J. Phys. Chem. B 2007, 111, 589-597
589
Insights into Hofmeister Mechanisms: Anion and Degassing Effects on the Cloud Point of Dioctanoylphosphatidylcholine/Water Systems Marco Lagi, Pierandrea Lo Nostro,* Emiliano Fratini, Barry W. Ninham,† and Piero Baglioni Department of Chemistry and CSGI, UniVersity of Florence, Via della Lastruccia 3, 50019 Sesto Fiorentino (Firenze), Italy ReceiVed: September 5, 2006; In Final Form: October 13, 2006
Water dispersions of dioctanoylphosphatidylcholine (diC8PC) exhibit upper consolute curves. How they are affected by some salts and other additives (D2O, urea) has already been explored and the phase separation has been interpreted within the framework of the Blankschtein-Thurston-Benedek (BTB) model. This deduces the chemical potential gain in micellar growth (∆µ) and the intermicellar interaction coefficient (C) as fitting parameters from the coexistence curves. But, the specific mechanisms that drive such phenomena have remained obscure. To identify these mechanisms, we investigate the effects of a range of anions on the cloud points of diC8PC/H2O systems and extract the phenomenological parameters of the BTB model. We show how these parameters, for micellar growth, i.e., surface, intramolecular interactions, and for interaggregate interactions, i.e., bulk solute effects, are connected to anionic polarizabilities. Nonelectrostatic (NES) quantum mechanical fluctuation (Lifshitz or dispersion) forces missing from conventional theories are then shown to regulate lyotropic Hofmeister effects, both explicitly and implicitly.
1. Introduction and Background Specific ion, or Hofmeister, effects are found in a plethora of different systems. Recent reviews illustrate the near universality of such phenomena.1 The term specific ion effects is a misnomer. It is used to encompass the fact that different electrolytes induce systematic changes in the properties of bulk solutions, interfaces, and polymeric and colloidal systems that are not explained by classical theories of electrolytes. The effects can differ, depending on the system, pH, and buffer, as well as the nature and concentration of the background electrolyte. To put our investigation within a wider context, we need to rehearse the nature of the problem and some of what is presently known and quantified. Typical Hofmeister phenomena show up for even moderate salt concentrations (above 10-2 M). These cannot be explained by electrostatic theories, with or without continuum solvent model approximations. Something then seems to be missing from current theories. Recent work has proven that, even within the confines of the primitive (continuum solvent) model approximation, part of the problem can be traced back to the inconsistent treatment of nonelectrostatic (NES) forces that were thought to be irrelevant. These are included in the Lifshitz theory of interactions but seemed to be negligible compared with electrostatic effects. But, the theory of interactions treats electrostatic interactions in a nonlinear theory, and it treats the NES interactions in a linear theory. It is thermodynamically inconsistent. Ion-substrate interactions can be extracted from the theory, and then, when treated at the same level as electrostatics to account for adsorption, the ion specificity shows up very strongly.2 Its extensions include also self-free energies * Corresponding author. Fax: +39 (055) 457-3036. E-mail: pln@ csgi.unifi.it. Internet: http://www.csgi.unifi.it. † Permanent address: Department of Applied Mathematics, Research School of Physical Sciences and Engineering, Institute of Advanced Studies, Australian National University, Canberra, Australia 0200.
of ions (hydration, salting-in, salting-out ions)2-4 and ion fluctuation forces.5 That theory takes into account contributions from all the frequency-dependent many body electrodynamic fluctuation forces. They are accessible in principle from bulk dielectric susceptibilities as a function of frequency. The familiar Keesom (orientation), Debye (induction), and London (dispersion) forces between molecules are special cases valid only in the limiting case of dilute gases. But, these NES forces are operative also in ionic solutions. These forces do act on and between ions and are responsible in the first approximation for hydration and hydration forces and contribute strongly to specific ion adsorption at interfaces. They depend on the ionization potentials and ion polarizabilities in solution, parameters that directly relate to the specific chemical nature (electronic configuration) of ions and molecules. For anions, there is much greater variation in excess polarizabilities than for cations. So one expects variations in such specific ion effects to show up more strongly with anions, as Hofmeister showed so long ago.1 The arguments above are acceptable for high electrolyte concentrations where, say at 1 M (Debye length ≈ 3 Å), electrostatic forces of interaction are strongly screened. Hofmeister effects might then be ascribed to water structure. Hofmeister himself remained bemused about whether the effects he observed with protein precipitation should be attributed to surface effects of specific anion or cation adsorption, or to bulk effects, what he termed the “withdrawing power of salts”. The latter opinion would still be favored, especially as the activity coefficients of bulk electrolytes follow the Hofmeister series at high concentrations. But, there remain some further problems. Some examples are these: The Hofmeister series shows up in surface tension measurements, inexplicable unless one allows both the NES forces that give rise to positive or negative adsorption excesses and the inhomogeneous profile at the water-air or water-oil interface.4-6 In pH measurements via glass electrodes with buffers, even at
10.1021/jp065769y CCC: $37.00 © 2007 American Chemical Society Published on Web 12/23/2006
590 J. Phys. Chem. B, Vol. 111, No. 3, 2007
Lagi et al.
millimolar buffer and high salt, the apparent pH can follow a direct or reverse Hofmeister series.7 The same is true for DNA restriction enzyme interactions for concentrations of salt greater than 0.15 M.8 Again with protein precipitation, the anion effectiveness follows a direct or reverse sequence depending on if the protein is above or below the point of zero charge (IP).9 So (counterintuitive) surface ion adsorption driven by NES forces is implicated, along with bulk solution effects. We shall see that both are involved. 2. Effects of Solutes and Ions on Cloud Points With this as background, we have chosen to study Hofmeister effects, the systematic variation of cloud points as a function of the anion in the zwitterionic surfactant diC8PC/H2O system. The salt concentration is chosen to be quite low, 0.05 M, so that modified Debye-Huckel theory gives a reasonable description of activities. At first sight, bulk water activity ought not to be the dominating factor. The occurrence of cloud points (CPs)si.e., the onset of turbidity upon addition of salts or temperature changessreflects phase separation that occurs in surfactant aqueous dispersions. By plotting the cloud point temperature as a function of the surfactant mole fraction, one obtains the coexistence curve that in the graph separates a monophasic micellar top region from a biphasic bottom region. Here, typically, a micellar concentrated phase is in equilibrium with a dilute micellar phase. It has been known for a long time that salts may have a relevant effect on the cloud point temperature and phase behavior of nonionic and zwitterionic surfactants.10 The salt is treated as part of the solvent medium, and the surfactant-salt-water is treated as a pseudobinary solution. The salt effects have been usually ascribed to hydration/dehydration processes of the polar head groups caused by the different electrolytes. Another interpretation relies on the surfactant solubility change due to a modification of the micellar aggregation number induced by the ions.11,12 The lyotropic, specific ion effect induced by electrolytes on the cloud point temperature of nonionic and zwitterionic amphiphiles is one of the best known examples of Hofmeister phenomena. The most typical cases are those of H(CH2)i(OCH2CH2)jOH (CiEj) nonionic amphiphiles,13 octoxynol 9 (Triton X-100),14 and of zwitterionic lecithins such as dioctanoylphosphatidylcholine (diC8PC). The literature offers plenty of papers on this subject.15-17 Some 20 years ago, Blankschtein, Thurston, and Benedek proposed a thermodynamic model that allows two phenomenological parameters, C and ∆µ, to be extracted from the experimental coexistence curve.18-20 The parameter C characterizes the magnitude of the effective attractive intermicellar free energy, and ∆µ determines the free energy gain associated with micellar growth, that is, when micelles keep growing in size and polydispersity with increasing surfactant.20 At the critical point, Cc and ∆µc are related to the critical surfactant mole fraction xc and the critical temperature Tc through the following relationships:
γCc 5 ) 1 + (3γ - 2)xc kBTc 3 ∆µc 9 ) ln xc-5(3γ - 2)-2 kBTc 16
[
(1)
]
(2)
Here, γ ) Ωs/Ωw is the surfactant-to-water molar volume ratio and kB is the Boltzmann constant. This model was applied to diC8PC/water dispersions, where the lipid forms rodlike ag-
gregates above the cmc (xcmc ≈ 3.6 × 10-6 at 20 °C). The quantity ∆µ is defined as ∆µ ) ∆ - N0δ, where ∆ is the free energy gain of forming a minimum size micelle containing N0 monomers and δ represents the chemical potential difference between a monomer that enters into the cylindrical, central region of the micelle and one that enters the spherical ends.20 Lipid molecules in the central and terminal areas will possess different values of the packing parameter p ) Vi/(liaP) where Vi, li, and aP represent the volume of the hydrocarbon core, the length of the fully stretched hydrophobic chain, and the area per polar headgroup, respectively. For spherical objects, p < 1/ , and for rods, 1/ < p < 1/ , depending on the value of a .21 3 3 2 P The addition of other cosolutes will modify both the properties of the solvent and the hydration of surfactant molecules. Specific cosolute adsorption will also modify the hydration of molecules in the micelles and therefore intrasurfactant interactions. This results in a significant variation of the coexistence curve. The problem is then exactly the same as for the addition of salts (Hofmeister effects). Equations 1 and 2 hold, and the phenomenological parameters can be evaluated directly from the experimental cloud point curves in the presence of different additives and at different concentrations. For example, the addition of urea22 or heavy water23 to diC8PC/water dispersions presumably results in a perturbation of the solvent and micellar surface hydrogen bonding, that significantly alters the values of C and ∆µ. Similarly, the presence of electrolytes produces a change in the phase separation behavior. In a comprehensive study, Huang et al. investigated the diC8PC/water system in the presence of aqueous solutions of LiCl, NaCl, KCl, LiI, NaI, and KI.24 The addition of salt results in a complex behavior, with a shift of the coexistence curve to lower or larger xc and Tc values with respect to the salt-free lipid dispersion. The changes depend on the specific ion pair and on the salt concentration. The fitting of the curves was carried out for lipid mole fractions lower than 2 × 10-3. The first work on short-chain lecithins was by Tausk and co-workers and dates back to 1974.25-27 They studied salt effects on the cmc values for diC6PC, diC7PC, and diC8PC. For the latter lipid however, only NaCl and LiI were tested, and the change of both the anion and the cation did not help to clarify the roles played by each ionic species. Sarmiento et al. reported the effect of KCl and pH on the cmc of diC8PC.28 They found that for 1.2 < pH < 10.0, the cmc ranges between 0.18 and 0.28 mM, while the polar headgroup area aP varies between 86.8 and 65.1 Å2/molecule (from surface tension measurements). However, the use of five different buffer compositions injects considerable uncertainty into the real meaning of the results. In fact, since electrolytes do affect the solution and self-assembly behavior of diC8PC, it is unclear if the findings reflect just pH changes or are due to the presence of different electrolytes as well, or more likely the competition in ion adsorption of buffer anion with the background electrolyte.8 This behavior has been shown also by Huang et al. by testing the effects of KCl, KI, MgBr2, MgCl2, CaBr2, CaCl2, LaCl3, and CeCl3.29 These studies were based on experiments carried out with different ion pair electrolytes at different concentrations. Any effects due a series of anions, (coions for such systems) at the same concentration and with the same counterion, could not be extracted from the data.30 In the present study, we have explored the phase behavior diC8PC/H2O dispersions in the presence of a series of sodium
Insights into Hofmeister Mechanisms
J. Phys. Chem. B, Vol. 111, No. 3, 2007 591
monovalent salts, all at a fixed concentration of 0.05 M. We chose to keep the cation constant and to change the anion. This is because negative speciesswith a much wider spread in electron numbersswould be expected to experience NES or dispersion forces more strongly than cations.29 Using the Blankschtein-Thurston-Benedek (BTB) thermodynamic model, we evaluated the phenomenological parameters C and ∆µ that characterize the micellar growth and phase separation processes. Our results indicate that these two quantities do indeed strongly depend on the nature of the anion. In order to quantify such effects, we relate the experimental findings to the anionic polarizabilities in solution. In fact, this parameter is directly involved in NES forces, and it has been shown to be at play in all phenomena where a Hofmeister (lyotropic) series effect occurs.1,31-35 This is to be expected. A further inspection of the data implicates bulk water changes induced by an indirect consequence of anionic adsorption at the micellar interface. 3. Materials and Methods 1,2-Dioctanoyl-sn-glycero-3-phosphocholine (dioctanoyl phosphatidylcholine, diC8PC) was obtained from Avanti Polar Lipids Inc. (Birmingham, AL) in powder form. The purity was stated to be greater than 99% in diC8PC content, and the lipid was used without any further purification. Bidistilled water was purified with a Millipore water purification system to remove ionic impurities. NaF, NaCl, NaBr, NaI, NaOAc, NaN3, NaNO3, NaClO4, N(CH3)4OAc, N(CH3)4Cl, N(CH3)4Br, LiI, and choline chloride were purchased from Sigma-Aldrich-Fluka (Milan, Italy) in analytical grade and used as received. Samples were prepared by weighing the lipid and the solvent (either H2O or 0.05 M aqueous salt solution) directly in a glass tube. The lipid concentration was varied by diluting the sample with the same solvent. After the preparation, the sample was kept in a refrigerator at 4 °C for at least 10 h, until a clear and foam-free solution was formed. The tube was then placed in a temperature-controlled bath ((0.03 °C). The phase separation was induced by heating the sample and then cooling it to near the transition temperature T0. Above T0, the sample contained an optically clear, monophasic region. The temperature was slowly decreased until the solution became slightly cloudy with visual inspection. The temperature was then increased until the lipid dispersion cleared again. The up-and-down temperature cycle was repeated several times in order to obtain a good average value. We found that the cloud point (cooling cycle) was always about 0.5 °C below the reclarification point (heating cycle), and the true T0 value was assumed to be the average of the cloud point and of the reclarification temperature. Once T0 had been determined for a specific sample, the lipid mole fraction was changed by dilution and the heating-cooling cycle was then repeated on the new sample. For T < T0, the dispersion separates into upper, transparent, and low-viscosity phases, and a lower phase that is more turbid and more viscous. Ion chromatography experiments for Cl-, Br-, and NO3were conducted with a Dionex AG4A/AS4A (Dionex Srl, Rome, Italy) column, a 250 mL loop for loading the column, and a HCO3- (1.7 mM)/CO32- (1.8 mM) buffer was used as eluent. In the case of F-, we used a Dionex AS11 column, a preconcentrator, and a solution of Borax as eluent. In order to avoid contamination of the columns with diC8PC micelles, the lower phase was diluted 1:10 000, while the upper phase was diluted 1:100. Density measurements were carried out with an Anton Paar DMA 5000 instrument as a function of temperature.
Figure 1. CPK minimized model of diC8PC. Carbons are gray, oxygens are red, nitrogen is blue, and phosphorous is yellow (hydrogens have been omitted for clarity).
For water degassing, we used an ultravacuum setup connected to a turbomolecular pump. The aqueous lipid dispersion was transferred into an ultrahigh vacuum (UHV) vial, frozen in liquid nitrogen at atmospheric pressure, brought to 1 × 10-5 hPa, and then warmed up to room temperature. The freeze-pump-thaw cycle was repeated four times. The cloud point of the degassed sample was measured in the same way as specified above; then, air was let in, and the sample was equilibrated for 24 h and measured again in the re-aerated sample. Results Structural Parameters. The minimized Corey-PaulingKoltun (CPK) model of diC8PC is shown in Figure 1. According to the literature, the amount of hydration water bound to lecithins depends on the number of methyl groups attached to the nitrogen atom.36 The phosphate negative group is usually surrounded by 3-5 solvating water molecules.37 Density measurements were performed to determine the partial specific volume of the surfactant in the monomer (V1) and in the micellar state (Vm), using the following equations, respectively:
F ) F0 + (1 - V1F0)c
(3)
F ) Fcmc + (1 - VmFcmc)(c - cmc)
(4)
where cmc is the surfactant critical micellar concentration and F, F0, and Fcmc are the density of the solution at concentration c, that of the pure solvent, and that at the cmc. The measurements were carried out at 20, 30, 40, and 50 °C. The cmc values were obtained by plotting F vs c, as the intersection point of the two straight lines.38 The cmc values increase with temperature (see Table 1). At 20 °C, the ∆Gmicellization is -22.92 kJ/ mol. Table 1 reports also the cmc values (in mole fraction) for diC8PC/water dispersions between 20 and 50 °C measured with different methods. The values obtained from density measurements in this work (second column in Table 1) are compared to the data acquired from static QELS experiments by Huang39 and from surface tension measurements by Martinez-Landeira (third and fourth columns, respectively).40 The differences between the values derived from the different methods are used to evaluate this parameter. Then, eqs 3 and 4 were used to
592 J. Phys. Chem. B, Vol. 111, No. 3, 2007
Lagi et al.
TABLE 1: Values of cmc and Ωs,1 and Ωs,m (in Å3/molecule) of diC8PC in Water at Different Temperatures, Obtained from Density Measurements xcmc T (°C) 20 30 40 50
this work
ref 39a
10-6
10-6
ref 40b 10-6
1.42 × 3.80 × 3.78 × 1.83 × 10-6 5.73 × 10-6 3.24 × 10-6 2.14 × 10-6 8.43 × 10-6 3.06 × 10-6 2.74 × 10-6 12.10 × 10-6
Ωs,1 Ωs,m 646 663 686 715
759 773 783 790
Ωw
γ
29.97 30.04 30.15 30.27
25.3 25.7 26.0 26.1
a
From static light-scattering experiments. b From surface tension measurements.
calculate V1 and Vm, respectively, from which the molar volumes of the surfactant in the monomer and in the micellar states were obtained (Ωs,1 and Ωs,m, respectively). Table 1 shows the results. The calculation of γ as Ωs/Ωw indicates that the effect of temperature on this parameter is quite small. Anion Effect on Phase Separation. Our results (see Table 2 and Figure 2) clearly indicate a specific ion effect that affects the value of the cloud point (CP) in diC8PC/water dispersions. Figure 2 shows the different coexistence curves obtained by plotting the cloud point as a function of the lipid mole fraction in the presence of 0.05 M salt solutions. The curves were fitted according to the BTB model (see Figure 2), and the critical consolute temperatures, Tc, are listed in Table 2. With a fixed cation concentration (Na+) and the salt concentration (0.05 M) held constant, we can directly relate the change in cloud point to effects induced by each anion in solution. All anions decrease the CP with respect to pure water according to the series:
(H2O) > F- > CH3COO- > Cl- > Br- > N3- > NO3- > Ias shown in Figure 3. It is interesting to compare these effects to those of heavy water23 and of urea at 0.5 M.22 These show a significant increment in, and lowering of, the cloud point, respectively. Those results were there ascribed to dielectric constant changes in the solvent and, therefore, to perturbation of the hydrogen-bonding network in pure water. By applying the BTB theory to diC8PC dispersions in the presence of these 0.05 M salt solutions, we extracted the critical values for the fitting parameters ∆µ/kB, C/kB, and γ (in the fitting procedure, we allowed γ to vary around its value of 26, which is approximately valid for aqueous dispersions of the lipid).22,23 The value of the critical lipid mole fraction, xc, and that of the aggregation number, gc, can be calculated, once the fitting parameters have been determined, through eqs 5 and 6:23,41,42
xc )
[
( )] ( )]
16(3γ - 2)2 ∆µ exp 9 kBTc
[
gc ≈ 55 + 2 (xc - xcmc) exp
5
∆µ kBTc
1/2
(5) (6)
Since the critical micellar concentration of the lipid in the different salt solutions (xcmc) is always much lower than xc (about 2 orders of magnitude), we approximated xc - xcmc as xc. Table 2 shows that Tc is always lower than the pure waterlecithin system for all anions (i.e., the solubility of the lipid increases in the presence of salt), xc increases slightly with the exception of I-, and micelles are smaller (lower values of gc) except for I-. The fitting parameters show an interesting trend. The lipid-to-water molecular volume ratio, γ, remains practically
constant within 2% with respect to the pure water dispersion, except for N3- and I-, where it increases. The free energy gain associated with micellar growth, ∆µ, decreases regularly from fluoride to iodide, while the parameter C decreases in a more consistent way, indicating smaller interactions between micelles in the presence of I-. In previous papers, Huang et al. reported different values for ∆µ/kB and for the product γ‚C/kB.24,39 In particular, assuming a molar volume for diC8PC of 900 Å3 (from its density value), they calculated γ as 30, and they attributed to ∆µ/kB a value of 8420, 8331, and 8485 K in the case of pure water, 47 mM KI, and 480 mM KCl solutions, respectively. In the present paper, we adopted γ ) 26 for diC8PC in pure water (obtained from previous SAXS experiments) and obtained different values.23 This is why we let γ change during the fitting procedure. In any case, although the particular values of the fitting parameters might differ, the trend reported by Huang for KI and KCl confirms what we found in this work. It is interesting to compare the effect of salts on aqueous dispersion of the zwitterionic surfactant diC8PC to that reported in the literature for a classic nonionic amphiphile, Triton X100 (see Table 2, last column).13 The trend for the two series of data is inverted, i.e., the ions that increase the CP temperature for Triton X100 cause a decrease in the case of the zwitterionic short-chain phospholipid. The inversion reflects the typical opposite behavior of zwitterionic and nonionic surfactants toward temperature changes. The concentration of some anions has been measured by ion chromatography in the two separated phases. Table 3 shows the concentration of fluoride, chloride, bromide, and nitrate ions after equilibration at constant temperature of the separated phases. Also, the ratio between the volume of the two phases is shown. The upper/lower concentration ratio and the upper/ lower volume ratio decrease regularly from fluoride to nitrate. In particular, fluoride ions partition almost evenly between the two phases, while bromide is two times as concentrated in the bottom phase. This finding supports the hypothesis of a strong interaction between the anions and the lipid molecules, which is specific for each electrolyte. Table 4 shows the cloud point temperature of diC8PC/water dispersions in the presence of different salts (0.05 M) at a constant lipid mole fraction (5 × 10-4). The data indicate that Tcp always decreases with respect to pure water, but not for deuterium oxide. The substitution of Na+ with N(CH3)4+ (a more hydrophobic cation) results in an interesting inversion; in fact in the case of acetates, Tcp with N(CH3)4+ is 1 °C higher than with sodium, for chlorides the two cations are almost equivalent, but for bromides N(CH3)4+ is 1.7 °C lower than Na+. Choline hydrochloride shows a 1.4 °C decrement in Tcp with respect to NaCl. The variation of the anion, keeping Na+ as the counterion and the lipid mole fraction at 5 × 10-4, parallels the typical direct Hofmeister sequence:
F- > Cl- > Br- > N3- > NO3- > I- > ClO4As expected, Li+ increases Tcp of 8 °C with respect to sodium (as iodides). These particular results show that the substitution of sodium with the more hydrophobic tetramethylammonium cation leads to Tcp values that do not simply parallel those obtained with sodium salts. This evidence strongly indicates that the effect induced by a specific anion depends on the counterion as well. This inference shows that in Hofmeister studies the choice of the investigated electrolytes is crucial, and no immediate comparison can be made on the basis of different
Insights into Hofmeister Mechanisms
J. Phys. Chem. B, Vol. 111, No. 3, 2007 593
TABLE 2: Values of ∆µ/kB (K), γ, and C/kB (K) at the Critical Point, Critical Temperature (Tc, K), and Critical Mole Fraction (xc), Aggregation Number at the Critical Point (gc) for diC8PC in Different Sodium Salt Solutions (0.05 M), and Mole Ratio of the Anion between the Upper and the Lower Phasea anion b
(water) fluoride acetate chloride bromide azide nitrate iodide urea 0.5 Md
∆µ/kB ( 20
γ ( 0.1
C/kB ( 0.2
Tc ( 0.5
xc × 10-4 ( 1 × 10-5
gc × 104
8900 8588 8580 8550 8515 8490 8520 8400 8250
26.0 25.5 25.4 25.4 25.7 27.3 25.6 30.1 45
13.25 13.67 13.70 13.57 13.22 12.36 13.10 10.69 7.33
319.2 318.5 317.6 315.2 312.4 310.9 310.1 298.6 297.8
5.97 7.23 7.17 7.01 6.80 6.56 6.51 5.35 4.94
5.7 4.1 3.9 4.1 4.3 4.4 4.7 5.9 4.6
up/down
N
-∆Ghydr
νs
σ
R
0.90
4.8 8.0 10.0 11.3
465 365 340 315 295 300 275
-2.35 39.7 16.6 23.5 28.3 27.7 35.0
3.60 2.64 1.63 1.31
1.31 3.77 5.07
0 -19.1 -13.7 -8.6 -3.0
1.18 1.02
4.47 7.41
-3.0 7.9
0.74 0.49 0.71
11.6 12.5
∆TTritonc
a Lyotropic number (N), Gibbs free energy of hydration (∆G -1 3 -1 hydr, kJ‚mol ), partial molar volume (νs, cm ‚mol ), molar surface tension increment σ (µN·m2‚mol-1), anion effective polarizability in solution (R, Å3), and cloud point temperature shift for Triton X-100 with respect to water (salt concentration 0.50 mol‚kg-1) for each anion.The values of N, -∆Ghydr, νs, σ, and R were taken from ref 1. b Values taken from ref 23. c Values taken from ref 14. d Values taken from ref 22.
TABLE 3: Concentration (in mM) of Fluoride, Chloride, Bromide, and Nitrate in the Upper and Lower Phase and Volume Ratio of the Up and Down Phases anion
upper phase (mM)
lower phase (mM)
up/down
up/down volume ratio
fluoride chloride bromide nitrate
49.9 51.3 43.5 32.2
55.3 69.4 87.6 45.1
0.90 0.74 0.50 0.71
2.43 2.16 1.98 1.92
TABLE 4: Cloud Point Temperature (Tcp, K) of diC8PC/ Water Mixtures in the Presence of Different 0.05 M Salt Solutions, at a Lecithin Mole Fraction (xs) of 5 × 10-4 salt a
Figure 2. Coexistence curves for diC8PC/H2O dispersions in the presence of different 0.05 M sodium salts: fluoride (0), acetate (b), chloride (2), bromide (1), nitrate (9), azide (O), and iodide ([). Lines represent the fitting curves obtained by applying the BTB model.
Figure 3. Hystogram showing the Tc values for diC8PC/water dispersions in the presence of electrolytes. The data for diC8PC/D2O and diC8PC/urea 0.5 M have been extracted from refs 23 and 22, respectively.
ion pairs. Our results on diC8PC/H2O systems prove that whatever the microscopic mechanism of Hofmeister phenomena, both ionic species take part in the molecular interactions that determine the lyotropic effect. Correlations with Other Physicochemical Parameters. In previous papers, we have shown that for each case where a Hofmeister effect is observed, the whole phenomenon can be
(H2O) NaF NaOAc NaCl NaBr NaN3 NaNO3 NaClO4 a
Tcp ( 0.3
salt
Tcp ( 0.3
318.3 318.1 317.2 314.6 312.0 310.6 309.8 274.6
NaI LiI N(CH3)4OAc N(CH3)4Cl N(CH3)4Br choline Cl (D2O)a urea 0.5 Mb
298.4 306.4 318.2 314.4 310.3 313.2 331.8 297.6
From ref 23. b From ref 22.
quantified and discussed in terms of a specific variable and that the variation of this quantity with the different electrolytes is well represented by some physicochemical properties that are specific for each ion. For example, by measuring the kinetics of formation of host-guest inclusion complexes (polypseudorotaxanes) obtained from cyclodextrins and linear polymers, the results clearly indicate that this phenomenon is progressively accelerated by anions that possess larger polarizabilities, smaller surface tension increments, and free energies of hydration.33 Other studies, dealing with the water absorption of natural fibers,31 lamellar phase transitions of some amphiphiles in aqueous dispersion,34 the properties of calixarenes Langmuir monolayers at the air/water interface,32 and the growth rate of two bacterial strains,35 indicate similar trends. Although this evidence does not imply a theoretical model for the description of the microscopic mechanism that regulates the observed phenomena, it is, however, a useful semiquantitative indication that a direct relationship between the measured variables and the characteristic properties of ions exist and is involved in the physical chemistry of their solutions. Adopting the same criteria for the present study, we will focus our attention on the anion effective polarizability in solution (R), the Gibbs free energy of hydration (∆Ghydr), and the molar surface tension increment (σ).
594 J. Phys. Chem. B, Vol. 111, No. 3, 2007
Lagi et al.
Figure 4. ∆µ/kB (O) and C/kB (9) at the critical point as a function of the anion polarizability in solution, R. Lines are guides for the eye.
Figure 7. Tc (O) and xc (9) at the critical point as a function of the anion polarizability in solution, R. Lines are guides for the eye.
Figure 5. ∆µ/kB (O) and C/kB (9) at the critical point as a function of the Gibbs free energy of hydration, -∆Ghydr. Lines are guides for the eye.
Figure 8. Tc (O) and xc (9) at the critical point as a function of the Gibbs free energy of hydration, -∆Ghydr. Lines are guides for the eye.
Figure 6. ∆µ/kB (O) and C/kB (9) at the critical point as a function of the anion molar surface tension increment, σ. Lines are guides for the eye.
Figures 4-6 report the variation of ∆µ/kB (left axis, full lines) and of C/kB (right axis, dotted lines) at the critical point as a function of R, -∆Ghydr, and σ, respectively. The two fitting parameters progressively decrease in going from fluoride (low R, high -∆Ghydr and σ); the data for NO3- in Figure 4 falls outside the general curve (a guideline for the eye), presumably for a discrepancy in the calculated values of the polarizability for this asymmetric and anisotropic anion. For R, we used the
Figure 9. Tc (O) and xc (9) at the critical point as a function of the anion molar surface tension increment, σ. Lines are guides for the eye.
values of the polarizability in solution, which is significantly lower than the corresponding value in the gas phase.43,44 Figures 7-9 show the variation of Tc (left axis, full lines) and of xc (right axis, dotted lines) at the critical point as a function of R, -∆Ghydr, and σ, respectively. Both the critical temperature and lipid mole fraction decrease progressively as R increases, or as -∆Ghydr and σ decrease. Dissolved Gases Effect. The effect of dissolved gases in the solvent was evaluated by measuring Tcp in regular water, then in degassed water, and finally after re-equilibration of the
Insights into Hofmeister Mechanisms
J. Phys. Chem. B, Vol. 111, No. 3, 2007 595
TABLE 5: Cloud Point Temperature (Tcp, K) of diC8PC/ Water Mixtures in Pure Water, Degassed Water, and Re-aerated Water at Two Lipid Mole Fractions Tcp solvent
xs ) 3.6 × 10-3
xs ) 1.0 × 10-3
water degassed water aerated water
292.0 ( 0.2 293.3 ( 0.2 291.9 ( 0.2
316.3 ( 0.2 317.1 ( 0.3 316.3 ( 0.2
lecithin dispersion in contact with a pure gas or air. The results, shown in Table 5 for two different mole fractions of lipid, indicate that when gases are removed under vacuum the cloud point increases by about 1.3 °C with respect to the normal aerated solvent, which is beyond the experimental error. Letting air gases dissolve again in the dispersion brings Tcp back to its normal values. The effect of other gases has been checked by exposing the deaerated dispersion to pure N2, O2, He, and Ar. Table 6 lists the detected Tcp values and the polarizability (in cubic angstroms) of each gas. With respect to the normal dispersion, the removal and addition of different gases leads to a lowering in the miscibility gap. As for the anions, the change in Tcp follows the variation of the gas polarizability. This result confirms the relevance of dispersion forces in this phenomenon, unless other kinds of stronger chemical interactions are involved. Discussion Mechanisms of Micellar Growth. The common mechanisms that are proposed for the interpretation of temperature-induced phase separation are the following two: (1) micelles grow indefinitely until the system phase separates (BTB)18-20 and (2) the intermicellar interactions become stronger and stronger while cooling until the assemblies get together and eventually separate.45 But, the mechanism that leads to phase separation probably also involves hydration and bulk water molecules. This can be seen by considering the cloud point phenomenon for alkyl-PEO/ water systems and, specifically, C8E5. At and above the cloud point, the headgroups give up two water molecules of hydration per EO group. Consequently aP reduces, the surfactant parameter increases, and the micelles grow. However, this is not enough to induce a phase transition, which requires the simultaneous onset of attractive forces between the aggregates. This indeed occurs, as direct force measurements46 show that the forces between monolayers of PEO surfactants change from repulsive to attractive.47 Although the surfactant micellar concentration is low, when the assemblies grow sufficiently to cylinderlike form, any weak force will act to force the aggregation. How this can occur can be comprehended by calculating the distance apart of micelles in the condensed phase. Surface-induced water structure that leads to either repulsive or attractive hydration forces can extend significantly at most about 6 water molecules between the surfaces. The distances between the surfaces of cylinderlike micelles in the condensed phase can easily be estimated for this system to be 21-23 Å. Furthermore, the presence of monomers between the micelles may act to extend the range of this interaction by propagating water structure from the micellar surface and give the required longer ranged attractive force. Therefore, both local water structure and bulk water structure seem to be at play and both are necessary in order to promote phase separation. The same phenomenon is involved in hydrophobic forces proper whose range is vastly extended by dissolved gas or any other sparingly soluble hydrophobic solute. The same mecha-
TABLE 6: Cloud Point Temperature (Tcp, K) of diC8PC/ Water Mixtures in the Presence of Different Gases at a Lipid Mole Fraction of xs ) 3 × 10-3 Gas
Tcp
R (Å3)
degassed He O2 N2 Ar air
297.9 ( 0.3 297.5 ( 0.2 297.2 ( 0.3 296.9 ( 0.2 296.8 ( 0.2 296.4 ( 0.2
0.21 1.57 1.73 1.63
nism must be involved in the other examples quoted. Indirectly, as we discuss below, further it is also involved in the case of electrolytes. In essence, we can expect that adsorption of cosolute or electrolyte at the micellar surface induces changes in the packing parameter due to changed head group interactions. Growth of micelles occurs. With electrolytes, excess anion adsorption induces a charging of the micelles and the high charge of the resulting electrostatic double layer causes an influx of coions (here sodium) and salt from the bulk ionic solution.48 The actual electrolyte concentration between the growing micelles is than very much higher than the bulk concentration would have led us to expect. The activity of the water between the micelles would then be strongly perturbed. But since the system is in equilibrium with a second bulk phase of monomers and salt, the balance for water activity is due to the different bulk hydration capacities of salts. Specific Anion Effect. The specific anion effect on the phase separation in aqueous diC8PC dispersions seems to indicate that the whole phenomenon is directly related to the different ion adsorption at the micellar interface, that depends on the nature of the anions (iodide more than fluoride). This effect will produce an increase in the repulsive electrostatic interactions between polar headgroups. As a resultsin the presence of strongly adsorbing anions such as iodide or nitratesthe area per polar headgroup, aP, will increase and then the packing parameter p will decrease, leading to a smaller micellar size, lower free energy gain, and weaker intermicellar interactions. Electrostatic interactions alone cannot explain these results, as the ionic charge and the electrolyte concentration are the same for all ions, and the missing NES (dispersion) forces must be taken into account. In fact, with an oil/water interface, ions experience, besides an image charge interaction, a (NES) quantum mechanical fluctuation (dispersion potential) given by Lifshitz theory.6 The potential includes image forces and many body dipole-dipole, dipole-induced-dipole and induced-dipoleinduced-dipole forces. This potential W((x) is given schematically as:31
W((x) ∝
(
)
R (iξ) w(iξ) - hc(iξ)
∫0∞ ((iξ)
1 x3
w
w(iξ) + hc(iξ)
dξ
(7)
where R((iξ) is the excess polarizability of the ions as a function of frequency on the imaginary frequency axis (ω ) iξ), x is the distance between the ion and the interface, and w and hc are the dielectric constant of water and of the hydrocarbon phase (hc > w), respectively.2 (In general, the integral is a sum over frequencies and includes temperature explicitly.) These forces that drive ionic adsorption appear to be a major player in several Hofmeister phenomena and depend strictly on the nature of the participating entities. The potential can be positive or negative depending on frequency-dependent dielectric susceptibilities of the substrate.
596 J. Phys. Chem. B, Vol. 111, No. 3, 2007 That inference is supported also by another observation. In their studies of diC8PC cmc values on the added salt concentration, Huang29 and Sarmiento28 show that the cmc is related to the ionic strength I of the electrolyte solution according to ln(cmc/cmc0) ) k1xI + k2I, where k1 and k2 reflect the contributions of the Debye-Hu¨ckel and Kirkwood theories, respectively. As pointed out by Huang in his paper, this means that, for low salt concentration (and ionic strength), the term in xI and hence electrostatic interactions dominate. At higher concentrations, the linear term takes over and (many-body) iondipole interactions become prominent. This is precisely what happens with Hofmeister phenomena, where the nature of the added salts emerges at moderately high concentrations and cannot be explained by simply invoking electrostatic forces. In order to describe in a semiquantitative way the trend shown by the experimental results, we correlate our results to some other physicochemical parameters that are expressions of classical Hofmeister phenomena. In fact, parameters, such as the Gibbs free energy of hydration, partial molar volume, viscosity B coefficient, activity and osmotic coefficients, molar surface tension increment, polarizability, and molar refractivity, all depend on dispersion forces more or less directly.1 It is worth noting here that the viscosity of moderately concentrated electrolyte solutions is well described by η/η0 ) 1 + Axc + Bc,49,50 where η and η0 are the viscosity of the solution and that of pure water, respectively, while both A and B depend on the specific salt, on the solvent, and on temperature. Analogously, the ion activity coefficients of salt solutions are well fitted by similar equations. And more recently, we found that the optical activity of serine and glucose enantiomers follow a similar trend.51 This seems to indicate that ion-solvent interactions and electrostatic and dispersion forces determine the physical chemistry of the solutions in moderately concentrated solutions. In summary, our results clearly indicate the following: (1) Different anions at the same concentration produce significant changes in the cloud point curve of diC8PC/water dispersions and follow the Hofmeister series. (2) These changes affect all the phenomenological parameters that can be calculated according to the micellar growth BTB model. (3) The variation of the parameters related to the liquidliquid phase separation phenomenon (i.e., Tc, xc, γ, ∆µ/kB, and C/kB) reflects the increment of physicochemical parameters (R, ∆Ghydr, and σ) that identify the nature of each anion. (4) The specific ion effect on the cloud point of diC8PC/ H2O dispersions can be interpreted in terms of ion adsorption at the micellar interface, that alters the free energy gain related to micellar growth and the intermicellar interaction parameter. (5) The specific effect produced by each ion can only be explained by including dispersion forces among the effective interactions that affect ions, interfaces, and solvent molecules. (6) After phase separation, the partition of anions between the upper and lower phase is asymmetric (depending on the specific anion), due to adsorption at the micellar interface. (7) Degassing the solvent produces an increment in Tcp; instead, the solution of gases leads to a lowering in Tcp, that depends on the gas polarizability. Conclusions The cloud point curve of a surfactant micellar dispersion is greatly affected by changes in the concentration and composition of electrolytes. In this work, we performed an extensive study on the specific ion effect on the cloud point of diC8PC/water
Lagi et al. dispersions as a function of the lipid mole fraction, in the presence of different anions at the same concentration (50 mM) and with the same cation (Na+). The results clearly indicate that the phenomenon follows the classic Hofmeister series. The experimental data were fitted according to the BlankschteinThurston-Benedek theoretical model that provides the value of the free energy gain related to micellar growth (∆µ/kB) and of the intermicellar interaction parameter (C/kB). These parameters, together with the critical upper consolute temperature Tc, the critical lipid mole fraction xc, and the critical micellar aggregation number gc, all depend on the specific anion used and regularly change with some physicochemical parameterss such as polarizability, free energy of hydration, molar surface tension increment, partial molar volume, or lyotropic numbers that identify the nature of anions. These findings support the hypothesis that dispersion forces, due to the frequency-dependent electrodynamic fluctuations, are directly involved in this phenomenon that cannot be explained simply in terms of electrostatic interactions. In fact, at these moderate concentrations and in the presence of an oil/water micellar interface, ions experience ionic dispersion potentials and adsorb at interfaces in different manners, depending on dielectric susceptibilities and polarizabilities. In more detail, according to this interpretation, anions will be differently affected by the micellar interface through the dispersion ionic potentials and, therefore, will differently adsorb in the proximity of the interface, changing the interactions between the amphiphilic polar headgroups, modifying the value of ap and the packing parameter p ) Vi/ liaP. Changes in p will then affect the micellar growth depending on the kind of anions, by enhancing or inhibiting the micellar growth (measured by ∆µ/kB) at either the central rodlike region or at the hemispherical ends and by affecting the intermicellar interactions and the parameter C/kB. This conclusion is in agreement with previous studies that ascribe to dispersion forces a relevant role in determining a significant specific ion effect in the liquid-liquid phase transition phenomenon taking place in short-chain phospholipid aqueous dispersions and offers a plausible interpretation of the mechanisms behind the specific ion effects on cloud points of surfactant micellar systems in water. Acknowledgment. Partial financial support from Consorzio Interuniversitario per lo Sviluppo dei Sistemi a Grande Interfase (CSGI, Italy) and Ministero dell’Istruzione, Universita` e Ricerca (MIUR, Italy; PRIN2003) is gratefully acknowledged. References and Notes (1) Kunz, W.; Lo Nostro, P.; Ninham, B. W. Curr. Opin. Colloid Interface Sci. 2004, 8, 1-18. (2) Ninham, B. W.; Yaminsky, V. Langmuir 1997, 13, 2097-2108. (3) Bostro¨m, M.; Williams, D. R. M.; Ninham, B. W. Phys. ReV. Lett. 2001, 87, 168103/1-4. (4) Bostro¨m, M.; Williams, D. R. M.; Ninham, B. W. Langmuir 2001, 17, 4475-4478. (5) Ninham, B. W. AdV. Colloid Interface Sci. 1999, 83, 1-17. (6) Ninham, B. W. Progr. Colloid Polym. Sci. 2002, 120, 1-12. (7) Salis, A.; Pinna, M. C.; Bilanicova, D.; Monduzzi, M.; Lo Nostro, P.; Ninham, B. W. J. Phys. Chem. B 2006, 110, 2949-2956. (8) Kim, H. K.; Tuite, E.; Norde´n, B.; Ninham, B. W. Eur. Phys. J. E 2001, 4, 41-417. (9) Bostro¨m, M.; Tavares, F. W.; Finet, S.; Skouri-Panet, F.; Tardieu, A.; Ninham, B. W. Biophys. Chem. 2005, 117, 217-224. (10) Kenkare, P. U.; Hall, C. K.; Kilpatrick, P. K. J. Colloid Interface Sci. 1996, 184, 456-468. (11) Deguchi, K.; Meguro, K. J. Colloid Interface Sci. 1975, 50, 223227. (12) Deguchi, K.; Meguro, K. J. Colloid Interface Sci. 1974, 49, 1015.
Insights into Hofmeister Mechanisms (13) Weckstro¨m, K.; Rosenholm, J. B. J. Chem. Soc., Faraday Trans. 1997, 93, 569-576. (14) Schott, H. Tenside, Surfactants, Deterg. 1997, 34, 304-314. (15) Nilsson, P. G.; Pacynko, W. F.; Tiddy, G. J. T. Curr. Opin. Colloid Interface Sci. 2004, 9, 117-123. (16) Alexandridis, P.; Holzwarth, J. F. Langmuir 1997, 13, 6074-6082. (17) Cacace, M. G.; Landau, E. M.; Ramsden, J. J. Quart. ReV. Biophys. 1997, 30, 241-277. (18) Thurston, G. M.; Blankschtein, D.; Fisch, M. R.; Benedek, G. B. J. Chem. Phys. 1986, 84, 4558-4562. (19) Blankschtein, D.; Thurston, G. M.; Benedek, G. B. J. Chem. Phys. 1986, 85, 7268-7288. (20) Blankschtein, D.; Huang, Y. X.; Thurston, G. M.; Benedek, G. B. Langmuir 1991, 7, 896-897. (21) Israelachvili, J. N.; Mitchell, D. J.; Ninham, B. W. J. Chem. Soc., Faraday Trans. 2 1976, 72, 1525-1568. (22) Carvalho, B. L.; Briganti, G.; Chen, S. H. J. Phys. Chem. 1989, 93, 4282-4286. (23) Lo Nostro, P.; Stubicar, N.; Chen, S. H. Langmuir 1994, 10, 10401043. (24) Huang, Y. X.; Thurston, G. M.; Blankschtein, D.; Benedek, G. B. J. Chem. Phys. 1990, 92, 1956-1962. (25) Tausk, R. J. M.; Karmiggelt, J.; Oudshoorn, C.; Overbeek, J. Th. G. Biophys. Chem. 1974, 1, 175-183. (26) Tausk, R. J. M.; Oudshoorn, C.; Overbeek, J. Th. G. Biophys. Chem. 1974, 2, 53-63. (27) Tausk, R. J. M.; Overbeek, J. Th. G. Biophys. Chem. 1974, 2, 175179. (28) Martinez-Landeira, P.; Ruso, J. M.; Prieto, G.; Sarmiento, F. J. Chem. Eng. Data 2002, 47, 1017-1021. (29) Huang, Y. X.; Tan, R. C.; Li, Y. L.; Yang, Y. Q.; Yu, L.; He, Q. C. J. Colloid Interface Sci. 2001, 236, 28-34. (30) Collins, K. D. Biophys. J. 1997, 72, 65-76. (31) Lo Nostro, P.; Fratoni, L.; Ninham, B. W.; Baglioni, P. Biomacromolecules 2002, 3, 1217-24.
J. Phys. Chem. B, Vol. 111, No. 3, 2007 597 (32) Lonetti, B.; Lo Nostro, P.; Ninham, B. W.; Baglioni, P. Langmuir 2005, 21, 2242-2249. (33) Lo Nostro, P.; Lopes, J. R.; Ninham, B. W.; Baglioni, P. J. Phys. Chem. B 2002, 106, 2166-2174. (34) Lo Nostro, P.; Ninham, B. W.; Ambrosi, M.; Fratoni, L.; Palma, S.; Allemandi, D.; Baglioni, P. Langmuir 2003, 19, 9583-9591. (35) Lo Nostro, P.; Ninham, B. W.; Lo Nostro, A.; Pesavento, G.; Fratoni, L.; Baglioni, P. Phys. Biol. 2005, 2, 1-7. (36) Jendrasiak, G. L. J. Nutr. Biochem. 1996, 7, 599-609. (37) Pullman, B.; Berthod, H.; Gresh, N. FEBS Lett. 1975, 53, 199204. (38) Lo Nostro, P.; Chen, S. H. J. Phys. Chem. 1993, 97, 6535-6540. (39) Huang, Y. X. J. Chem. Phys. 1997, 107, 9141-9145. (40) Martinez-Landeira, P.; Besada, L.; Seoane, L.; Ruso, J. M.; Prieto, G.; Sarmiento, F. Mol. Phys. 2002, 100, 1633-1639. (41) Lin, T. L.; Chen, S. H.; Gabriel, N. E.; Roberts, M. F. J. Phys. Chem. 1987, 91, 406-413. (42) Lin, T. L.; Chen, S. H.; Roberts, M. F. J. Am. Chem. Soc. 1987, 109, 2321-2328. (43) Pyper, N. C.; Pike, C. G.; Edwards, P. P. Mol. Phys. 1992, 76, 353-372. (44) Jungwirth, P.; Tobias, D. J. J. Phys. Chem. A 2002, 106, 379383. (45) Corti, M.; Minero, C.; Degiorgio, V. J. Phys. Chem. 1984, 88, 309317. (46) Kawanishi, N.; Christenson, H. K.; Ninham, B. W. J. Phys. Chem. 1990, 94, 4611-4617. (47) Yaminsky, V. V.; Ninham, B. W.; Christenson, H. K.; Pashley, R. M. Langmuir 1996, 12, 1936-1943. (48) Petrache, H. I.; Kimchi, I.; Harries, D.; Parsegian, V. A. J. Am. Chem. Soc. 2005, 127, 11546-11547. (49) Jenkins, H. D. B.; Marcus, Y. Chem. ReV. 1995, 95, 2695-2724. (50) Marcus, Y.; Ben-Naim, A. J. Chem. Phys. 1995, 83, 4744-4759. (51) Lo Nostro, P.; Ninham, B. W.; Milani, S.; Fratoni, L.; Baglioni, P. Biopolymers 2006, 81, 136-148.
