Journal of Colloid and Interface Science 422 (2014) 54–57
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
pH and the surface tension of water James K. Beattie a,⇑, Alex M. Djerdjev a, Angus Gray-Weale b,⇑, Nikola Kallay c, - Selmani c Johannes Lützenkirchen d,⇑, Tajana Preocˇanin c, Atida a
School of Chemistry, University of Sydney, NSW 2006, Australia School of Chemistry, University of Melbourne, Victoria 3000, Australia c Department of Chemistry, Faculty of Science, University of Zagreb, Horvatovac 102a, HR-10000 Zagreb, Croatia d Institut für Nukleare Entsorgung, Karlsruher Institut für Technologie (KIT), Herrmann-von-Helmholtz-Platz 1, 76433 Eggenstein-Leopoldshafen, Germany b
a r t i c l e
i n f o
Article history: Received 20 November 2013 Accepted 4 February 2014 Available online 22 February 2014 Keywords: Surface tension Air/water interface pH Gibbs isotherm Salt effect
a b s t r a c t Despite the strong adsorption of hydroxide ions, the surface tension of water is almost independent of pH between pH 1 and 13 when the pH is adjusted by addition of HCl or NaOH. This is consistent with the Gibbs adsorption isotherm which measures the surface excess of all species in the double layer, if hydronium ions and hydroxide ions are adsorbed and sodium and chloride ions are not. The surface tension becomes pH dependent around pH 7 in millimolar NaCl or KCl solutions, for now sodium ions can replace hydronium ions as counterions to the adsorbed hydroxide ions. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Surface tension is a fundamental property of water that affects many phenomena . Textbooks ascribe the molecular origin of the surface tension to the difference in energy of molecules between the bulk and the pristine, surfactant-free surface. Yet it has been known for many years that air bubbles, as well as oil drops and inert polymer surfaces, acquire a negative charge in water . This is attributed to the spontaneous adsorption of hydroxide ions at the interface. The pH dependence of the pristine surface tension of water now presents an apparent paradox. On the one hand is the compelling evidence, including the pH dependence of the f potential, that the interface is charged by the strong adsorption of hydroxide ions . On the other hand there is indirect evidence that the surface tension is nearly pH independent around neutral pH. These two observations appear to be contradictory and have created an impression that the question of the charge at the interface is controversial . The Gibbs adsorption isotherm provides an exact relationship between bulk concentrations, total surface coverage of adsorbed species, and the surface tension changes produced by adsorption [4,5]. Its careful application leads to the prediction that the surface tension around pH [ 7 is independent of pH, despite the high surface concentra⇑ Corresponding authors. Fax: +61 2 9351 3329 (J.K. Beattie). E-mail addresses: [email protected]
(J.K. Beattie), [email protected]
edu.au (A. Gray-Weale), [email protected]
(J. Lützenkirchen). http://dx.doi.org/10.1016/j.jcis.2014.02.003 0021-9797/Ó 2014 Elsevier Inc. All rights reserved.
tion of the hydroxide ion . Only in the presence of additional electrolyte is a pH dependence of the surface tension for expected pH < 7. An explanation of this phenomenon, the absence of a pH effect in the presence of hydroxide ion adsorption, and the report of new surface tension data, are the subjects of this Article. We present experimental results that resolve the apparent contradiction about the pH dependence of the surface tension, and that provide further evidence that the negative surface charge is created by adsorbed hydroxide ions. In fact, the appearance of pH dependence of the surface tension only in the presence of added salt rules out many mechanisms for the negative surface charge other than hydroxide adsorption . There are surprisingly few data on the surface tension of water around neutral pH. The value for neat water is of course very well established (72.7 mN m1 at 293 K ) and recommended as a standard for surface tension measurements. It has generally been assumed that this is the surface tension of an uncharged, pristine interface, but two of us have recently described it as the surface tension of the hydroxide-adsorbed interface which forms from the pristine state in a few milliseconds . The fact that this gives a reproducible constant value that can be used as a standard implies that the surface tension must be independent of pH around pH 5–7. Similarly, there are few data available in the present literature for the surface tension of water over a wide pH range. Experimental data for surface or interfacial tensions were traditionally collected for the very extreme pH ranges and for neat water (i.e.
J.K. Beattie et al. / Journal of Colloid and Interface Science 422 (2014) 54–57
pure water without added electrolyte) and with variable electrolyte concentrations. Classical examples are measurements of the surface tensions of electrolyte solutions of concentrations 0.1– 1.0 M and studies on the Jones-Ray effect , a minimum in the surface tension at millimolar electrolyte concentrations in neutral water. The Jones-Ray effect has recently been conﬁrmed spectroscopically . The effect was originally, and again recently, interpreted in terms of electrolyte ion adsorption, but an alternative view is that it reﬂects salt effects on hydroxide ion adsorption [6,11,12]. Work by Lorenz is a rare example of the measurement of the pH dependence of surface tension from [H+] = 1 mM to [H+] > 1 M, but this study reported an inverse Jones-Ray effect, a maximum in the surface tension of water with HCl concentrations below about 30 mM HCl . The purpose of the present work is to clarify the interpretation of the pH dependence of the surface tension, with additional experiments as required, and with particular attention to the surface adsorption of hydroxide.
2. Experimental methods Surface tension measurements on the pH dependence of neat water solutions (Fig. 1) were done using the Cahn Radian 321 set-up with the plate method. Before each measurement, the platinum plate was cleaned by rinsing with water, rinsing with acetone, a ﬁnal rinsing with ethanol, and then exposure of the side that was cleaned to UV-light for 5 min. Then the procedure was repeated for the other side of the plate. The initial surface tension value of neat water could slightly vary on a daily basis between 71.8 and 73.4 mN/m. For all measurements, a pH-series was only started when the result for the pure water was 72.8 mN/m as expected, and remained so for at least three consecutive measurements. In some of the series we also prepared multiple samples for analysis in parallel. These were (i) kept in the initial state for a ﬁnal check after the measurement series to verify that the sample and the set-up still produced the initial value, (ii) treated in the same way as the measurement sample and measured at the end of the series, and (iii) treated in the same way as the measurement sample, but used exclusively for pH-measurement. The above precautions were taken to avoid variations in surface tension due to immersion of the pH electrode in a measurement solution . The pH measurements were done using a combination electrode
Fig. 2. The measured surface tension, c, varying with pH in the presence of two monovalent salts at 10 mM concentration and 298 K. The error bars are twice the standard error in the mean calculated for the 20 independent measurements.
that was calibrated daily against commercial pH buffers (Merck, buffers in the range from pH 1 to 13). Error bars shown in Fig. 1 are the standard deviations of at least 5 independent experiments for each solution. The solution was prepared by ﬁlling a container that had been ﬁlled with argon by water. The solutions were kept always under an argon cushion, and the set-up was always ﬂushed with argon. The argon was sent through a number of washing bottles that contained acid, base (to remove carbon dioxide), and water. Some long-term experiments with pure water were also conducted that showed that one solution could produce a constant value of the surface tension over 10 h. Surface tension measurements of the 10 mM salt solutions began with 10 mM NaCl or KBr solutions in Milli-Q water at pH 6–7 and were conducted under an atmosphere of argon with the pendant drop method on a CAM200 instrument (Fig. 2). Each data point is the average of 20 independent measurements with error bars given by twice the standard error in the mean, and represents an experiment with a single addition of HCl, HBr, NaOH or KOH to the measured pH; there was no incremental addition to obtain multiple pH values with the same sample. It should be noted that surface active impurities would generally lower the surface tension, not increase it as we observed in the acidic solutions. The surface tensions vary smoothly from pH less than 7 to pH greater than 7, where added salt lowers the surface tension. This smooth variation is consistent with similar surface composition above and below pH 7. We conclude then that amphiphilic impurities are not responsible for the observed surface tension behaviour. 3. Results and discussion
Fig. 1. pH dependence of surface tension of neat water at 298 K with added HCl, HClO4, HNO3, or NaOH. The error bars are the standard deviations of at least 5 independent measurements. Note that the upper panel of this plot is an expansion of the lower.
The results of our new measurements of the surface tension of water at 298 K in the absence of added electrolyte are shown in Fig. 1. Only an acid (HCl, HClO4, or HNO3) or NaOH was added to adjust the pH. As described above, measures including work under argon were taken to exclude carbon dioxide in these experiments. The range of pH studied is enough to close the gap between the pH values of 1 and 13 for which data are already available . Except for a small local minimum around pH 4, a probable manifestation of the Jones-Ray effect, the surface tension is essentially constant from pH 1 to pH 13. Langmuir  disputed the surface tension minimum seen by Jones and Ray, but it was later seen also by Dole and Swartout , and recently surface spectroscopic methods were used to verify the effect . Nearly constant surface tension could be interpreted to indicate that none of the ions
J.K. Beattie et al. / Journal of Colloid and Interface Science 422 (2014) 54–57
present are surface active, but in fact as discussed below this is not a valid inference. At pH < 0 the surface tension decreases signiﬁcantly, indicating adsorption of HCl from strong acid solutions. At pH > 13 the surface tension increases, indicating a lower surface concentration of hydroxide ion than in the bulk solution. These limiting cases are consistent with previous observations of surface tension in electrolyte solutions of 0.1 M or greater . As noted above, the measurement shown in Fig. 1 cover the pH gap between data previously published . Consider ﬁrst the data for pH < 7 shown in Fig. 1. These measurements are consistent with strong adsorption of hydroxide ion according to the Gibbs adsorption isotherm. In acidic solutions where HCl is the only electrolyte there are three possible ions in the double layer: hydroxide, hydronium and chloride. The surface tension is given by the Gibbs Eq. (1):
dc=RT ¼ COH d ln aOH þ CH3O d ln aH3O þ CX d ln aX
where c is the surface tension (mN m ), Cx is the surface coverage of species x (mol m2) and aX is the activity of species X, proportional to the concentration for the dilute solutions employed here. If there is negligible adsorption of chloride anions
dc=RT ¼ C d ln ðaOH aH3O Þ ¼ C d ln K W ¼ 0
there is no pH dependence of the surface tension, despite the adsorption of hydroxide ions. In other words, in acidic solutions in which the hydroxide ion is adsorbed at the interface, an equal concentration of hydronium ions must serve as counterions to maintain interface electroneutrality. Now consider a change in pH, say from pH 6 to pH 4. If the surface concentration of hydroxide ions were to remain the same, its surface excess would increase by a factor of 100, as the pH is reduced. But the surface excess of hydronium ions would decease by the same amount, as the bulk concentration of protons increases with the lower pH. So by electroneutrality the two effects cancel and the surface tension remains the same. This is the ﬁrst paradox resolved: the interface can undergo a transition from highly charged by hydroxide ion adsorption at pH 7 to uncharged at the isoelectric point without any change in the surface tension, a consequence of the hydronium ion and hydroxide ion concentrations being linked by the autolysis equilibrium. An alternative explanation for the pH independence, that there is no signiﬁcant adsorption of hydroxide ions, is eliminated by the pH effect observed in 10 mM NaCl or KBr (MX) solutions (Fig. 2). Now the Gibbs equation is
dc=RT ¼ COH d ln aOH þ CH3O d ln aH3O þ CM d ln aM þ CX d ln aX ð3Þ Because the added salt is at a ﬁxed concentration, dlnaM is now zero but [M] is large enough for some adsorption of M to replace H3 Oþ . On addition of HCl dlnaCl is not zero. There are three possibilities. (i) An amount < D > of M replaces H3O and there is no adsorption of Cl. Then
dc=RT ¼ C d ln aOH þ ðC DÞd ln aH3O ¼ D d ln aH3O ¼ ln ð10Þ D dpH
Fig. 3. (A) Conventional view of the force creating surface tension. (B) Alternative view to include the surfactant activity of the adsorbed hydroxide ions.
This gives the opposite slope to that observed because aHCl changes in the opposite direction to pH. (iii) The third possibility is a mixture of these two, but the negative slope indicates that the adsorption of the cation, Na+ or K+, outweighs that of the halide anions. In the presence of salt the concentration of Na+ or K+ in the double layer changes with pH to keep the interface electrically neutral. The cancellation of terms in the Gibbs isotherm due to hydronium and hydroxide no longer occurs. Here the exact Gibbs isotherm predicts that a hydroxide surface layer produces a decrease in surface tension with pH. This is just the result that two of us predicted when presenting the ﬂuctuation correlation model the explain the force that leads to the hydroxide adsorption . In that case a Poisson–Boltzmann model was used to account for the pH dependence of the zeta potential. In the present case two different Poisson– Boltzmann models were used to account for the surface tension. In each case the same 20 kBT was found for the hydroxide surface binding energy. Details are given in the accompanying Supplementary Data. The results in Fig. 2 are consistent with a signiﬁcant adsorption of hydroxide ion over a wide pH range, but not with its absence. They are not explained by surface afﬁnities of the K+, Na+, Br, or Cl ions. The addition of HCl to neat water from pH 7 to 4 results in no change to the surface tension, as explained above (Fig. 1). But in the presence of NaCl the same addition of HCl does change the surface tension (Fig. 2). This is not the consequence of any strong surface afﬁnity of the sodium or chloride ions, for those afﬁnities would not lead to the observed pH dependence, but rather is a result of the strong, speciﬁc, surface afﬁnity of the hydroxide. This conclusion rests only on the Gibbs adsorption isotherm, an exact law for the dependence of surface tension on bulk concentration and surface coverages. It is important to note that the addition of salt leads to an increase in surface tension, so that the effect is not due to impurities, which would produce a decrease.
ð4Þ This gives the sign of the slope in Fig. 2, consistent with possibility (i). (ii) The second possibility is that an amount D of X replaces OH and there is no adsorption of M. Then
dc=RT ¼ ðC DÞd ln aOH þ C d ln aH3O þ D d ln aX ¼ D d ln ðaH3O aX Þ ð5Þ
4. Conclusions 4.1. Hydroxide adsorption is consistent with Gibbs Adsorption Isotherm The experimental data behave as expected from the exact Gibbs isotherm for a strong hydroxide surface afﬁnity. In the absence of cations other than the proton (Fig. 1, pH < 7), the hydroxide surface layer is neutralised by a diffuse layer of protons, and electroneu-
J.K. Beattie et al. / Journal of Colloid and Interface Science 422 (2014) 54–57
trality requires the coverage of these two species to be equal. A change in pH will then produce no change in surface tension, by the Gibbs isotherm, because the two coverages remain equal (C(H+) = C(OH)), and the two species’ contributions to dc cancel. With 10 mM added salt, other cations in the diffuse layer contribute to balance the hydroxide charge and a variation of surface tension with pH is observed. Claims that surface tension implies an attraction for the hydronium ion and a repulsion of the hydroxide ion are unwarranted ; they are based on behaviour at the limits of high and low pH , as shown in Fig. 1, and are not relevant for near-neutral solutions. Taken together, the results of Figs. 1 and 2 show conclusively that hydroxide is the adsorbing ion. 4.2. The dynamic surface tension relaxes to the equilibrium value in milliseconds Various experiments over many decades have shown that the surface tension of a newly formed, pristine surface is larger than the equilibrium value and relaxes to the equilibrium surface tension in about a millisecond. Of the alternative explanations for this effect, the generation and adsorption of hydroxide ion has the most appropriate time constant . The fact that the dynamic surface tension decreases implies that a surface active species is being adsorbed. From the surface charge density the pristine surface tension can be estimated to be 10–15 mN m1 larger than the equilibrium value of 72 at 298 K. 4.3. Simulations have been wrong There have been numerous simulations of water at the air or vapour/water interface. Almost all have given the wrong sign of the charge at the surface [18,19], or predict a very weak hydroxide ion afﬁnity . Recently several authors have observed that this is inevitable insofar as the simulations have only been possible with sample sizes too small to encompass the surface charge density , which has been known for at least ten years . 4.4. Predictions about the adsorption of ions have been fatally ﬂawed Simulations about the adsorption of cations and anions at the air/water interface have been in serious error because the charge of the adsorbed hydroxide ions has not been included in the models [23,24]. So all of the discussion and development of concepts of polarisability, dipole orientation, quadrupoles, etc. have been irrelevant. A gauge to estimate the difference in surface afﬁnities between hydroxide and iodide ions can be obtained by noting that the half-adsorption pH for the adsorption of hydroxide is 5.5, i.e. the hydroxide concentration is 108.5 while the half adsorption concentration for iodide ion observed by surface spectroscopy  is >1 mM, a difference of 105. In other words the iodide experiment measured the adsorption of iodide to a hydroxide charged interface. The experimental data need to be reexamined with the pH controlled, and the theoretical models must incorporate the hydroxide-charged interface. The so-called intense controversy about the charge at the air/water interface is a consequence of theoretical models that do not incorporate all of the experimental evidence .
4.5. Textbook revision required The standard textbook explanation for surface tension is illustrated by a ﬁgure which shows a force attracting surface molecules to the interior of a drop (Fig. 3A). This force arises because molecules at the surface are not subject to isotropic interactions as are molecules in the bulk interior. The present work shows that the equilibrium surface tension also involves the surfactant-like effect of hydroxide ions at the interface (Fig. 3B), so the equilibrium surface tension is the result of two opposing forces. Acknowledgments This work was supported by a Grant from the Australian Research Council. The calculations were supported by grants of computer time from the National Computational Infrastructure National Facility, and the Victorian Life Sciences Computing Initiative. A.S, T.P. and N.K. were supported by the Ministry of Science, Education and Sports of the Republic of Croatia (Project No. 119-1191342-2961). We thank Dr Richard O’Brien and Mr. Maoyuan Liu for helpful discussions and a Reviewer and the Co-Editor for critical and thoughtful suggestions to improve the presentation of the work. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcis.2014.02.003. References  R.J. Hunter, Foundations of Colloid Science, Oxford University Press, Oxford, 2001.  J.K. Beattie, A.M. Djerdjev, G.G. Warr, Faraday Discuss. 141 (2009) 31.  P. Jungwirth, Faraday Discussions 141 (2009) 9.  J. Lyklema, Fundamentals of Interface and Colloid Science, vol. I, Academic Press, London, Fundamentals, 1991.  J.P. Hansen, I.R. McDonald, Theory of Simple Liquids, Academic Press, London, 2006.  A. Gray-Weale, J.K. Beattie, Phys. Chem. Chem. Phys. 11 (2009) 10994.  N.B. Vargaftik, B.N. Volkov, L.D. Voljak, J. Phys. Chem. Ref. Data 12 (1983) 817.  M. Liu, J.K. Beattie, A. Gray-Weale, J. Phys. Chem. B 116 (2012) 8981.  G. Jones, W.A. Ray, J. Am. Chem. Soc. 63 (1941) 288.  P.B. Petersen, J.C. Johnson, K.P. Knutsen, R.J. Saykally, Chem. Phys. Lett. 397 (2004) 46.  M. Manciu, E. Ruckenstein, Adv. Colloid Interface Sci. 105 (2003) 63.  K.A. Karraker, C.J. Radke, Adv. Colloid Interface Sci. 96 (2002) 231.  P.B. Lorenz, J. Phys. Colloid Chem. 54 (1950) 685.  A.A. Fedorova, M.V. Ulitin, Russ. J. Phys. Chem. A 81 (2007) 1124.  P.K. Weissenborn, R.J. Pugh, J. Colloid Interface Sci. 184 (1996) 550.  I. Langmuir, Science 88 (1938) 430.  M. Dole, J.A. Swartout, J. Am. Chem. Soc. 62 (1940) 3039.  V. Buch, A. Milet, R. Vacha, P. Jungwirth, J.P. Devlin, PNAS 104 (2007) 7342.  M. Mucha, T. Frigato, L.M. Levering, H.C. Allen, D.J. Tobias, L.X. Dang, P. Jungwirth, J. Phys. Chem. B 109 (2005) 7617.  C.J. Mundy, I.F.W. Kuo, M.E. Tuckerman, H.-S. Lee, D.J. Tobias, Chem. Phys. Lett. 481 (2009) 2.  T. Cecchi, J. Phys. Chem. C 117 (2013) 19002.  J.K. Beattie, A.M. Djerdjev, Angew. Chem. Int. Ed. 43 (2004) 3568.  P. Jungwirth, D.J. Tobias, J. Phys. Chem. B 106 (2002) 6361.  C. Caleman, J.S. Hub, P.J. van Maaren, D. van der Spoel, Proc. Natl. Acad. Sci. USA 108 (2011) 6838.  R.K. Saykally, Nat. Chem. 5 (2013) 82.