the apparent stability of nanobubbles. While the bubbles are truly nanoscopic in terms of height, they are highly flattened (large contact angle) and hence have a radius of curvature much greater than both the height and the lateral dimension. For example, for a typical bubble with a height of 50 nm, the radius of curvature may be 1000 nm. This has considerable implications for their stability. If the radius of curvature is much larger, then the Laplace pressure, which drives the dissolution of nanobubbles, is considerably smaller, and this will result in a lifetime prediction orders of magnitude longer. However, analysis suggests that the predicted bubble lifetime is still considerably shorter than we observe.11 Is it possible that the negative charge of the gas-water interface somehow accounts for this paradox? We note that it is unlikely (57) Amirfazli, A.; Neumann, A. W. AdV. Colloid Interface Sci. 2004, 110, 121.

Nanobubbles on Hydrophobic Surfaces in Water

that the substrate surface is charged at all in air (inside nanobubbles). Regardless, our results show that the nanobubbles are insensitive to electrolytes and surfactants of different charge, which is in agreement with the notion that the electrostatic forces do not play an important role here. Second, the van der Waals force across the air layer (or nanobubbles) is expected to be attractive and cannot account for the apparent stability. Nonetheless, it is still conceivable that the nanobubbles persist because either (1) they are trapped in a local minimum of the free energy landscape, or (2) the time scale that characterizes the disappearance of nanobubbles is much longer than the experimental time scale and that predicted theoretically.

Conclusions We have reproduced the earlier findings reported by Hu and co-workers.24,27 In particular, we used OTS silicon as the hydrophobic substrate in combination with the ethanol-water exchange technique for the first time and obtained results consistent with those on HOPG. We have indicated that variations reported in published results may arise as a consequence of different sample preparation procedures. We also extensively studied the effects of imaging conditions to estimate the real profile of the nanobubbles and quantify the contact angle. We found that nanobubbles are most likely deformed by the AFM tip during imaging. In addition, we studied the effects of different types of salts and the solution pH and found that they have only a minimal influence on nanobubble morphology. The addition

Langmuir, Vol. 22, No. 11, 2006 5035

of surfactant increases the compliance of the interface, as would be expected from the accompanying decrease in surface tension. Physical properties of nanobubbles are found to be comparable to those of macroscopic bubbles in all respects other than the contact angle, which is surprisingly large for nanobubbles. The contact angle anomaly remains unresolved and is perhaps associated with the line tension or the surprising stability observed for nanobubbles through the commensurate reduction in Laplace pressure. Acknowledgment. X.H.Z. gratefully acknowledges Australian Endeavor Postdoctoral Research Fellowship. N.M. gratefully acknowledges CRC Smart Print for financial support. V.S.J.C gratefully acknowledges the support of the ARC through a Research Fellowship. We thank Chiara Neto for assistance with sample preparation and Mika Kohonen for macroscopic contact measurements. Tim Senden, Drew R. Evans, and Esben Thormann are acknowledged for assistance in AFM. Supporting Information Available: CCD image of micronsized bubbles on an OTS silicon surface undergoing AFM imaging after ethanol-water exchange; TM-AFM images of nanobubbles scanned using different set points; lift-mode images of nanobubbles in ∼0.5 CMC Tween 20 solution; and details from the profiles of nanobubbles in water imaged at different set point amplitude ratios. This material is available free of charge via the Internet at http://pubs.acs.org. LA0601814