Nanobubbles at the Interface between Water and a Hydrophobic Solid

Mar 27, 2008 - ... Center, University of Melbourne, Melbourne 3010, Australia .... Dongjin Seo , Sean R. German , Tony L. Mega , and William A. Ducker...
0 downloads 0 Views 549KB Size
4756

Langmuir 2008, 24, 4756-4764

Nanobubbles at the Interface between Water and a Hydrophobic Solid Xue Hua Zhang, Anthony Quinn, and William A. Ducker* Department of Chemical and Biomolecular Engineering and Particulate Fluid Processing Center, UniVersity of Melbourne, Melbourne 3010, Australia ReceiVed NoVember 23, 2007. In Final Form: January 17, 2008 A very thin layer (5-80 nm) of gas phase, consisting of discrete bubbles with only about 40 000 molecules, is quite stable at the interface between a hydrophobic solid and water. We prepare this gas phase from either ambient air or from CO2(g) through a solvent exchange method reported previously. In this work, we examine the interface using attenuated total internal reflection infrared spectroscopy. The presence of rotational fine structure in the spectrum of CO2 and D2O proves that molecules are present in the gas phase at the interface. The air bubbles are stable for more than 4 days, whereas the CO2 bubbles are only stable for 1-2 h. We determine the average gas pressure inside the CO2 bubbles from the IR spectrum in two ways: from the width of the rotational fine structure (Pgas < 2 atm) and from the intensity in the IR spectrum (Pgas ) 1.1 ( 0.4 atm). The small difference in gas pressure between the bubbles and the ambient (1 atm) is consistent with the long lifetime. The dimensions and curvature of a set of individual bubbles was determined by atomic force microscopy. The pressures of individual bubbles calculated from the measured curvature using the Laplace equation fall into the range Pgas ) 1.0-1.7 atm, which is concordant with the average pressure measured from the IR spectrum. We believe that the difference in stability of the CO2 bubbles and the air bubbles is due to a combination of the much lower pressure of CO2 in the atmosphere and the greater solubility of CO2 in water, compared to N2 and O2. As expected, smaller bubbles have a shorter average lifetime than larger bubbles, and the average pressure and the curvature of individual bubbles decreases with time. Surface plasmon resonance measurements provide supporting evidence that the film is in the gas state: the thin film has a lower refractive index than water, and there are few common contaminants that satisfy this condition. Interfacial gas bubbles are not ubiquitous on hydrophobic solids: bubble-free and bubble-decorated hydrophobic interfaces can be routinely prepared.

Introduction The response of liquid water when faced with a large hydrophobic solid (e.g., solid alkane) is the subject of active research.1,2 Some conclude that there is a reduced density of water in the thin layer adjacent to hydrophobic solids,3-6 whereas others conclude that there is no evidence for a reduced density or a gas phase.7-10 Recent compelling evidence from X-ray reflectivity measurements show that there is a very thin (0.20.4 nm) layer with a reduced density of water adjacent to solids that have been coated with a hydrophobic monolayer.11,12 There is a parallel controversy over the possible existence of very small, discrete bubbles at the interface between a hydrophobic solid and water.1,13 If interfacial bubbles were to exist under ambient conditions, they should be thermodynamically un* Address correspondence to this author. E-mail: [email protected] (1) Ball, P. Nature 2003, 423 (6935), 25-26. (2) Christenson, H. K.; Claesson, P. M. AdV. Colloid Interface Sci. 2001, 91 (3), 391-436. (3) Steitz, R.; Gutberlet, T.; Hauss, T.; Klosgen, B.; Krastev, R.; Schemmel, S.; Simonsen, A. C.; Findenegg, G. H. Langmuir 2003, 19 (6), 2409-2418. (4) Schwendel, D.; Hayashi, T.; Dahint, R.; Pertsin, A.; Grunze, M.; Steitz, R.; Schreiber, F. Langmuir 2003, 19 (6), 2284-2293. (5) Subramanian, S.; Sampath, S. J. Colloid Interface Sci. 2007, 313 (1), 6471. (6) Maccarini, M.; Steitz, R.; Himmelhaus, M.; Fick, J.; Tatur, S.; Wolff, M.; Grunze, M.; Janecek, J.; Netz, R. R. Langmuir 2007, 23 (2), 598-608. (7) Doshi, D. A.; Watkins, E. B.; Israelachvili, J. N.; Majewski, J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (27), 9458-9462. (8) Seo, Y. S.; Satija, S. Langmuir 2006, 22 (17), 7113-7116. (9) Mao, M.; Zhang, J. H.; Yoon, R. H.; Ducker, W. A. Langmuir 2004, 20 (5), 1843-1849. (10) Takata, Y.; Cho, J. H. J.; Law, B. M.; Aratono, M. Langmuir 2006, 22 (4), 1715-1721. (11) Mezger, M.; Reichert, H.; Schoder, S.; Okasinski, J.; Schroder, H.; Dosch, H.; Palms, D.; Ralston, J.; Honkimaki, V. Proc. Natl. Acad. Sci. U.S.A. 2006, 103 (49), 18401-18404. (12) Poynor, A.; Hong, L.; Robinson, I. K.; Granick, S.; Zhang, Z.; Fenter, P. A. Phys. ReV. Lett. 2006, 97 (26). (13) Attard, P. AdV. Colloid Interface Sci. 2003, 104, 75-91.

stable.14,15 The interfacial tension of pure water against air (72 mJ m-2) is greater than against common hydrophobic solids, such as plastics and alkanes (∼50 mJ m-2), so there is no energetic gain from covering a flat hydrophobic solid with a thick layer of air, whether uniform or discrete. Additional energy is required to create the solid-gas interface. Thus, energy must be supplied to create an interfacial bubble. In addition, the curvature required to close a discrete bubble on a flat solid requires a greater pressure inside the bubble than outside. If this pressure is supplied by the gas, then there is also a problem with the kinetic stability of the bubble. (See, for example, refs 16 and 17, which describe the stability of bubbles in bulk solution.) In spite of these objections, experimental evidence for the existence of discrete interfacial nanobubbles is accumulating. Parker et al. originally invoked their existence from surface forces measurements in 1994.18 In 2000, Hagashitani reported AFM images of interfacial bubbles,19 but this result remained controversial, because it was reproduced in some laboratories20-26 (14) Eriksson, J. C.; Ljunggren, S. Colloids Surf. A 1999, 159 (1), 159-163. (15) Eriksson, J. C.; Ljunggren, S. Langmuir 1995, 11 (6), 2325-2328. (16) Epstein, P. S.; Plesset, M. S. J. Chem. Phys. 1950, 18 (11), 1505-1509. (17) Ljunggren, S.; Eriksson, J. C. Colloids Surf. A 1997, 130, 151-155. (18) Parker, J. L.; Claesson, P. M.; Attard, P. J. Phys. Chem. 1994, 98 (34), 8468-8480. (19) Ishida, N.; Inoue, T.; Miyahara, M.; Higashitani, K. Langmuir 2000, 16 (16), 6377-6380. (20) Holmberg, M.; Kuhle, A.; Garnaes, J.; Morch, K.; Boisen, A. Langmuir 2003, 19 (25), 10510-10513. (21) Simonsen, A.; Hansen, P.; Klosgen, B. J. Colloid Interface Sci. 2004, 273 (1), 291-299. (22) Otsuka, I.; Yaoita, M.; Higano, M.; Nagashima, S. Surf. ReV. Lett. 2003, 10 (2-3), 337-343. (23) Tsionsky, V.; Kaverin, A.; Daikhin, L.; Katz, G.; Gileadi, E. Phys. Chem. Chem. Phys. 2005, 7 (8), 1830-1835. (24) Yang, J. W.; Duan, J. M.; Fornasiero, D.; Ralston, J. J. Phys. Chem. B 2003, 107 (25), 6139-6147. (25) Borkent, B. M.; Dammer, S. M.; Schonherr, H.; Vancso, G. J.; Lohse, D. Phys. ReV. Lett. 2007, 98 (20).

10.1021/la703475q CCC: $40.75 © 2008 American Chemical Society Published on Web 03/27/2008

Nanobubbles at the H2O-Hydrophobic Solid Interface

but not others.9,10,27,28 At that point it was not clear whether the features observed by AFM were simply very rare or artefacts. Interfacial nanobubbles are not expected to arise spontaneously, so a more fruitful line of enquiry has developed around rational methods to cause the formation of interfacial bubbles by creating supersaturation of gases.29,30 AFM has now been used to image features that were formed at the interface between water and hydrophobic solids after the water was supersaturated with gas by a number of techniques, including solvent exchange,29-35 temperature changes,35,36 and electrochemical or chemical reactions.37 These experiments have been performed on a variety of solids, both hydrophobic (HOPG and OTS-Si) and hydrophilic (mica). All the experiments described in the previous paragraph used AFM imaging to identify the interfacial nanobubbles. Although AFM has the requisite resolution to confirm the existence of nanoscopic features, AFM images are not good at identifying the phase state of an object. Much of the interest in nanobubble work arises from the possibility that they are actually in the gas state, rather than a liquid or a solid. Indirect evidence for a gas state has been provided by the fact that the features in the AFM images can be removed by degassing the solutions31 and that they do not form when degassed solvents are used.30,34 Supporting evidence for the existence of nanobubbles comes from the fact that nanobubbles form under same conditions as micron and larger-sized bubbles.33 Clearly, the situation would be improved if the identity and phase of these interfacial fluids could be confirmed by spectroscopy. In this paper, we describe direct evidence for the existence of interfacial nanobubbles using attenuated total internal reflection infrared spectroscopy (ATR-IR). In ATR-IR, the electrical field decays exponentially from the interface (decay length ∼200 nm) so the spectrum is sensitive to adsorbed material. We also use ATR-IR to measure the pressure inside the gas. This is a more complete account of work that was briefly described in a recent letter.38 The key element is the formation of interfacial nanobubbles containing CO2, because the phase state of CO2 can be uniquely identified from the infrared spectrum.39 The gaseous CO2 spectra have characteristic fine structure due to the rotational freedom of gas molecules, while CO2 dissolved in water has only has a single sharp absorbance peak around 2345 cm-1. We extend the initial report with both an additional measurement of the gas pressure from line broadening and a study of the kinetics of the bubble dissolution. (26) Jeon, S. M.; Desikan, R.; Tian, F.; Thundat, T. Appl. Phys. Lett. 2006, 88 (10). (27) Evans, D. R.; Craig, V. S. J.; Senden, T. J. Physica A 2004, 339 (1-2), 101-105. (28) McKee, C. T.; Ducker, W. A. Langmuir 2005, 21 (26), 12153-12159. (29) Lou, S. T.; Ouyang, Z. Q.; Zhang, Y.; Li, X. J.; Hu, J.; Li, M. Q.; Yang, F. J. J. Vacuum Sci. Technol. B 2000, 18 (5), 2573-2575. (30) Zhang, X. H.; Zhang, X. D.; Lou, S. T.; Zhang, Z. X.; Sun, J. L.; Hu, J. Langmuir 2004, 20 (9), 3813-3815. (31) Zhang, X. H.; Li, G.; Maeda, N.; Hu, J. Langmuir 2006, 22 (22), 92389243. (32) Zhang, X. H.; Li, G.; Wu, Z. H.; Zhang, X. D.; Hu, J. Chin. Phys. 2005, 14 (9), 1774-1778. (33) Zhang, X. H.; Maeda, N.; Craig, V. S. J. Langmuir 2006, 22 (11), 50255035. (34) Zhang, X. H.; Wu, Z. H.; Zhang, X. D.; Li, G.; Hu, J. Int. J. Nanosci. 2005, 4 (3), 399-407. (35) Zhang, X. H.; Zhang, X.; Sun, J.; Zhang, Z.; Li, G.; Fang, H.; Xiao, X.; Zeng, X.; Hu, J. Langmuir 2007, 23 (4), 1778-1783. (36) Yang, S. J.; Dammer, S. M.; Bremond, N.; Zandvliet, H. J. W.; Kooij, E. S.; Lohse, D. Langmuir 2007, 23 (13), 7072-7077. (37) Zhang, L. J.; Zhang, Y.; Zhang, X. H.; Li, Z. X.; Shen, G. X.; Ye, M.; Fan, C. H.; Fang, H. P.; Hu, H. Langmuir 2006, 22 (19), 8109-8113. (38) Zhang, X. H.; Khan, A.; Ducker, W. A. Phys. ReV. Lett. 2007, 98 (13), 136101. (39) Gong, W. Q.; Stearnes, J.; Fornasiero, D.; Hayes, R. A.; Ralston, J. Phys. Chem. Chem. Phys. 1999, 1 (11), 2799-2803.

Langmuir, Vol. 24, No. 9, 2008 4757

Figure 1. Experimental protocol. Bubbles are formed when the fluid surrounding the hydrophobic solid is switched from ethanol (stage 1) to water (stage 2). The previous stages (in air and stage 0) are for control experiments and to allow recording of the background IR spectra.

In all our experiments, the interfacial nanobubbles are generated by starting with a hydrophobic solid in one solvent and exchanging that solvent with a second solvent29 (Figure 1, stages 1 and 2). The important characteristic is the relative solubility of gas in the two solvents.29,33 The first solvent is ethanol and the second solvent is water. N2, O2, and CO2 each have low solubility in both water and ethanol, but the solubility in ethanol is about 10-20 times higher than in water at 1 atm.40 (Water at equilibrium with air has about 1 mM of air molecules.41 The solubility of CO2 is about 20 times higher than N2 in water.) The hypothesis is that, during the exchange of solvents, gas that was originally dissolved in the ethanol becomes supersaturated in the water and “precipitates” on the solid surface.29,30,33,35,38 Prior to generating the nanobubbles, we first performed the measurements on the solid in air and then in water (Figure 1, stage 0). We do this step to examine the following two hypotheses: (A) that nanobubbles are always present at the interface between water and a hydrophobic solid and (B) that nanobubbles form when the solvent quality (i.e., solubility for the gas) is improved. We find both of these hypotheses to be false. Stage 0 also provides background spectra for later stages. Experimental Methods Preparation of Solutions. The CO2-saturated solutions were prepared by bubbling CO2(g) through ethanol or H2O or H2O/D2O mixtures for about 2 h. The final pH of CO2-saturated water was 3.8 ( 0.2. Note that in H2O/D2O mixtures, H+ and D+ exchange rapidly between molecules, so DOH was also present. Normal Milli-Q water and ethanol (100%, Anala R, Merck Pty Ltd) were used as the air-equilibrated solutions. Hydrophobic Solids. Our model hydrophobic material was silicon wafers (Mitsubishi Silicon, America-Mod 2 polished wafers) coated with a self-assembled monolayer of octadecyltrimethylchlorosilane (OTS).35 In brief, polished silicon wafers (Mitsubishi Silicon, America-Mod 2) were cleaned in piranha solution [H2SO4 (70%): H2O2 (30%)] and all the glassware used for the solution was dried (40) Gerrard, W. Gas Solubilities, 1st ed.; Pergamon Press: 1980; p 365. (41) Maeda, N.; Rosenberg, K.; Israelachvili, J.; Pashley, R. Langmuir 2004, 20 (8), 3129-3137.

4758 Langmuir, Vol. 24, No. 9, 2008

Zhang et al.

Figure 2. Tapping-mode AFM images of the surface of OTS-coated silicon/water (A) in air; (B) stage 0, in water before ethanol; (C) stage 1, in ethanol; (D) stage 2, in water after exchange. All solutions were saturated with CO2(g). The cross section (E) shows measured points through the bubble apex (shown by the dotted line in D) and the best fit of an arc of a circle (solid line, radius:6.2 µm) to the cross section. for 2 h at 120 °C in the oven. The silicon was dried at 120 °C for 1.5 h, and soaked in 0.5 vol % OTS in toluene for ∼12 h. The surface was then rinsed with chloroform, sonicated in toluene and ethanol, and dried with nitrogen. Before use, the OTS-Si was sonicated in chloroform, ethanol, and water sequentially and then dried with nitrogen. Water droplets in air on OTS-Si form an advancing contact angle of 112 ( 3° and a receding angle of 101 ( 3°. The rms roughness was 0.2 nm over an area of 6 µm2 in water (Figure 2B). The silicon wafer used for IR measurements was cut into parallelepiped prisms (80 mm × 8 mm × 3.2 mm) by Rohm and Haas (Blacksburg, VA) before coating with OTS. Surface plasmon resonance (SPR) experiments were performed on a gold film coated with a layer of 1-decanethiol. The gold was sputter-coated onto glass with an adhesion layer of chromium and then exposed to a 1 mM solution of 1-decanethiol in ethanol for 2 h. The surface was then cleaned in ethanol in an ultrasonic bath. The advancing angle of water on the thiol-coated gold surface was 104°, and receding angle was 93°. Solvent Exchange Procedure. All experiments were performed at 25-27 °C unless otherwise specified. This temperature was chosen so that the gas supersaturation was not too high and the formation of macron-sized bubbles could be avoided. The solvent was exchanged three times in each experiment as follows. Stage 0: We immersed an OTS-Si wafer in water. This acted as a control for stage 3; it was not required to generate bubbles. Stage 1: we replaced the water with ethanol. Stage 2: we replaced the ethanol with water. In air-equilibrated experiments, all solvents were equilibrated with air; in CO2-saturated experiments, all solvents were saturated with CO2 as described above. Infrared Spectroscopy. IR spectra were recorded in Attenuated total reflection (ATR) configuration on a Nicolet 5700 FTIR (Thermoelectron Corp.) at 25-27 °C with a liquid nitrogen cooled MCT detector, incident angles in the range 56°-60°, between 46 and 52 reflections, and a spectrum resolution of 1 cm-1, except in Figures 4B, 7, and 11, where the resolution was 0.125 cm-1. The decay length of the evanescent field created by total internal reflection is about 300 nm at 2000 cm-1 in water. The absorption coefficient of CO2(g) was measured in transmission experiments on the same instrument at the resolution of 1 cm-1. Ambient CO2 and H2O were prevented from entering the light path in the sample chamber by maintaining a purge of either high-purity nitrogen or air from which CO2 and water were removed, except in Figures 5B and 10. Atomic Force Microscopy. Tapping-mode mode AFM was used to image the solid-liquid interface in a closed fluid cell, as described

previously,29-31,33,35 using either a MultiMode Nanoscope IIIa SPM (Digital Instruments Veeco Metrology Group, Woodbury, NY) or MFP-3D Atomic Force Microscope (Asylum Research, Santa Barbara, CA). Nominally, 0.32 N/m cantilevers (NP probe, Veeco) were treated by UV in air for 20 min before use to remove organic contamination. The drive frequency for the TM-AFM imaging was typically 6-12 kHz in water. The set point ratio (A/A0, where A0 is the free amplitude of the cantilever and A is the set point amplitude) was usually above 0.8 during imaging. Surface Plasmon Resonance. A laser beam (631.6 nm, ppolarized) was coupled into a prism (SF10 glass, Schott) in the Kretschmann configuration. A surface plasmon spectroscope (Optrel GBR, Berlin, Germany) equipped with a homemade fluid cell was used.

Results Interfacial Bubbles Are Not Always Present on Hydrophobic Solids. After a smooth OTS-Si wafer has been transferred from air to water, AFM images do not reveal interfacial nanobubbles (Figure 2). Likewise, when the water is exchanged for ethanol AFM images do not reveal interfacial nanobubbles. This applies both to solvents that have been equilibrated in air and those that are saturated with CO2. There is also no evidence of CO2(g) in ATR-IR spectra measured at the water/OTS-Si interface. Thus, we have shown that nanobubbles are, at best, extremely rare when smooth OTS-Si wafers are immersed in water or in ethanol. Interfacial Bubbles Form When the Solvent Quality Is Rapidly Decreased. When we change the solvents bathing the OTS-Si from ethanol to water (both CO2-saturated), AFM images show a set of small lumps that have previously been assigned as nanobubbles (Figure 2). These interfacial nanobubbles have a cross section that is very similar to an arc of a circle. The heights are in the range of 10-80 nm with a typical value of 20 nm, the diameters of the three phase circle are in the range of 0.6-6 µm, and the fitted curvature of the “gas”-liquid interface is in the range of 2-45 µm. These features are formed for both CO2-saturated solutions and for air-equilibrated solutions as described previously.33 ATR-FTIR spectra of the surface of an OTS-silicon prism are shown in Figures 3 and 4. All IR spectra shown are the

Nanobubbles at the H2O-Hydrophobic Solid Interface

Figure 3. ATR-FTIR spectra of the surface of an OTS-silicon prism after CO2-saturated water has displaced ethanol (stage 2). The background is CO2-saturated water before ethanol (stage 0). The spectrum shows four features: negative water bands (>2800 cm-1, 2000-2300 cm-1) due to the exclusion of water from the interface, positive CO2(g) bands (2300-2380 cm-1), the absence of ethanol bands (2980 cm-1), and very little change to the C-H stretch region (∼2900 cm-1).

Figure 4. ATR-FTIR spectra of the interface between the OTSsilicon prism and various fluids. (A). CO2 gas (background is argon) at a spectral resolution of 1 cm-1. The gas exhibits rotational fine structure. CO2-saturated water, stage 0. (Background is airequilibrated water). CO2-saturated water, stage 2. (Background is stage 0). The spectrum in stage 2 shows the rotational fine structure, proving that there is CO2 gas at the interface. The negative CO2(aq) peak shows that the aqueous phase is excluded from part of the interface. (B) High resolution (0.125 cm-1) spectrum in stage 2.

difference between spectra measured in two different conditions; this method of presentation removes the features that are common in both spectra. For example, the strong adsorption by water is mainly removed from the spectrum by subtraction of a background spectrum of water or an aqueous solution. The difference between the spectra in stage 2 and stage 0 in Figure 3 clearly shows four features as follows: (1) A set of positive gaseous CO2 peaks is shown; there is now gas at the interface. (2) A negative aqueous CO2 peak and a negative water band are visible. The interfacial gas has excluded water from part of the evanescent zone (∼200 nm deep), resulting in less IR absorption from both water and CO2 that is dissolved in the water [which is mainly CO2(aq)].42 The exclusion of water from the interface is consistent with the formation of bubbles at the interface. (3) No absorption of ethanol (2980 cm-1) is seen; the nanofeatures in the AFM image are not ethanol. (4) There is very little change to the C-H stretch region (∼2900 cm-1); there is no significant organic contamination. The expanded frequency scale in Figure 4 clearly shows that the CO2 at the interface has the spectrum of CO2 gas. In fact the (42) Stumm, W.; Morgan, J. J. Aquatic Chemistry, 3rd ed.; John Wiley & Sons, Inc.: New York, 1996; pp 150-163.

Langmuir, Vol. 24, No. 9, 2008 4759

CO2 band in stage 2 is the sum of CO2(g) and a negative CO2(aq) band (2340 cm-1) due to CO2(aq) being excluded from the interface. A simple explanation for the presence of the CO2(g) spectrum would be macroscopic bubbles in the IR cell in stage 2. To check for macroscopic bubbles, we have replaced the OTScoated prism with an OTS-coated glass slide so that we can view the interface in the same fluid cell with optical microscopy (resolution ∼1.5 µm). When we repeat stage 0, 1, and 2 we do not observe bubbles. Note that in reference 38 we incorrectly stated that we monitored the silicon prism whereas we monitored a glass slide. So in summary, when we change from a better solvent (ethanol) to a poorer solvent (water) for gases, CO2(g) spectra appear at the interface; there are no macroscopic bubbles and there are microscopic features in the AFM image. Therefore, we have formed nanobubbles of CO2 at the interface. This one example showing the existence of nanobubbles negates any generic arguments against the existence of interfacial nanobubbles. Water vapor also exhibits rotational fine structure in the IR spectrum,43,44 and the fine structure is visible in the O-H bands. However, we are not able to make definite conclusions that the bubbles contain some water vapor from this observation, because (a) the water absorbs so much of the IR that the transmittance is very weak and (b) it is difficult to be certain that all water has been completely removed from the IR beam outside the fluid cell and this amount can fluctuate. So, instead of relying on H2O spectrum, we doped the water with D2O and examined the O-D band. Figure 5A is the IR spectrum in stage 2 when the experiment is repeated with a CO2-saturated 5 vol % D2O solution. There is a large negative O-D band (∼2500 cm-1), due to exclusion of DOH/D2O from the interface; a negative aqueous CO2 band (∼2343 cm-1); and a positive gaseous CO2 band. The gaseous CO2 band proves the existence of the gaseous phase, and the negative aqueous CO2 and O-D bands demonstrate that the aqueous phase is excluded from the surface; i.e., they are consistent with the presence of a gas phase. There is not enough D2O to observe the gas fine structure, but later we use this spectrum to calculate the pressure of CO2 in the interfacial bubbles. When 30 vol % D2O solution is used, the O-D band in stage 2 (Figure 5B,C) clearly shows rotational fine structure. This fine structure with the same peak positions as in gas-phase O-D superimposed on the negative liquid-phase O-D band demonstrates that the interfacial air bubbles contain water vapor. A negative O-H band is also present. It is reasonable to find gaseous phase water because the bubble is in direct contact with the surrounding water. The vapor pressure of water (either H2O or D2O) at room temperature is about 0.03 atm,45 so water should make only a small contribution to the total gas pressure in the bubble. Further evidence for the existence of interfacial air bubbles was obtained from surface plasmon resonance (SPR) experiments. The angle of resonance in the SPR curve depends on the refractive index of thin films at the interface and the thickness of the film. This technique is usually used to detect the binding of organic molecules to interfaces. The refractive index of common organic contaminants is greater than that of water, so their adsorption causes the resonance to shift to a higher angle. When we compare the resonant angle in air-equilibrated water in stage 0 and stage 2, we see that, after exposure to ethanol, the resonant is shifted to a lower angle (Figure 6), which demonstrates the displacement of water (index 1.33) by a lower index material. This is consistent (43) Gailar, N.; Plyler, E. K. J. Chem. Phys. 1956, 24 (6), 1139-1165. (44) Marechal, Y. J. Chem. Phys. 1991, 95 (8), 5565-5573. (45) Harvey, A. H.; Lemmon, E. W. J. Phys. Chem. Ref. Data 2002, 31 (1), 173-181.

4760 Langmuir, Vol. 24, No. 9, 2008

Zhang et al.

Figure 5. ATR-FTIR spectra of the interface between OTS-silicon and H2O/D2O solution for stage 2 with the spectrum for stage 0 as the background. (A) 5 vol % D2O solution saturated with CO2. (B, C) 30 vol % D2O solution equilibrated with air.

Figure 7. ATR-IR spectrum on rough OTS-silicon substrate. Vaporphase CO2 peaks are observed for stage 0. The background for the spectrum is air-equilibrated water. Resolution: 0.125 cm-1.

Figure 6. SPR at the interface between 1-decanethiol-coated gold film and air-equilibrated water for stage 0 (black line) and stage 2 (red line). (A) SPR curves in large scale. (B) The critical angles in the two curves are the same, which indicates that the refractive index of the bulk medium is the same after the exchange. (C) The resonance angle always shifts to a lower angle after the exchange, which demonstrates that a thin layer with refractive index lower than bulk water has been formed in stage 2. The extent of resonance angle shift is not constant from experiment to experiment, suggesting variation in the coverage or size of the bubbles.

with the adsorption of a gaseous phase (refractive index ∼ 1.00). A quantitative estimate of the amount of air is precluded by distribution of gas in a heterogeneous set of discrete bubbles. Meanwhile, the critical angle (∼44.4°) in stage 0 and stage 2 is the same, which indicates that the refractive index of the bulk liquid is similar, i.e., ethanol has been removed from the system. Interfacial Bubbles Are More Easily Formed on Rough Solids. On smooth solids, we do not observe interfacial gaseous phase from the IR spectrum until we have exchanged a better solvent for a poorer solvent. On a rough solid, we sometimes observe a gaseous phase before exchanging solvents (stage 0, Figure 7). Possible mechanisms for the formation of these bubbles include the following: (a) during immersion, there are some concave surface features in which it is locally unfavorable for the water to penetrate and (b) lower curvature structures can be formed on the rough surfaces, requiring only small fluctuations in CO2 density.

Figure 8. AFM height images of interfacial nanobubbles over the same area 20 min and 44 h after formation. The larger bubbles are very stable, but the several smaller ones have decreased in volume or dissolved.

Interfacial Bubbles Have Long Lifetimes. AFM images show that the air bubbles are stable for days. Figure 8 shows two pairs of tapping-mode AFM images of the same area taken 44 h apart. All the large bubbles remain after 44 h, but some of the very small bubbles dissolve after about 2 days. After 44 h the water was replaced with 20% ethanol aqueous solution to increase the CO2 solubility; we still did not observe a change in the bubbles over the next 20 h (data not shown). Figure 9 shows a time series over 4 days at different ambient temperature. Over 4 days, images

Nanobubbles at the H2O-Hydrophobic Solid Interface

Langmuir, Vol. 24, No. 9, 2008 4761

Figure 9. Stability of nanobubbles with time at different ambient temperature, where the temperature of ethanol and water is 30 °C: (A) 20 min at 21 °C, (B) 2 days at 21 °C, (C) an additional day at 23 °C, (D) a second additional day at 25 °C (the start of day 5). Some small bubbles have dissolved on the third day.

Figure 10. ATR-IR spectrum after the exchange of ethanol by water (15% D2O in H2O). The negative D2O peak is due to the depletion of liquid by interfacial nanobubbles. There is no obvious change of the D2O absorbance with time. (There is no vapor purge in this experiment, so CO2 is uncontrolled.)

Figure 11. ATR-IR spectra after the exchange of ethanol for D2O: H2O (5:95 in volume) over time. The background is stage 0 with the same solution. The interval between spectra is 17-18 min. All solutions were initially saturated CO2.

of the same area look the same. Close inspection (Figure 9, elliptical label) reveals that some very small bubbles disappeared on the third day. AFM images of the air-bubble lifetimes are supported by monitoring the IR spectrum from D-O bands in experiments that are doped with D2O (Figure 10). In stage 2 we observe negative D-O bands that we attribute to exclusion of water from the evanescent zone (∼200 nm of water nearest the solid surface). Over a period of 3 days, the D-O bands have approximately constant intensity, i.e., there is a constant zone at the interface that is inaccessible to liquid water. So both the AFM images and the IR experiments show that the interfacial air bubbles last for at least days. Interfacial CO2 bubbles are much less stable than air bubbles. The time series of IR spectra in Figure 11 shows that the CO2(g) band decreases rapidly and is not resolved 2 h after the fluids

were exchanged to form the bubbles. Likewise, the negative D2O peak decreases rapidly, showing the loss of bubble volume. The explanation for the difference between the stability of air and CO2 nanobubbles is revealed by the negative CO2(aq) band in the time series shown in Figure 11. The amount of CO2 in solution falls rapidly; clearly our IR fluid cell was not gastight. At the start of the experiment, the aqueous CO2 was in equilibrium with about 1 atm of CO2 gas, whereas later, during the IR experiment, the exterior of the cell contained air with most of both the CO2 and water removed. Our explanation for the rapid loss of the interfacial CO2 nanobubbles is that, as the gas leaked out of the IR cell, a gradient in CO2 chemical potential opened up between the bubbles and the water, and the CO2 diffused out of the bubbles and into the water. In contrast, the air bubbles were formed and remained in contact with approximately 1 atm of air (mainly N2 and O2), so there is a much smaller gradient in air concentration. Moreover, the solubility of CO2 (∼0.8 L/L) in water is greater than N2 (0.015 L/L) or O2 (0.03 L/L) at 25 °C at 1 atm,46,47 so the transport from the bubbles through the water should be greater for CO2 than for air. The Pressure Inside Nanobubbles is Close to 1 atm. The fine structure in the gas-phase spectrum arises because collisions between gas-phase molecules are sufficiently rare that the molecules exist in discrete rotational states between collisions. The uncertainty principle relates the lifetime of these rotational states to the range of rotational energy changes that are measured. If the peak width is limited by lifetime broadening, then the peak width can be used to determine the time between collisions and thus the pressure using calibration measurements for CO2. Unfortunately, our measured peak widths are similar to the resolution of our instrument. For example, the peak at 2341 cm-1 in Figure 4B has a half width at half-height of 0.11 cm-1 and the resolution is 0.125 cm-1. The dependence of half-width at half-height varies with pressure as 0.120 cm-1 atm-1,48 so the pressure corresponding to the width is 1 ( 1 atm. Thus, our measured peak width sets an upper bound of 2 atm for the pressure in the bubbles. The pressure in the CO2 bubbles can be obtained independently from the intensity of the spectra. The intensity of the CO2 band can be used to measure the concentration of CO2, if we know the molar absorption coefficient and the path length. The molar absorption coefficient was obtained from an independent experiment with ambient CO2 by transmission. The modified (46) Postigo, M. A.; Katz, M. J. Solution Chem. 1987, 16 (12), 10151024. (47) Pollack, G. L. Science 1991, 251 (4999), 1323-1330. (48) Oodate, H.; Fujioka, T. J. Chem. Phys. 1978, 68 (12), 5494-5497.

4762 Langmuir, Vol. 24, No. 9, 2008

Zhang et al.

version of Beer’s law that is applicable to evanescent waves in the ATR configuration is49

ACO2 ) NCO2FCO2d where A is the absorbance in the CO2 band, N is the number of reflections of the IR beam, CO2 is the absorption coefficient, FCO2 is the density of gas in the nanobubbles, and d is the effective path length. The effective path length accounts for the fact that the electric field decays exponentially with distance from the interface. In the current application, the effective path length must also account for the fact that the bubbles do not cover the entire surface and they have a distribution of thicknesses in each experiment. We do not attempt to calculate d, but instead calibrate d by measurement of the loss of absorption by D2O in the same spectrum as the CO2. This has the advantage that the same set of bubbles is used to measure both the effective path length and the pressure. The frequencies of the absorptions are also close, so the correction for the difference in penetration depth is negligible.49 The path length is obtained from

AD2O ) ND2OCD2Od where CD2O is the concentration of D2O solution. D2O is obtained from stage one, using H2O as the background spectrum. This calculation is possible because D2O(aq) is excluded from exactly the same region that CO2(g) occupies. We do make a subtle assumption as follows. The change in the D2O spectrum on formation of the bubbles arises from (a) the loss of D2O that formerly occupied the space now in the bubbles and (b) the change in the spectrum of the D2O elsewhere caused by the decrease in refractive index by insertion of the bubble. This change in refractive index will scatter and refract the light. The penetration depth is smaller through the air, which should decrease the absorbance by the overlying D2O. Ignoring the change in refractive index is reasonable when the thin intervening films are much thinner than the wavelength of light. A typical bubble has a height of 20 nm, whereas the decay length in water is about 250 nm at 2500 cm-1. Using a two-layer optical model,50 we find that the average density of the CO2 gas is 44 ( 16 mol/m3. The large error arises from uncertainty in determining the areas of the partially overlapping CO2 and D2O peaks. The low density of CO2 allows us to accurately calculate the pressure from the ideal gas equation. This density corresponds to a partial pressure of carbon dioxide inside the bubbles of 1.1 ( 0.4 atm. Note that, although our error range includes values less than 1 atm, pressures below 1 atm are unreasonable, because they require a negative curvature of the bubble, which is not observed in the AFM image and is impossible on a flat solid. We used a time series of IR measurements to monitor the pressure in the CO2 nanobubbles as a function of time (Figure 12). The pressure decreases with time, which is consistent with both the disappearance of small bubbles and the loss of gas from the large bubbles. We can also estimate the pressure in the bubbles from the radius of curvature measured in AFM experiments. We perform a least-squares fit of an arc of a circle to a cross section through the apex of the bubble and convert this to a pressure difference across the bubble using the Laplace equation and the surface tension of pure water. To obtain the bubble pressure, we add 1 atm for the pressure in the liquid. The radius of curvature of (49) Hansen, W. N. In AdVances in Electrochemistry and Electrochemical Engineering; Delany, P., Tobias, C. W., Eds.; Wiley: New York, 1973; Vol. 9. (50) Citra, M. J.; Axelsen, P. H. Biophys. J. 1996, 71 (4), 1796-1805.

Figure 12. CO2 pressure inside CO2 nanobubbles as a function of time. The average pressure in the nanobubbles is proportional to the ratio of the area under the CO2 band to the area under the D2O band. There is a large systematic error in the pressure, so we have normalized all data by the ratio at zero time. At time zero the pressure is between 1.0 and 1.7 atm.

the bubbles varies in the range 2-45 µm with a typical radius of ∼4 µm. This corresponds to a pressure inside the bubbles in the range 1.7-1.0 atm, with a typical value of 1.4 atm for CO2. In summary, the three measures of pressure made in the first half hour after bubble formation, 20 nm). The van der Waals force across the air is attractive, so the only candidate is an electrostatic force. It is thought that the air-water interface is negatively charged.51 It is possible that accumulation of negative charge at the airwater interface prevents shrinkage of the interfacial area (i.e., decreases the interfacial tension), leading to stability. This effect would require adsorption of insoluble charged contamination. If the solid-gas interface remains neutral, the interaction between the solid-gas interface and the liquid-gas interface will be an attractiVe image charge interaction. To generate a repulsive electrostatic force, a like (negative) charge would need to accumulate at both the solid-air and liquid-air interfaces. Such a force would stabilize the bubble. Contact Angle of Interfacial Nanobubbles. As described previously,24,33 the contact angle of interfacial nanobubbles is much greater (164° measured through the water) than the macroscopic contact angle for macroscopic droplets (101°-112°). One possible reason for this difference is that forces between the solid-gas and liquid-gas interfaces are significant for the smaller bubbles. As discussed above, both the van der Waals and electrostatic force between the solid-gas and liquid-gas interfaces are expected to be attractive. These attractive forces should lead to an increase in the contact angle (measured through the water) compared to the macroscopic bubble as observed, but also to a dependence of contact angle on the size of the bubble. However, in contrast to earlier work by Yang et al.24 (who also used AFM to determine the contact angle), we find that the contact angle does not vary with the radius of the three-phaseline within the error of the measurements (See Figure 13). That is, we cannot resolve any line tension of the air bubbles. This observation makes it difficult to support the hypothesis that the contact angle is influenced by an electrostatic or van der Waals force. This observation also poses problems for explanations of the stability of the bubble in terms of surface forces. (51) Healy, T. W.; Fuerstenau, D. W. J. Colloid Interface Sci. 2007, 309 (1), 183-188.

4764 Langmuir, Vol. 24, No. 9, 2008

A second possible explanation for the large contact angle for interfacial nanobubbles is that the contact angle hysteresis is different on macroscopic and microscopic scales. The contact angle hysteresis is small (∼11°) for macroscopic droplets (a ∼ 1 mm) on this very smooth solid (0.2 nm rms over an area of 6 µm2). Although smooth on a macroscopic scale, the roughness is significant compared to the bubble height, ∼20 nm, and may contribute to contact angle hysteresis. It is possible that the bubbles began their existence with a larger volume and curvature and a higher contact angle. Then, as gas diffused out, they quickly lost volume and curvature, and the contact angle (through the water) increased to the values that we observe. Unfortunately, we are unable to image the early stages of bubble development to provide evidence of this effect. In contrast, our measurements of the time evolution of bubble at later times, between 1 and 4 days, show that the radius of the three phase line decreases at approximately constant contact angle. (See Table 1.) Effect of Solvents on the Formation of Interfacial Bubbles. If the formation of nanobubbles depends on the degree of gas supersaturation, then the formation of bubbles should not only depend on the difference in solvency of a pair of solvents but it should also depend on the shape of the solvency-composition curve. For this reason, we point out that the change of the solubility of many gases (including CO2) is nonlinear with the concentration of ethanol aqueous solution,46 thereby creating a greater gradient in solubility than that for the ideal solutions. Meanwhile, if the exothermic mixing of ethanol occurs without adequate heat transfer, this will raise the temperature and further decreases the gas solubility.47 This combination of properties makes it particularly easier to form nanobubbles.

Conclusion Interfacial nanobubbles are not always present at the interface between hydrophobic solids and water, but they can be formed under certain conditions. An important example is that interfacial nanobubbles form when air-equilibrated ethanol is used to wash a hydrophobic solid before the solid is exposed to water. The exchange of fluids leaves the solvent supersaturated with gases and causes the precipitation of interfacial nanobubbles. Such changes in solvency are common when cleaning or heating solids and may lead to inadvertent generation of interfacial bubbles. It is much easier to form interfacial nanobubbles on a rough solid: they sometimes occur even without solvent exchange. These gas bubbles are not ubiquitous: we can prepare hydrophobic solids under water that are decorated with bubbles and hydrophobic solids with no bubbles. The nanobubble-free hydrophobic interfaces can be prepared by either (1) not decreasing the solubility of gas in the liquid exposed to the solid, (2) degassing the liquids, or (3) decreasing the solubility of gas gradually. Our strongest evidence for the existence of interfacial gas

Zhang et al.

bubbles is based on ATR-IR. The presence of rotational fine structure in the spectrum of CO2 and D2O proves that molecules are present in the gas phase at the interface. SPR measurements provide supporting evidence that the film is in the gas state: the thin film has a lower refractive index than water, and there are few common contaminants other than gases that satisfy this condition. The gas pressure inside the CO2 bubbles is low. The width of the rotational fine structure shows that the pressure of CO2(g) is less than 2 atm, and the intensity in the IR peaks shows that the pressure of the gas is 1.1 ( 0.4 atm. These results are concordant with the pressure calculated from the curvature using the Laplace equation (Pgas ) 1.0-1.7 atm). The air bubbles are stable for at least days, whereas the CO2 bubbles are only stable for 1-2 h. The longer lifetime of air bubbles is explained by the fact that they are attached to a reservoir of water with a pressure of about 1 atm of gas, whereas the CO2 bubbles are attached to a reservoir of only about 3 × 10-4 atm of CO2. As expected, smaller bubbles have a shorter average lifetime than larger bubbles, and the average pressure of the bubbles decreases with time. For the individual bubbles that we monitored, the curvature decreased with time. This is consistent with Ostwald ripening via gas transport through the solution. The long lifetime of the air bubbles must be due to a small pressure difference across the bubble interface: our IR measurements are consistent with a small pressure difference. However, our AFM measurements of the bubble shape introduce two unsolved problems. The contact angle through the water is much greater than for macroscopic bubbles and the curvature of small bubbles is high. The high contact angle for the bubble suggests that there is an attractive force acting between the solid-air and liquid-air interface. This is consistent with the expected attractive van der Waals force but is at odds with measurements that show (a) constant curvature of each bubble and (b) invariance of the contact angle with radius when we compare different bubbles. Acknowledgment. We thank Abbas Khan for assistance with ATR-FIR measurements, Tim Senden and Nobuo Maeda for assistance in AFM measurement, Darwin Lau for help in programming software used for peak area calculation, and Dalton Harvey and Evan Bieske for useful discussions. This research was supported under Australian Research Council’s Discovery Project funding scheme (DP0664051, DP0880152). W.D. is the recipient of an Australian Research Council Federation Fellowship (FF0348620). X.H.Z. is the recipient of Australian Postdoctoral Fellowship (DP0880152). Travel was funded by the Particulate Fluids Processing Centre, an ARC Special Research Centre. LA703475Q