Langmuir 1996,11, 1422-1426
1422
Surface Tension and Bubble Coalescence Phenomena of Aqueous Solutions of Electrolytes Peter K. Weissenborn” and Robert J. Pugh Institute for Surface Chemistry, Thin Films Group, Box 5607, S-114 86 Stockholm, Sweden Received December 30, 1994@ The inhibition of bubble coalescence by electrolytes beyond a critical transition concentration has been assessed by the relationship between transition concentration and surface tension gradients of electrolytes (d(Ay)ldc)and gas solubility. The correlationbetween transition concentration and [d(Ay)/d~]-~ was mediocre (0.74),suggestingthat the Gibbs-Marangoni effect cannot fully account for inhibition ofbubble coalescence. Preliminarybubbling experiments on mixed electrolytessupported this conclusion. A much better correlation was found between the transition concentration and gas solubility (represented by the exponential decay coefficient of 0 2 solubility with increasing electrolyte concentration). Hence, the inhibition of bubble coalescence may be linked with the decreased dissolved gas concentration in the electrolyte solution.
Introduction The fact that common inorganic electrolytes such as NaCl dramatically reduce the coalescence of gas bubbles above a critical concentration has been known for many years.1-6 Explanations of the observed effects have included electrical repulsive forces,l a reduction in the hydrophobic a t t r a ~ t i o nand , ~ the Gibbs-Marangoni effect/ surface elasticity.8 Underlying the explanations are ionwater interactions and effects on water structure, first pointed out by Lessard and Z i e m i n ~ k i . ~ Recently, it has been shown t h a t only a particular combination of ions can inhibit coale~cence.~ This led to a “combining rule” based on the nature of the cationicanionic pair which predicted whether or not the electrolyte would inhibit coalescence. Of the electrolytes which did inhibit coalescence, some were significantly more effective than others when concentration was expressed in terms of molarity; however all were approximately equally effective when concentration was expressed as the inverse Debye length. Surface tension, viscosity, and electrically repulsive forces were ruled out as possible explanations, with the preferred reason being a reduction in the hydrophobic attraction between bubbles due to the presence of electrolyte. A link between the long-range hydrophobic attraction and dissolved gas concentration in electrolyte solutions was also implicated. An explanation for the inability of some electrolytes to inhibit coalescencewas not apparent. The results have generated much controversy and a resurgence in research on electrolyte effects on bubble coalescence. Even though Craig et aL7ruled out surface tension a s an explanation for the above results, we have chosen to measure the surface tension of a large range of electrolytes using the maximum bubble pressure method in a continuing effort to understand bubble coalescence. Christenson and Yaminsky8 have shown that surface tension gradients (Gibbs-Marangoni effect) play a role in bubble
* Corresponding author
Abstract published in Advance A C S Abstracts, April 15,1995. (1)Marrucci, G.; Nicodemo, L. Chem. Eng. Sci. 1967,22, 1257. (2)Zieminski, S. A,;Caron, M. M.; Blackmore, R. B. Ind. Eng. Chem. Fundam. 1967,6,233. (3) Lee, J. C.; Meyrick, D.L. Trans. Inst. Chem. Eng. 1970,48,T37. (4) Lessard, R. R.; Zieminski, S. A. Ind. Eng. Chem. Fundam. 1971, 10,260. ( 5 ) Zieminski, S.A.; Whittemore, R. C. Chem. Eng. Sci. 1971,26, @
509. (6) Prince, M. J.; Blanch, H. W. AIChE
J. 1990,36, 1425. (7) Craig, V. S. J.; Ninham, B. W.; Pashley, R. M. J. Phys. Chem. 1993,97, 10192. (8)Christenson, H. K.; Yaminsky, V. V. Submitted for publication in J. Phys. Chem. 0743-746319512411-1422$09.00/0
coalescence in electrolyte solutions. Expansion or contraction of bubbles generates the surface tension gradients and the consequent movement of ions and surrounding sheath of water molecules to, from, and along the gas/ water interface slows drainage of solution between the bubbles and hence inhibits coalescence. It could also be suggested that the magnitude of the surface tension gradients (expressed as dyldc, y being surface tension and c the electrolyte concentration) gives a n approximate guide to the efficiency of the electrolyte at inhibiting coalescence. In this Letter we do not intend to offer a n explanation for the mechanism of inhibition of coalescence but want to show that the contrasting behavior of electrolytes in inhibiting coalescence (combining rule) can be explained in terms of positive and negative adsorption of ions at the gaslwater interface. Further, the efficiency of electrolytes which do inhibit coalescence can be linked to the effect of electrolyte concentration on gas solubility. These results have implications on the mechanism of inhibition of coalescence but are not sufficient to offer a complete explanation.
Experimental Section Stock solutions(2 M) of approximately40 inorganicelectrolytes were prepared in water. The electrolytes were of analytical reagent quality and, if available, of purity greater than 99 or 99.5%. Water was purified by a Milli-Q plus 185 system. All glasswarewas cleaned in chromic acid, rinsed 3 times with Milli-Q water, and handled with gloves to avoid contamination. Surface tension was measured using a SensaDyne 6000 tensiometer and version 4.0 software. The tensiometer uses the maximum bubble pressure method with dual capillaries to measure surface t e n ~ i o n .The ~ use of dual capillaries allows measurements to be made independent of solution height or volume. Glass capillariesof diameters 4.0 mm and 0.5 mm were used and bubble pressure was measured at the small capillary. High-purity nitrogen was used to blow the bubbles. Calibration was carried out using Milli-Q water and NaCl (2 M), both of known surface tension.1° Surface tension of the electrolyte solutions was measured relative to water by measuring the surface tension ofwater and then adding stock electrolyte solution in selected increments to give surface tensions for at least five concentrations in the range 0.05-1 M. Temperature was maintained at room temperature to within +0.2 “C. A plot ofthe change in surface tension relative to water ( A y )versus electrolyte concentration ( c ) gave the surface tension gradient (d(Ay)/dc). (9) Mysels, K. J. Colloids Surf. 1990,43,241. (10)CRC Handbook of Chemistry and Physics, 69th ed.; Weast, R. C., Ed.; CRC Press: Boca Raton, FL, 1988; pp F-33, D-272.
0 1995 American Chemical Society
Letters The advantage ofthis method over conventional methods such
as the Wilhemy plate or Du Nouy ring is that bubble pressure (surface tension) is measured for every bubble blown. Typically about 30 to 100 bubbles, depending on the bubble interval, were blown and an average surface tension was obtained. Because a new gadwater interface is generated with every bubble, the method is less sensitive to impurities or contamination. Nevertheless there was some concern over possible organic impurities in the reagents. Earlier workers,l1JZhad to purify their reagents by roasting and recrystallization to get reliable surface tensions of solutions using the Wilhemy plate method. To eliminate any doubt in our results the surface tension of solutions of KCl and NaBr were measured before and after roasting of the solids and no differences in d(Ay)/dc(within experimental error) were found. Since this comparison was not made for all solids, there is still a possibility that some samples (Le., the acetate and tetramethylammonium salts) may contain impurities; however, we felt that the combination ofthe quoted reagent purity and method of surface tension measurement was adequate to give reliable data. A further advantage of the method is that bubble growth more closely resembles the dynamic environment which bubbles experience during coalescence or collision. The rate of bubble growth or bubble interval can also be controlled to give dynamic surface tensions, although this asset is more appropriate to studying diffusion limited adsorption of bulky surfactants. Surface tension of all the electrolytes was, nevertheless, measured at two bubble intervals-1.5 s (relatively slow) and 0.15 s (relatively fast). For selected electrolytes the effect of bubble interval on surface tension was measured over a wider range. At relatively fast bubble rates (maximum rate was 0.10 s) there was a slight change in surface tension relative to slower rates, but this was related to solution viscosity effects. Results for 1.5 s are reported here only. The precision of the bubble method was tested by measuring t h e surface tension of water, NaBr (1 M) and MgS04 (1 M) 7 times, alternating between each solution, at a bubble interval of 1.5 s and 22.0 "C. The mean surface tension ofwater was 72.37 mN m-l with standard deviation of 0.03. The mean change in surface tension relative to water (Ay = y - yo, where yo is for water) and standard deviation of NaBr and MgS04 were 2.06 & 0.05 and 2.71 f 0.05 mN m-l, respectively.
Results of Surface Tension Experiments Graphs of surface tension relative to water versus electrolyte concentration (c = 0 to 1MI gave straight lines (fitted by least-squares regression analysis) of gradient d(Ay)ldc. Values of d(Ay)/dcfor selected electrolytes are shown in Table 1, along with literature values where available, and transition concentrations for bubble coalescence obtained from Craig et aL7 A complete set of results and figures of Ay versus c will be given in a subsequent publication.l4 Table 1shows negative and positive values of d(Ay)/dc. Negative values of d(Ay)/dc(decrease in y ) indicate positive surface excess concentrations of solute or positive adsorption of the solute a t the gaslwater interface. Positive values of d(Ay)/dc(increasein y ) indicate negative surface excess concentrations of solute or negative adsorption (depletion) of the solute from the gadwater interface.15 Negative or positive values of d(Ay)/dcdo not necessarily mean that both cations and anions are positively or negatively adsorbed but that one ion may dominate over the other in terms of overall adsorption and effect on surface tension. Figure 1shows a plot of Ay versus c for (11)Johansson, K.; Eriksson, J. C. J . Colloid Interface Sci. 1974,49,
469.
(12) Ralston, J.;Healy, T. W. J . Colloid Interface Sci. 1973,42,629. (13)Abramzon, A. A,; Gaukhberg, R. D. Russ. J . Appl. Chem. 1993, 66 (61, 1139; 66 (71, 1315; 66 (8), 1473. (14) Weissenborn, P. K.; Pugh, R. J. To be submitted for publication in J . Colloid Interface Sci. (15) Hiemenz, P. C. Principles of Colloid and Surface Chemistry, 2nd ed.; Marcel Dekker: New York, 1986; p 391.
Langmuir, Vol. 11, No. 5, 1995 1423 Table 1. Comparison between Experimental and Literature Values of d(Ay)ldcfor a Range of Electrolytes and Transition Concentrations As Measured by Craig et ~~
d(Ay)/dc electrolyte experimentala HC1 -0.27 f 0.04' LiCl 1.98 f 0.09 NaCl 2.08 f 0.08 KCl 1.85 f 0.05 RbCl nd CSCl 1.52 f 0.07 'NH&1 1.59 f 0.09 0.94 f 0.03 (CH3)4NC1 MgClz 4.06 0.10 CaClz 4.02 f 0.08 SrClz nd Lac13 5.91 f 0.30 0.44 f 0.06 Li~S04 3.58 f 0.05 Nazi304 2.90 f 0.08 MgS04 2.44 f 0.05 HN03 -0.83 f 0.10 NH4N03 1.15 f 0.04 Ca(N03)z 2.47 f 0.11 Cr(NOd3 4.13 f 0.10 HClO4 -2.15 f 0.08 LiC104 0.27 f 0.06 0.22 f 0.06 NaC104 CH3COOH -38 f lf CH3COONa 0.93 f 0.03 0.76 f 0.09 CH3COOK (CH3COO)zMg 0.48 f 0.07 CH3COO(CH3)4N -0.51 f 0.09 0.85 f 0.06 NaI 1.23 f 0.06 KBr nd NaC103 0.89 f 0.05 KOH 1.98 f 0.04 NaCL/HC104 mixg -0.49 f 0.04
d(Ay)/dc literatureC -0.29 1.63 1.55 1.60, 1.65 1.56 1.56 1.34 0.6 3.1420 3.22 3.3820 4.73 0.64 2.6818 2.96 2.241° -0.70 2.0620 nae na -1.64 na 0.73 na 0.54 0.45 na na 0.98 1.45, 1.21 1.97 0.5715 1.7717 na
transition concentrationd (M) no transition nde 0.078 0.120 nd nd 0.100 no transition nd 0.037 nd nd no transition 0.025 nd 0.020 no transition 0.140 0.040 0.046 0.070 nd no transition 0.002 no transition no transition no transition 0.125 nd nd 0.083 no transition 0.053 nd
a Bubble interval 1.5 s. Temperature 21-28 "C, temperature for any one electrolyte within f 0 . 2 "C. The dependence of d(Ay)/dc on temperature between 20 and 30 "Cis within experimental error, e.g., based on literature datal3 for KC1 d(Ay)/dcat 20 "C is 1.56 and at 30 "C it is 1.66. The f error is the standard deviation of data points from the line ofbest fit. Experimental error in d(Ay)/dcwas estimated to be fO.l. Literature values of d(Ay)/dccalculated from data compiled by Abramzon and Gaukhberg,13 except the second mentioned values for KC1, NaI1l and (CH&NCl, CH&OOK.8 Temperature is 25 "C,unless specifiedin superscript. Taken from Craig et aL7 Transition concentration was defined as the electrolyte concentration at which 50% coalescence occurs, on a scale where 100%coalescenceis for water and 0%coalescenceis where no further change in coalescence is measured with increasing electrolyte concentration. Hofmeier et al.19 have since shown that the socalled transition concentration is due to a combination of the initial size of the bubbles emerging from the frit, bubble coalescence, and bursting of larger bubbles to form smaller bubbles. e nd means not determined and na not available. ,Estimated over the range 0.01 to 0.095 M. 8 Stock solution was a 1:l mixture of 2 M NaCl and 2 M HC104. Ionic strength was used to represent c and obtain d(Ay)/ dc.
selected electrolytes. The values of d(Ay)/dc have been distinguished as either > + or l < - 1and the cations and anions classified as either negatively (--) or positively (+) adsorbed. The basis for the classification of cations and anions was the value of d(Ay)/dc. For electrolytes with d(Ay)/dc > +1 both cation and anion are strongly negatively adsorbed (- -) and for electrolytes with d(Ay)/ dc < -1 both cation and anion are strongly positively adsorbed (+ +). A special situation arises for electrolytes which combine, for example, a cation classified as negatively adsorbed and an anion classified as positively adsorbed (-- +) or vica versa (+ -). These electrolytes gave d(Ay)/dcbetween -1 and +l. The actual value of
Letters
1424 Langmuir, Vol. 11, No. 5, 1995 -300
1
-250
promotion and compactionof water SVLLcNre
d(Ay)/dc > + I
-- combination of ions cooperative negative adsorption of ions
1
d(Ay)/dc between + I and-1
I50
-+ or +- combination of ions
200
competitive adsorption of ions
d(Ay)/dc < -1
HCIOY 1
0.2
0.4
-
++ combination of ions
.
cooperative positive adsorption of ions
-
0.8 1 Electrolyte concentration (M)
0.6
Figure 1. Effect of electrolyte concentration on the change in surface tension relative to water. Bubble interval was 1.5 s. Electrolytes were classified accordingto the magnitude of d(Ay)l dc (see text). A full account of the effects of electrolyte concentration on surface tension will be given in a subsequent
publication.l4 Data points have been eliminated for clarity. Experimental error in d(Ay)ldc was estimated to be fO.l.
d(Ay)/dcin all cases is the summation of the adsorption affinities (positive or negative) of the individual ions. The value of f l has been chosen arbitrarily to distinguish between a strong ( > + or l 1:2 1:3. Within any group (e.g., 1:l) there is no significant change in oxygen solubility for different cations, but Li+ concentration seems to have consistently the least effect. The mineral acids had the least effect on oxygen solubility. For chloride electrolytes common to our surface tension measurements, a plot of the exponential decay coefficient for oxygen solubility versus d(Ay)/ dc gave a straight line with correlation coefficient of 0.96 (Figure 3). Taking the effect of dissolved gases further, a plot of the transition concentration from Craig et aL7 versus exponential decay coefficient for oxygen solubility
+
-
OHF-
I
B ~ -
-
N
Y insol
insol insol insol insol
Y
insol insol insol
I I IYIY I I I
I I
IinsolI
I I
I
a + + catiodanion pairs have d(Ay)idc less than -1, - - cation/ anion pairs have d(Ay)idc greater than +1, + - or - + cation/ anion pairs have d(Ay)ldc between +1 and -1. CH3COOMe4N is the only exception. Y = yes and d(Ay)/dc > +l or < -1, N = no and d(Ay)dcbetween -1 and fl,insol means electrolyte is insoluble at 1M, blank space means electrolyte commercially unavailable o r not determined.
and the nature of the counterion. Further, in a dynamic system the expanding or contracting gadwater interface influences transport of ions to or away from the interface (Gibbs-Marangoni effect). No matter what the underlying forces are, it is the outcome of positive or negative ion adsorption and the combination of ions which determines the magnitude of the surface tensiodconcentration gradient (Figure 1). For - - combinations of ions, both ions move away from the interface (cooperative negative combinations both ions move toward adsorption). For the interface (cooperative positive adsorption). For or - combinations the ions want to move in opposite directions (competitive adsorption). Relationship between Transition Concentration for Bubble Coalescence and d(A.y)/dc. Using the and - classification, Table 2 summarizes the results for all electrolytes measured in terms of d(Ay)/dcbeing greater or less than h l . A comparable table was constructed by Craig et al.7which predicted the coalescence behavior of electrolyte solutions, except ions were designated as either a or p. The link between Table 2 and that of Craig et aL7 is that electrolytes with d(Ay)/dcbetween approximately -1 and +1gave no transition in bubble coalescence (the only anomaly was CH3COO(CH3)4N,classified as with d(Ay)/dcof -0.51 and a transition concentration of 0.125 M). Further, ions designated as a are negatively adsorbed a t the interface and ions designated as p are positively adsorbed. Christenson and Yaminskys and Hofmeier et al.19 alluded to the same designation. In an attempt to explain bubble coalescence in electrolytes Christenson and Yaminskf plotted the transition concentrations from Craig et aL7against [d(Ay)/d~]-~, using surface tension data from literature. A straight line of slope 6.5 and correlation coefficient of 0.92 was obtained. Using our experimentally determined values of [d(Ay)/ we plotted the same graph and obtained a slope of 4.1 h 0.7 with correlation coefficient of 0.74, ignoring the
++
+
+
+
++
(19) Hofmeier, U.; Yaminsky, V. V.; Christenson, H. K. J. Colloid Interface Sci., in press.
+
(20)Ninham, B. W. Personal communication. (21)Ninham, B. W.; Claesson, P. M.; Eriksson, J. C. Personal
communication.
Letters
1426 Langmuir, Vol. 11, No. 5, 1995 -0.8
’
0
2
1
3 d(Ay)/dc
5
4
Ldf.
6
Figure 3. Correlation between exponential decay coefficient for oxygen solubility in cationic chloride solutions and d(Ay)/ dc. Correlation coefficient = 0.96.
4
4
0.03
0 ‘ -0.8
-0.6
-0.4
-02
I
0
exponential decay coefficient for oxygen solubility
Figure 4. Correlation between transition concentration from
Craig et aL7 and exponential decay coefficient for oxygen solubility. Oxygen solubility data are for 37 “C and taken from ref 10. Correlation coefficient = 0.95. gives a n exponential curve with correlation coefficient of 0.95 (Figure 4). Using surface force techniques, researchers have obtained conflicting results on the effect of electrolytes on the hydrophobic interaction between hydrophobic surf a c e ~ . ~The ~ , role ~ ~of- dissolved ~~ gases in the hydrophobic (22) Meagher, L.; Craig, V. S. J. Langmuir 1994,10, 2736. (23) Parker, J. L.; Claesson, P. M.; Attard, P. J . Phys. Chem. 1994,
98,8468.
(24) Tsao, Y.-H.; Fennel1Evans, D.; Wennerstrom, H. Langmuir 1993, 9,779. (25)Christenson, H. K.; Claesson, P. M.; Parker, J. L. J.Phys. Chem. 1992,96,6725.
interaction is a topic of interest. Experiments show that vapor cavities form on contact between hydrophobic surfaces, and recently there are indications that bridging cavities may be one reason for the long-range attraction observed between silanated glass surfaces.23In this case, no decrease in the attraction upon the addition of up to 5 M NaCl was observed.23 Recent atomic force microscope (AFM) measurements have shown that removal of dissolved gas decreases the long-range hydrophobic interaction between polypropylene surfaces in aqueous dilute solutions of NaC1.22 However, measurements in nondegassed solutions showed that NaCl concentration (1 x and 1M) had no effect on the interaction. 1x In contrast, Ducker et a1.18have also used AFM to suggest that a long-range attractive force is present between a bubble and hydrophilic surface and that this force is decreased (based on jump distances) in going from to M NaC1. The results obtained by most surface force techniques are based on interactions between two solid surfaces separated by a thin liquid film and obvious caution should be used in using these results to explain interactions between two gas bubbles dispersed in solution. Our present study indicates a good correlation between the exponential decay coefficient for oxygen solubility and both d(Ay)/dc and transition concentration, suggesting that dissolved gas concentration has an important influence on the interaction between two bubbles. The dissolved gas concentration is dependent on electrolyte concentration and hence so is the interaction between two bubbles.
Conclusions The effect of electrolyte concentration (up to 1 MI on surface tension has been measured for a wide range of electrolytes using the maximum bubble pressure method. The dependence of surface tension on electrolyte concentration was related to the entropy of hydration of the ions in solution. Strongly hydrated ions such as La3+had the largest effect on surface tension. The surface tension gradients (d(Ay)/dc)and bubbling experiments were used to investigate the bubble coalescence results of Craig et al.7 Their “combining rule” can be restated in terms of positive and negative adsorption of ions a t the gadwater interface and the so-called transition concentration for bubble coalescence correlates well with the exponential decay coefficient for oxygen solubility in electrolytes and to a lesser extent with [d(Ay)/d~]-~. The results suggest that dissolved gas concentration plays an important role in the coalescence of bubbles in aqueous electrolyte solutions, but a contribution due to the Gibbs-Marangoni effect and surface elasticity cannot be ruled out. Acknowledgment. We thank Professors B. N. Ninham and J. C. Eriksson, Associate Professors P. M. Claesson and B. Kronberg, and Dr. H. K. Christenson for support, discussion, and interest in this work. Special thanks to P. M. Claesson and B. Kronberg for critically reading the manuscript and the Wenner-Gren Foundation for a research award to P.K.W. LA941045E