2226
The Journal of Physical Chemistry, Vol. 83, No. 17, 7979 I
1
EC LOCAL1ZED STATES
SOLUTION REDOX LEVEL
S. Ross and R. E. Patterson
may result from additive-induced defect structures, We are encouraged, nevertheless, by the results in Table I1 as a verification of our original design philosophy. In order to tailor the bulk properties of photoelectrodes for water decomposition, it will be important to consider the radii, internuclear separations, and local symmetries of the dn ions. These factors are key to determining whether a d band is formed, what its width will be, and whether it is partially or completely filled.
References and Notes Figure 4. Schematic model for long wavelength photoresponse of semiconductors with localized intragap electrons.
the long wavelength sensitization have a single source. The formation of "d bands" can occur only if the d orbitals of the incorporated transition metal ions show significant overlap. This overlap might be achieved, for example, if these ions were ordered in a given plane. Otherwise, the d" additives can form localized states lying within the band gap, as shown in Figure 4. Carriers in the localized levels have extremely low mobilities. Hence, excitation of an electron from the localized level to the conduction band in the bulk would not lead to efficient separation to the electron and hole. Only holes produced at the surface would be active in producing photocurrent via reaction with the solution species, in this case H20.This is why the uisible wavelength photocurrent is so low. These localized partially filled energy levels can also function as recombination centers for electrons and holes produced by band gap excitation; hence, the reduced total photocurrents. The increased conductivities of many of the d" compounds relative to the parent compounds may be related to a hopping process between localized states, or
(1)K. Rajeshwar, P. Singh, and J. DuBow, Electrochim. Acta, 11, 1 1 17 (1978). (2) A. J. Nozlk, Annu. Rev. Phys. Chem., 29, 189 (1978). (3) L. A. Harris and R. H. Wilson, Annu. Rev. Mater. Sci., 8,99 (1978). (4) M. A. Butler and D. S. Ginley, J. Ektrmhem. Scc., 125,228(1978). (5) H. Hovel, "Solar Cells: Semiconductors and Semimetals", Vol. 11,
R. K. Willardson and A. C. Beer, Ed., Academic Press, New York, 1975. (6)J. J. Loferski, J . Appl. Phys., 27, 77 (1958). (7) F. S.Fllip6v and E. G. Fesinko, Sov. Phys. Crystakbgr.(Engl. Trans/.), 10,243,532 (1965). (8) H. Tributsch, Ber. Bunsenges. Phys. Chem., 81,362 (1977);82, 169 (1978). (9)H. Trlbutsch, Ber. Bunsenges. Phys. Chem., 82, 169 (1978). (IO) H. Tributsch, J. Electrochem. Soc., 125, 1086 (1978). (11) J. G. Goodenough, Prog. Solid State Chem., 5 , 145 (1972). (12) H. Okomoto and T. Aso, Jpn. J. Appl. Phys., 6, 779 (1967). (13)D. Rogers, R. Shannon, and J. Gillson, J. SolMSfate Chem., 3, 314 (1971). (14) R. D. Rauh et al., Flnal Report, Contract EC-77-C-01-5060, US. Department of Energy, Feb 1979. (15) A. K. Ghosh and H. P. Maruska, J . Electrochem. SOC.,124,1516 (1977). (16)J. B. Goodenough and J. M. Longo, "Crystallographic and Magnetic Properties of Perovsklte-Related Compounds", Landolt-Bornstein Tabellen, New Series II1/4a, Springer-Verlag, Berlin, 1970,p 126 ff. (17) C. N. R. Rao and G. V. S. Rao, Phys. Status Sola/ A , 1, 597 (1970). (18) L. A. Bursill and B. G. Hyde, Prog. SolEdstate Chem., 7 , 177 (1972). (19)R. J. H. Voorhoeve in "Advanced Materials in Catalysis", J. J. Burton and R. L. Garten, Ed., Academic Press, New York, 1977,p 129.
Innate Inhibition of Foaming and Related Capillary Effects in Partially Miscible Ternary Systems Sydney Ross" and Robert E. Patterson' Department of Chemistty, Rensselaer Polflechnlc Institute, Troy, New York 12185 (Received June 28, 1978; Revlsed Manuscript Received April 27, 1979)
Capillary activity of solutions and phenomena dependent thereon are found to appertain to certain sets of conditions of temperature and composition corresponding to the same features of phase diagrams regardless of the system. More particularly, regions of innate inhibition of foaminess are located within the miscibility gap wherever one solution, which acts as the foam inhibitor, has a positive spreading coefficient with respect to its equilibrium conjugate solution, but even then only where the foam-inhibiting conjugate is the internal phase of the liquid/liquid dispersion formed by the two conjugate solutions. These relations are confirmed in the ternary systems benzene-water-ethanol and n-hexane-water-ethanol at 20 "C.
Capillary Phenomena and Phase Diagrams The feasibility of mapping phase diagrams so as to include capillary (Le., surface-active) phenomena has only recently been apprehended. One such connection has indeed been known for a long time to metallurgists, namely, the observation that frozen eutectic melts have a particularly fine-grained structure; but this feature has t PQ Corporation, Research and Development Center, P.O. Box 258, Lafayette Hill, PA 19444.
0022-365417912083-2226$01 .OO/O
not been recognized as being only one of a general class of capillary phenomena that can be associated with certain well-defined regions of phase diagrams. Yet one can readily deduce the existence of more such phenomena, based on the general principle that interfacial energies tend toward zero a t critical temperatures and at consolute points. After that, a further advance might well require the genius of Langmuirl to recognize that surface activity inheres in the phase-diagram vicinity of such points; actually, this brilliant insight was formulated and lay around 0 1979 American Chemical Society
Capillary Effects
in Ternary Systems
unattended to for half a century. The colligation of various capillary effects into a general field of enquiry was suggested by Ross and N i s h i ~ k awho , ~ ~found ~ that the only correlations to be made of their measurements of foaminess of multicomponent solutions lay in linking the observations to the phase diagrams. They discovered that one-phase compositions near the consolute point of a three-component system, or near the critical temperature of a two-component system, are surface active, as is shown by their being able to produce a relatively stable foam. At about the same time, Gahn4demonstrated that, on a solid substrate such as the container wall, strong adsorption of a component in a binary solution occurs in the region of the critical temperature; and tentatively sketched the position of such capillary-active solutions on a phase diagram. One can now readily appreciate that more information than has yet been exploited about the capillary behavior of materials can be drawn from phase diagrams; and that investigations of these and other phenomena, related to well-defined regions of phase diagrams, might yield generalities of behavior, and so become amenable to prediction and control. Such phenomena include foam stability and the defoaming action of conjugate solutions; emulsion type and stability of immiscible liquid phases; and particle size and size distribution of nucleated liquids and solids. Practical applications range from the control of foaming in distillation and fractionation towers to the design of experiments under free-fall conditions to produce states of matter in a degree of fine dispersion otherwise unattainable. Capillary activity of solutions makes itself manifest in many different effects, such as foaming, emulsification, spreading, wetting, etc., but its fundamental characteristic is the lowering of surface and interfacial tensions. An attempt to read capillary-active phenomena into phase diagrams begins, therefore, with the measurement of the surface tensions of conjugate solutions and of the interfacial tension between them. From such measurements can be determined spreading coefficients, adhesion tensions, and all the other related functions that measure the extent of molecular attraction at the interface between immiscible liquidsU5
Spreading Coefficients and Foam Inhibition The foaminess of a homogeneous solution increases at compositions near the plait point of ternary systems.2 At compositions within the heterogeneous region of the phase diagram, however, foam was sometimes found to be inhibited.3 This effect, when it occurs, may be attributed to the spreading action of the dispersed immiscible phase over the surface of its conjugate, which action provides the mechanical shock by which liquid films are ruptureda6By this hypothesis, if droplets of the dispersed phase can spread on the foamable matrix of their conjugate solution, they are able to destroy liquid films. A slightly different mechanism was put forth by Robinson and Woods? who showed that the mere action of an insoluble droplet entering the surface is sometimes all that is required to rupture the liquid film. The “rupture coefficient” of Robinson and Woods, which has also been called the “entering coefficient”, is the same, except for sign, as the spreading coefficient of the matrix on the dispersed phase, as a droplet entering the surface from the interior is exactly the same phenomenon as the dewetting of the droplet by its matrix.8 Ross and Nishioka have shown,g for example, that the dewetting of hydrophobic particles of silica is the functioning mechanism of the silicone antifoams of commerce, which are dispersions of
The Journal of Physical Chemistry, Vol. 83,NO. 17, 1979 2227
silica in silicone oil. Other examples of foam-inhibiting action depend just as clearly on the spreading of the droplet of insoluble liquid agent;. Which of the two mechanisms operates may well depend on the particle size of the dispersed phase; dewetting of a relatively large solid particle or liquid droplet may be effective to rupture the liquid film, whereas a smaller droplet would require the additional force conferred by the action of spreading, and a still smaller droplet might spread without causing the liquid film to rupture. The present work uses spreading and dewetting coefficients, calculated from isothermal data for surface and interfacial tensions, of two ternary systems to correlate foaminess, foam inhibition, and other capillary phenomena in different compositions with their location on the phase diagram. Spreading coefficientslO are defined as follows: S (0 on a) = (T, - B, - pint (1) S (a on 0 ) = u, - u, - uint (2) where S (0 on a) is the spreading coefficient for the organic phase spreading on the aqueous phase, S (a on 0 ) is the spreading coefficient for the aqueous phase spreading on the organic phase, u, is the surface tension of the organic phase, u, is the surface tension of the aqueous phase, and pint is the interfacial tension, A positive value of S predicts that spreading of one phase on the surface of the other will occur spontaneously; if negative, that spreading will not occur. Dewetting coefficients (the “rupture coefficients” of Robinson and Woods) are defined as D (0 by a) = (T, - u, qnt (3) D (a by 0 ) = (T, - (T, uint (4) A positive value of D predicts that a droplet of one phase will be dewetted by its matrix phase, and a negative value predicts that it will not be so dewetted. From these equations D ( 0 by a) - S (0 on a) = 2uht (5) D (a by 0 ) - S (a on 0 ) = 2uint (6) Since the right-hand sides of eq 5 and 6 are always positive we can distinguish the following three cases: (i) D is negative and S is negative. The droplet is not dewetted, or in the terminology of Robinson and Woods it cannot rupture the liquid film because it does not enter the surface of the matrix. If it does not enter, it cannot spread on the surface. (ii) D is positive and S is negative. The droplet is dewetted, hence enters, but does not spread on the surface. (iii) D is positive and S is positive. The droplet is dewetted, hence enters, and t h e n spreads on the surface. A fourth case, D negative and S positive, is both mathematically and physically impossible, as the droplet could not spread on a surface that does not allow it to enter. An important relation is obtained by combining eq 1and 2, giving S (0 on a) S (a on 0 ) = -2uint (7) Equation 7 tells us that both spreading coefficients may be negative; or one may be negative and the other positive; but that they cannot both be positive. This is pertinent to a series of dispersions of immiscible conjugate solutions that vary only in the relative amounts of each phase. Within a narrow range of compositions described by points on the tie line between two conjugate solutions, a morphological change occurs: what was previously the dis-
+ +
+
2228
The Journal of Physical Chernistty, Vol. 83, No. 17, 1979
S.Ross and R. E. Patterson
5.1
i
'O
0
--20 lo;
t
e S(a o
ON 0)
s (0 ON a )
$11 E
/
v
-'I
'-401
'-601
-80
- 60
-70L 0
I
I
0.2
I
,
,
I
0.4
0.6
I
, 0.8
0
s(0 ON a )
:/
f
-100
I
1.0
1 - 17 LO
Flgure 1. Variation of spreading coefficients as a function of tie line length in the system benzene-water-ethanol at 20 O C .
persed phase becomes the continuous phase, and vice versa. The spreading coefficient that determines foam inhibition, i.e., the ability of the dispersed droplet to spread over its matrix, switches at this inversion point from one to the other, e.g., from S ( 0 on a) to S (a on 0). If the former is positive, the latter must be negative, which means that the new type of dispersion will not be defoamed because the dispersed phase is now unable to spread on the surface of its conjugate matrix.
Results Spreading Coefficients and Foaminess. We report elsewhere"J2 the purification of materials and the measurements of surface and interfacial tensions of conjugate solutions in two isothermal ternary systems: benzenewater-ethanol and n-hexane-water-ethanol, using the pendent-drop method,12J3which was selected after due consideration as the most suitable for the intended purpose. From these results, spreading coefficients can be calculated for pairs of conjugate solution throughout the miscibility gap in each system. These are presented graphically in Figures 1 and 2. The abscissae in these figures are a function of tie line length, L; Lo is defined as the length of the longest tie line, which in these systems is practically the line joining the vertices of the triangle. The function [l- (L/LO)] equals zero when no ethanol has been added to the other mutually saturated components, and equals unity at the plait point. Figure 1 shows a series of inversions of signs of the spreading coefficients, first a t point b and then at point c. In Figure 2 an inversion of signs occurs at point b and the slopes of the curves plainly suggest that a second point of inversion lies just beyond the last measured data point, in a region of compositions beyond the reach of our experimental techniques. Interestingly enough, the inversions in the two systems do not correspond sequentially;
l
0
0.1
~
I
0.2
1--
'
013
'
0.4
~
0.5
LO
Flgure 2. Variation of spreading coefficients as a function of tie line length in the system n-hexane-water-ethanol at 20 OC.
point b in Figure 1shows the water-rich phase beginning to spread on the organic phase, while point b in Figure 2 shows the organic phase beginning to spread on the water-rich phase. This reversal of the sequence of the inversions in the two systems has implications for capillary behavior, determining that various effects occur a t compositions in quite different regions of the phase diagrams of the two systems, although the two diagrams themselves closely resemble one another. Extrapolations or predictions of behavior based on the assumption of similarities between systems have, therefore, to be made cautiously. The foaminess of a solution was tested by bubbling nitrogen gas through it. We found that if one of the conjugate solutions sustains a relatively stable foam, its conjugate phase acts as a foam inhibitor if and only if its spreading coefficient on the medium is positive (Le., case iii). We did not find it to be sufficient in these systems merely for the dewetting coefficient to be positive (i.e., case ii).l2 Foamable and Foam-Inhibited Compositions. The potential regions of foaming and defoaming are mapped on the phase diagrams in Figures 3 and 4. Tie lines bb' and cc' in Figure 3 correspond to points b and c of Figure 1. All the tie lines between P (the plait point) and CC' represent conjugate solutions that have a positive S (0 on a), and all the tie lines from cc' to bb' represent conjugate solutions that have a positive S (a on 0). The tie lines from bb' to the base line represent conjugate solutions in which both spreading coefficients are negative. Our foam tests show that solutions represented along the line Pc in Figure 3 are defoamed by droplets of composition represented along the line Pc'. That corresponds to a region where S ( 0 on a) is positive. Solutions represented along the line c'b' are defoamed by droplets of composition represented
/
~
Capillary Effects in Ternary Systems
The Journal of Physical Chemistry, Vol. 83, No. 17, 1979 2229
ETHANOL
I WAl E R
-
\
1
I
BENZENE
Flgure 3. Capillary effects in the system benzene-water-ethanol at 20 ‘C. Shaded areas are regions of innate inhibition of foaminess.
F.I.C. = foam-inhibiting conjugate.
ETHANOL
A
WATER
I
“-HEXANE
Capillary eftects in the system n-hexane-water-ethanol at 20 ‘C. Shaded area is a region of innate inhibition of foaminess. F.I.C. = foaminhibiting conjugate. Positions of P and I are estimated. Flgure 4.
along the line cb, a region where S (a on 0)is positive. Tie lines between bb’ and the base line represent mixtures of solutions where no defoaming occurs. Figure 4 represents potential foaming and defoaming compositionsin the n-hexane-water-ethanol system. Not shown on Figure 4, but reasonably suspected to exist from circumstantial or indirect evidence, is a tie line cc’ very near point P, correspondingto a pair of conjugate solutions whose two spreading coefficients are very close to zero. Conjugate pairs represented by tie lines closer to P than cc’ have a positive S (a on 0);those represented by tie lines between cc’ and bb’ have a positive S (0on a); and those represented by tie lines from bb’ to the base line have two negative spreading coefficients. Our foam tests show that solutions represented along the line cb are defoamed by droplets of their conjugate composition represented along the line c’b’, a region where S (0 on a) is positive. The limitations of our experimental technique prevented us from doing foam tests on any solutions along the bypothetical line Pc‘; these presumably would be defoamed by their conjugates. Tie lines between bh’ and the base line represent solutions where no defoaming occurs. Figures 3 and 4 represent areas of potential foaming and defoaming, as predicted by the spreading coefficients. Not all one-phase compositions on the solubility curve are foamy solutions, no matter what may be the spreading coefficient of their conjugates.
Influence of Morphology. Two conjugate solutions form two different types of dispersion, conjugate a dispersed in 0,designated a/o; or Conjugate o dispersed in a, designated o/a. If the spreading coefficient of a is positive with respect to 0,then that of o is negative with respect to a, by eq 7. When a is dispersed in o it can, in this case, act as an inhibitor of the foaminess of o because it can spread on 0;but o dispersed in a does not inhibit the foaminess of a because it cannot spread on a. The type of the dispersion, whether a/o or o/a, depends on the relative amounts of the two phases. At a certain phase ratio, or rather, within a more or less narrow range of compositions, a morphological change occurs: what was previously the dispersed phase becomes the continuous phase and vice versa. The conjugate with S > 0, which when dispersed acts as foam inhibitor, when continuous becomes a foamable (or perhaps unfoamable but not foam-inhibited) matrix; the conjugate with S < 0, which when it is the continuous phase is defoamed by its conjugate, when dispersed cannot reciprocate the defoaming action. The foam behavior of the heterogeneous system therefore switches from foam-inhibited to foamable a t the composition at which the dispersion inverts. In Figure 3, for the benzene-water-ethanol system, the locus of the compositions of the inversion points is shown as the curve PI. The shaded areas denote regions of composition in which innate inhibition of foaminess occurs in this system. Qualitative foam tests were made by means of a simple apparatus, consisting of a glass tube 20 cm long and 2.2 cm in diameter, with a porous glass frit separating the upper from the lower half. About 10 mL of solution under test is placed in the upper half and nitrogen gas is admitted at the lower end. The flow rate of the gas is adjusted to produce a convenient foam height above the liquid surface (in cases where the liquid could support even an evanescent foam). Foam-iuhihiting action is tested by adding a few drops of the agent to be tested. This apparatus was used in an air-conditioned room adjusted to 20.0 i 0.5 ‘C. The conjugate phases to be tested are carefully separated after removal from a constant-temperature bath and tested independently. Special precautions are required at compositionsnear the plait point. In this region of compositions the temperature sensitivity is much greater: small changes of temperature cause the system to shift between one-phase and two-phase regimes, as shown by the appearance or disappearanceof cloudiness. The results of these tests are reported in Tables I and 11. They verify the described relation between foam inhibition and emulsion inversion. In addition, quantitative determinations of the foam stabilities of benzene-water-ethanol compositions, using the foam meter described in ref 3, were made on compositions represented by tie line dd’ drawn through the two-phase region and extrapolated on hoth sides to include one-phase compositions. The tie line in question lies just below cc’ in Figure 3. These foam results are reported in Figure 5, which shows two compositions where foam stability changes abruptly to foam inhibition. One of these is a t composition d’, at which point droplets of the water-rich phase first appear in the benzene-rich medium; and the other is a t the composition represented by point I, the point of inversion of the liquid/liquid dispersion. These results point to the water-rich conjugate as the foam inhibitor; which agrees with the condition that the foam inhibitor have a positive spreading coefficient with respect to the medium. Another point of interest in the results presented in Figure 5 is that the solutions in the one-phase region, whether water-rich or benzene-rich, are foamable,
2230
The Journal of Physical Chemistry, Vol. 83, No. 17, 7979
S. Ross and R.
E. Patterson
TABLE I: Equilibrium Spreading Coefficients, Foaminess, and Defoaming Action in the System Benzene-Water-Ethanol at 20.0 C O
aqueous phase tie line no.a 0 1 2 3
4 5 6 7 8
S (0on a), mN/m -0.62 - 1.99 -3.26 -4.00 -3.52 - 2.62 -0.75 0.05 0.47
organic phase
foaminess
defoamed by conjugate?
S (a on o), mN/m
foaminess
defoamed by conjugate?
none medium medium medium medium high medium none medium
no no no no no no
-67.54 -29.81 -10.80 -3.52 - 0.60 0.74 0.29 -0.25 -0.49
none none none none none none medium very low medium
Yes no no
Yes
Compositions of the conjugate solutions corresponding to these tie line numbers are reported in ref 11. TABLE 11: Equilibrium Spreading Coefficients, Foaminess, and Defoaming Action in the System n-Hexane-Water-Ethanol at 20.0 C aqueous phase S
tie line no.a
(0on a), mN/m
- 1.5
3.9 3.2 3.8 3.1 4.2 4.2 3.4 0.1 a
organic phase
foaminess
defoamed by conjugate?
none high high high high medium medium low high
Yes Yes Yes Yes Yes Yes Yes Yes
S (a on a),
mN/m
foaminess
defoamed by conjugate?
-98.0 -33.3 -22.8 - 19.5 -16.4 -14.7 -11.8 - 6.4 -0.3
none none none none none none none none very low
no
Compositions of the conjugate solutions corresponding to these tie line numbers are reported in ref 11.
thus testifying to capillary activity in solutions represented by those two regions of the phase diagram. In the water-rich solutions the organic solute is surface active, and in the benzene-rich solutions the water is surface active. Figure 4, for the n-hexane-water-ethanol system, shows that the region of innate inhibition of foaminess extends over a large area of the two-phase portion of the diagram. Curve PI, the locus of inversion-point compositions, is only approximate, as no quantitative location of this property was made for this phase diagram.
i‘t
I
Explaining Some Hitherto Unexplained Observations Some prior empirical observations find their rationale in the light of the present results. Forty years ago SasakP reported what he described as “a rare and interesting example of a foam-nonfoam system”. He worked with a stoppered test tube containing air, 2.33 cm3of acetic acid, 3.79 cm3of diethyl ether, and 3.88 cm3of water. He found that he could obtain either an a/o emulsion covered with a layer of foam or an o/a emulsion without foam by changing the motion of shaking from swinging to upand-down jerks. In another set of observations18with the ternary system benzene-water-acetic acid, the type of emulsion obtained and whether it foamed, again depended on the mode of shaking. When the system was composed of equal volumes of water and benzene and a small amount of acetic acid, the production of the o/a type of emulsion was accompanied with foam but the a/o type was made without foam. The change in the morphology was solely responsible for the change in the foam properties of the system. Sasaki’s results indicate that the inversion point of an emulsion, in terms of the ratio of oil to water, is not immutable but can be made to vary, probably within a narrow range of compositions near 50:50 by volume, by using different modes of shaking. Although no data are
T
I
I
I
1
I 0 0
0.2
0.6
0.4
WATERRICH
0.8
1.0 BENZENERICH
R Figure 5. Foam stability, 2, in seconds, of various compositions of benzene-water-ethanol at about 25 O C . R is the relative position on an extended tie line passing through points d and d’ on the bincdal curve. This tie line lies just below cc’ in Figure 3. R is defined as the length of the extended tie line from the point representing the overall composition to the water-ethanol side of the phase-diagram triangle divided by the entire length of the extended tie line from the waterethanol side to the benzene-ethanol side. 2 is defined as the ratio of the steady-state volume of foam to the flow rate of gas bubbled through the 1 i q ~ i d . lThe ~ phase diagrams for benzene-water-ethanol at 20’‘ and at 25 OC‘’ are nearly identical.
available for these systems from which to calculate spreading coefficients,the behavior observed is consonant with that of the emulsions between conjugate solutions reported in the present work, if the inversion point shown in Figure 5 be deemed able to be moved a little to the right or to the left of its reported position by changes in the
Capillary Effects in Ternary Systems
mode of mixing. Our interpretation of his observations on the presence or absence of foam is that, in one type of emulsion, the dispersed phase has a positive spreading coefficient with respect to the medium and therefore acts as a foam inhibitor; and, in the other type of emulsion,the other conjugate solution as the dispersate necessarily has a negative spreading coefficient with respect to the medium and therefore has no foam-inhibiting effect. Sasaki’s “foam-nonfoam systems” certainly retain their interest, as he commented, but are not as rare as he thought, now that the number of examples of that behavior is mounting. We now know how to locate them. All those on record are systems of two or three components with two immiscible liquid phases, and they occur in the vicinity of conditions that determine a critical point or a plait point. Other Capillary Effects Stability of Emulsions. While the ability of the dispersed phase to spread on the surface of its matrix phase is a condition for the inhibition of foam, the inverse ability, i.e., that of the matrix phase to spread on the surface of the dispersed liquid phase, is a condition that promotes the stability of an emulsion, because it means that the adhesion between the droplet and its matrix is greater than the cohesion of the matrix. Coalescence of droplets is resisted, therefore, by the presence of the matrix phase, which does not withdraw from the space between the droplets on their close approach. The compositions denoted by the unshaded areas above tie line bb’, which are bounded by the capillary-active conjugates (marked F.I.C. in Figures 3 and 4), are those in which emulsions are stabilized. Recently the existence of stable oil-continuous microemulsions composed of hexane, water, and 2propanol has been reported,l9 which confirms that this kind of capillary effect does occur in ternary systems, even in the absence of conventional amphipathic solutes. Critical-Point Wetting. The present results also demonstrate the analogue in a ternary system of the critical-point wetting4 in a binary system, the addition of cosolvent taking the place of an increase in temperature. As the plait point is approached, on addition of cosolvent, the interfacial tension and the difference in the surface tensions of conjugate pairs both tend to zero, but the tendency of the former is greater than that of the latter. It therefore follows that at compositions often well removed from that of the plait point 2 dint The range of compositions where this inequality holds begins with the tie line where l o a - do1
= dint This tie line is shown in Figure 3 as line bb’. Two other such tie lines almost coincide at cc’. Throughout the range of compositions from bb’ to very near cc’, S (a on 0 ) is positive; (shortly before cc’ it changes back to negative); and from cc’ to the plait point P, S (0 on a) is positive: i.e., in the first range, the conjugate a intrudes into the o/vapor surface and perfectly wets it; in the second range, the conjugate o intrudes into the a/vapor surface and perfectly wets it. In Figures 2 and 4,similar but not identical relations exist; in the range of compositionsfrom bb’ to very near cc’, the conjugate o intrudes into the a/vapor surface and perfectly wets it; in the compositions from cc’ to the plait point P, the conjugate a intrudes into the o/vapor surface and perfectly wets it. Surface-Active Solutions. Cahn4 has deduced on theoretical grounds that, in any two-phase mixture of liquids near their critical point, an adsorbed surface layer loa -
got
The Journal of Physical Chemistry, Vol. 83, No. 17,
7979 2231
of the spreading phase continues to exist under conditions where it is no longer stable as a bulk phase, which is another way of saying that he predicts capillary activity in a one-phase solution. In the benzene-water-ethanol system there are two points on the binodal curve, namely, c and b’, where the spreading phase would begin to persist as an adsorbed film into the one-phase region of the diagram. Positive evidence for capillary activity of one-phase solutions in the region of compositions near point b’ is given by the foaminess reported by Ross and N i ~ h i o k a ; ~ ~ ~ and in compositions near point c by the foaminess of those solutions observed by the present authors and reported previously in this paper. Foam stability, as explained by the Marangoni effect, depends on the presence of an adsorbed layer at the surface of a solution,8 and is useful here as an index of capillary activity. The capillary activity of these solutions, according to Cahn’s theory, would diminish steadily as their compositions, originating at that represented by point c, include larger concentrations of ethanol; and a similar diminution of capillary activity would begin at b’ and extend into the one-phase region with larger concentrations of ethanol. This behavior too is accurately reflected in the observed decrease of stability of foam as compositions change in these directions. The presence of two such loci of compositions of maximum surface activity at a given weight fraction of ethanol in the one-phase region of compositions can arise only from a double inversion of the signs of the two spreading coefficients at two compositions almost coinciding at c; and it means that the hydrocarbon is capillary active in the water-rich solutions and the water is capillary active in the organic-rich solutions. This behavior is demonstrated for the benzene-water-ethanol system both in Figure 3 and in Figure 5, and is apparently indicated as holding also for the n-hexane-water-ethanol system, although not shown in Figure 4 because it is not actually observed owing to the difficulty of obtaining measurements so close to the plait point. A precisely similar condition can be deduced to hold in some twocomponent systems, where each component A and B acts as a capillary-active solute in conjugate solutions at temperatures near critical, Le., where A is positively adsorbed at the surface of the B-rich conjugate and B is positively adsorbed at the surface of the A-rich conjugate. Figures 6 and 7 of ref 3 show two examples of binary systems where both conjugate solutions sustain relatively stable foams, a sure indicator that each has a capillaryactive solute. Capillary activity in the one-phase region is the last effect of the spreading conjugate, when its composition is altered just enough to make it soluble in its medium: spreading of an insoluble layer is then transformed into adsorption of a soluble film, both phenomena being manifestations of different kinds of capillary activity. A similar change of behavior is observed at a certain point in a series of capillary-active solutes such as an homologous series of fatty acids;” precisely where two adjacent members of the series change from insoluble to soluble in water, the two-phase property of spreading as an insoluble layer (or “perfect wetting” as the term is used by Cahn), manifested by the insoluble member, is transformed into that of surface activity, i.e., the spontaneous forming of a soluble adsorbed film at the surface or interface of the one-phase solution, manifested by the soluble member. The capillary activity of solutions as a function of composition is added to the other capillary effects that can be predicted from the spreading coefficients of conjugate solutions, when mapped on the phase diagram. These
2232
The Journal of Physical Chemistry, Vol. 83, No. 17, 1979
additions to the phase diagram locate regions of foaminess, of preferential wetting, of capillary activity, of innate inhibition of foam, of emulsion stability, and also show which conjugate phase appears as the matrix of the liquid/liquid dispersion. Acknowledgment. The data represented by line PI in Figure 3 were obtained by Mr. R. E. Kornbrekke. The data shown in Figure 5 were obtained by Miss Carrie Woodcock.
References and Notes (1) I. Langmuir in "Coiioid Symposium Monograph", Vol. 3, H. N. Holmes, Ed., Chemical Catalog Co., New York, N.Y., 1925, p 62. (2) S. Ross and G. Nishioka, J . Phys. Chem., 79, 1561 (1975). (3) S. Ross and G. Nishioka in "Foams", R. J. Akers, Ed., Academic Press, New York, N.Y., 1976, pp 17-31. (4) J. W. Cahn, J . Chem. Phys., 66, 3667 (1977). (5) S. Ross, J . Colloid Interface Sci., 42, 52 (1973).
Ulmius et ai.
(6)S. Ross, J. Phys. Colloid Chem., 54, 429 (1950); W. E. Ewers and K. L. Sutherland, Aust. J. Sci. Res., Ser. A , 5, 697 (1952). (7) J. V. Robinson and W. W. Woods, J. SOC.Chem. Ind. (London), 67, 361 (1948). (8) S. Ross, Chem. Eng. Prog., 63 (9), 41 (1967). (9) S.Ross and G. Nishioka, J. Colloid Interface Sci., 65, 216 (1978). (10) W. D. Harkins, J . Chem. Phys., 9, 552 (1941). (11) S.Rossand R. E. Patterson, J . Chem. Eng. Data, 24, 111 (1979). (12) R. E. Patterson, Ph.D. Dissertation, Rensseiaer Polytechnic Institute, Troy, N.Y., 1978. (13) R. E. Patterson and S.Ross, Surface Sci., 81, 451 (1979). (14) J. J. Bikerman, Trans. Faraday Soc., 34, 634 (1938). (15) Landolt-Bornstein, "Zahlenwerte und Functionen", Vol. 2, Part 2, Section C, Springer-Veriag, Berlin, 1964, p 610. (16) W. D. Bancroft arid S.C. Hubard, J. Am. Chem. Soc., 64, 347 (1942). (17) T. Sasaki, Bull. Chem. Soc. Jpn., 14, 63 (1939). (18) T. Sasaki and S. Okazaki, Kolloid-Z., 159, 11 (1958). (19) G. D. Smith, C. E. Doneien, and R. E. Barden, J. Colloid Interface Sci., 60, 488 (1977). (20) N. K. Adam, "The Physics and Chemistry of Surfaces", 3rd ed,Oxford University Press, London, 1941, pp 117-1 18 (see also Dover reprint, New York, 1968).
Viscoelasticity in Surfactant Solutions. Characteristics of the Micellar Aggregates and the Formation of Periodic Colloidal Structures Jan Ulmlus, HAkan Wennerstrom,* Lennart B.-A. Johansson, Goran Lindblom, Division of Physical Chemistry 2, Chemical Center, 5-220 07 Lund, Sweden
and Signe Gravsholt Fysisk-Kemisk Instituf, The Technical University of Denmark, DTH 206, DK 2800 Lyngby, Denmark (Received November 6, 1978) Publicatlon costs assisted by the University of Lund
The properties of viscoelastic dilute aqueous solutions containing the hexadecyltrimethylammonium cation (CTA+) are investigated by monitoring the proton magnetic resonance spectrum and the linear dichroism induced in a shear gradient. It is found that the viscoelastic behavior of the dilute solutions correlates with the formation of rod-shaped micellar aggregates. This correlation is manifested both with respect to changes in composition and in temperature. It is suggested that the behavior of the solutions is caused by the presence of a periodic colloidal structure formed due to the repulsive force between the aggregates.
Introduction Aqueous solutions of some amphiphilic molecules show a striking viscoelastic behavi0r.l The viscoelasticity can, for example, be seen by simply swirling the solution and visually observing the recoil of air bubbles trapped in the solution after the swirling is stopped. The viscoelasticity is manifested in a number of other properties as, for example, a non-Newtonian viscous behavior2 and a flow-induced optical anisotropy as illustrated in Figure 1.2b Although viscoelasticity3has been observed for a number of aqueous solutions containing ionic amphiphiles (see ref 1 and references cited therein), the property is of a rather rare occurrence. Furthermore, the chemical difference between solutions which show viscoelasticity and those which do not is sometimes smal1.l Usually the viscoelastic behavior is observed when a third component is added to a rather dilute (
