J. Phys. Chem. B 2006, 110, 2949-2956
2949
Specific Anion Effects on Glass Electrode pH Measurements of Buffer Solutions: Bulk and Surface Phenomena Andrea Salis,† M. Cristina Pinna,† Dagmar Bilanicˇ ova´ ,† Maura Monduzzi,*,† Pierandrea Lo Nostro,‡ and Barry W. Ninham†,‡,§ Dipartimento di Scienze Chimiche, UniVersita` di Cagliari - CSGI, Cittadella Monserrato, S.S. 554 BiVio Sestu, 09042 Monserrato, Italy, Dipartimento di Chimica, UniVersita` di Firenze - CSGI, Via della Lastruccia 3, 50019 Sesto Fiorentino, Italy, and Department of Applied Mathematics, Australian National UniVersity, Canberra, Australia ReceiVed: August 17, 2005; In Final Form: December 12, 2005
The effect of electrolytes on pH measurements via glass electrodes is explored with solutions buffered at pH 7 (phosphate and cacodylate). Salt and buffer concentrations are varied. Direct and reverse Hofmeister effects are observed. The phenomena are significant for salt concentrations above 0.1 M and for buffer concentrations below 20 mM. Changes in measured pH show up most strongly with anions. They can be related to the usual physicochemical parameters (anion molar volumes, molar refractivity, and surface tensions) that are characteristic of Hofmeister series. They correlate strongly with anionic excess polarizabilities; this suggests the involvement of non-electrostatic, or dispersion, forces acting on ions. These forces contribute to ionic adsorption at the glass electrode surface, and to the liquid junction potential.
1. Introduction 1.1. Background. The presence, type, and concentration of a background electrolyte can affect the pH of a solution.1-6 With a glass electrode, the measured pH changes with added electrolyte, buffered or not. The changes depend on salt type.6 Such effects are not accommodated within the standard IUPAC Recommendations on pH measurements and procedures.7 These recommendations rely on the extended Debye-Hu¨ckel theory of electrolytes and the Nernst equation (see sections 5.1 and 11.1 in that reference). Neither theoretical assumptions embrace specific ion effects. For the best up-to-date account of the situation on hydronium activity determination see ref 8. In his pioneering work, starting in 1888, Hofmeister studied specific ion effects of various salts, at a fixed ionic strength, for precipitation (salting-out) of ovalbumin.9,10 The precipitation efficiency of the anions of sodium salts was in the following order:
H2PO4- >SO42- > F- > Cl- > Br- > NO3- > I- > ClO4- > SCNThis usual Hofmeister series ordering is qualitative. Sometimes, depending on the kind of the protein or other colloidal suspension studied, it occurs in reverse order.11-13 Moreover, while the salting-out efficiency follows this sequence when the pH is higher than the isoelectric point (pI), it follows the opposite order at pH < pI.11,14 The phenomenon of reversal of Hofmeister series, reviewed in the 1920s,12,13 provides a benchmark test for any theory. Anions as coions or as counterions that lead to this reversal of Hofmeister series implicate anion-substrate interactions. Occasionally some anions show anomalous be* Corresponding author phone: +39 070 6754385; fax: +39 0706754388; e-mail:
[email protected]. † Universita ` di Cagliari - CSGI. ‡ Universita ` di Firenze - CSGI. § Australian National University.
havior.15 Several investigations into the role of added salts in influencing protein behavior and enzyme activity have recently appeared with a focus on Hofmeister anion series. Among many such studies, it is worth mentioning a surprising buffer and anion effect on the cutting efficiency of a DNA-restriction enzyme with varying electrolyte background pairs.16 Other interesting effects of anions in the Hofmeister series were shown in the case of cytochrome C and lysozyme solutions.17,18 Hofmeister effects also show up even in such complex biological systems as water uptake of wool fibers and bacterial growth rates of Staphylococcus aureus and Pseudomonas aeruginosa.19,20 Lipase activity increases as a result of salt addition and shows a clear specific optimal dependence on NaBr ionic pair, independently of salt-induced pH decrease.21,22 The same occurs for the superactivity of horseradish peroxidase.23,24 In the framework of the conventional language for the description of water structure and electrolytes, ions are designated as kosmotropic or chaotropic. The former are supposed to act as structure makers, and they form the left-hand side of the Hofmeister series. The sequence then merges into the chaotropic species, the supposed structure breakers.25 Ions with small size and high charge (fluoride, sulfate, calcium, and aluminum) belong to the kosmotropic class, while large and monovalent ions (thiocyanate, iodide, and cesium) are considered to be chaotropes. This classification relates the observed Hofmeister effects to the ability of ions to strengthen or to break hydrogen bonds in aqueous solutions. In essence ion specificity is attributed to a nonlocal bulk effect that is determined by solvation processes and energies. However, it has been recently shown that ions affect the first few hydration shells. Ions neither enhance nor weaken the hydrogen bond network, at least over the time scale experienced by femtosecond pump spectroscopy.26-29 The assignment of the major carriage of Hofmeister phenomena to surface or bulk effects induced by ions is a long standing question. For a review of the present state of affairs with Hofmeister effects interpretation, and modeling see ref 30.
10.1021/jp0546296 CCC: $33.50 © 2006 American Chemical Society Published on Web 01/21/2006
2950 J. Phys. Chem. B, Vol. 110, No. 6, 2006
Salis et al.
Germane to the present work, some evidence of specific ion effects emerged in investigations on the effect of ionic strength on pH measurements.5,6 Another study on the role of added salts in pH measurements invokes additional ion-specific potentials to explain the measured bulk pH values.31 It was shown that the measured changes in pH could be quantitatively accommodated by modifying conventional double layer theory that predicts the potential and hence, via charge regulation, the hydronium ion concentration at the electrode surface. However the analysis of pH in ref 31 ignores dissolved CO2 and its equilibria, to which such variation is usually attributed. Indeed the amount of CO2 dissolved in water depends significantly on the concentration and on the presence of specific ion pairs.32 The main point was nevertheless met in that work;31 it has been demonstrated that the inclusion of the surface forces can accommodate the magnitude and specificity of the effects. 1.2. Present Work. This work is an attempt to take together the effects described above that bear on the question of pH and to make some more systematic observations. Our focus is on the role of added salts, their type and concentration, and on the modification of pH in different buffer solutions (phosphate and cacodylate). We propose that pH changes are accommodated in terms of a theoretical approach that accounts for both surface and bulk specific ion contributions. In particular, adsorption potential contributions, that are surface effects at the glass electrode membrane, must include non-electrostatic (NES) or many body dispersion forces. By the term dispersion forces we mean the totality of NES quantum mechanical fluctuation forces. These are accessible from and included in Lifshitz theory and its extensions.33 They are temperature and salt dependent and also depend on frequency dielectric susceptibilities of the interacting entities and on ion fluctuation effects. In a proper treatment of these forces not yet reported in the literature, they automatically include ion specific contributions to the self-free energies (hydration) of ions, and interactions of these hydration profiles (also cf., Born electrostatic energies of ion transfer).34 The dispersion potential (Udispersion) felt by an ion at a distance x from an interface is in first approximation:
Udispersion(x) ≈
(nwater2 - nsubstrate2)R/i (0)pωi 8x3
(1)
It changes sign accordingly as the refractive index of the substrate (nsubstrate) is larger or smaller than that of water (nwater). An event similar to this happens for the static excess polarizability of the ion in water (R/i (0)). In general the complete expression involves a sum over all frequency contributions. As the distance goes to zero, the potential goes over to the change in hydration free energy of the ion on contact. pωi is the electron affinity (or ionization potential) for the ion. The electron affinity of the ions in water is not known experimentally, but it should be between an IR and a UV frequency.35 The general formulas for these many body forces are complicated, but the simpler result shown above for dispersion forces effectively captures the essence of the problem. These forces include, in the special limiting case of dilute media, the dipole-dipole (Keesom), dipole-induced dipole (Debye), and dispersion (London) forces. These take on a completely different form in condensed media. Moreover when such forces have been included in theories like DLVO theory of colloid stability, they have been invoked in a way that handles them in a linear theory, whereas, the double layer or other
electrostatic (ES) forces are treated in a nonlinear theory. This result is that ion specificity is missed.30,36 These forces are missing from conventional theories of electrolytes, of the electrical double layer, and of interactions.36 In electrolytes, most such effects are strongly correlated to anion polarizability. This reflects the operation of dispersion forces and seems to be a recurring key factor in understanding specific ion effects.15,20,37-39 In addition, the same NES forces account for many specific ion effects hitherto inexplicable, including Hofmeister effects with interfacial tensions, Born free energies of transfer of ions, and direct and reverse Hofmeister series.18,34,40 In the present paper, except for a series of pH measurements in water solutions at different ionic strength, phosphate and cacodylate buffer media at different concentrations, and initially at neutral pH, were used. This avoids problems related to CO2 dissolution in water because after air equilibrium, the total concentration of carbon dioxide-related species is only about 1.2 × 10-5 M. 2. Experimental Section 2.1. Chemicals. Sodium di-hydrogen orthophosphate 99%, di-sodium hydrogen orthophosphate 99%, potassium phosphate monobasic 99.5%, and di-potassium hydrogen orthophosphate 99% were from Carlo Erba. Cacodylic acid 98% (CH3)2AsOOH, sodium hydroxide 98%, and potassium thiocyanate 99% were from Sigma. Potassium iodide >99.5%, potassium perchlorate 99%, and potassium bromide >99.5% were from Fluka. Potassium chloride 98% was from Aldrich. Sodium perchlorate 99%, sodium bromide >99%, sodium nitrate 99%, and sodium iodide 99% were from Acros. Sodium chloride 99.5% and potassium nitrate 99% were from Merck. Glass electrode calibration buffer solutions, pH 4 (20 °C), (citric acid/sodium hydroxide/hydrogen chloride) and pH 7 (20 °C), (di-sodium hydrogen phosphate/potassium di-hydrogenphosphate), both traceable to SRM from NIST and PTB, were from Merck. 2.2. Preparations of the Buffer/Salt Solutions. Buffer solutions contained either sodium phosphate buffer and sodium salts, sodium cacodylate buffer and sodium salts, or potassium phosphate buffer and potassium salts. The pH was 7 with buffer concentrations of 0.1, 1, 5, and 20 mM. Distilled water was purified through a Millipore system (Simplicity 185) with conductivity ClO4- > Br- ≈ Cl- > NO3-, is obtained for the measured pH values. Around and above 1 M, gas solubility of O2, N2, or CO2, which decreases linearly with salt concentration, is very low. It is remarkable that at high concentration, the different salts modify the measured pH differently. These results show trends similar to those reported by Bostro¨m et al.31 There is here an apparent Hofmeister effect on pH measurements with an increasing salt concentration. In this context, “apparent” seems to be the most appropriate definition, since it is not possible to distinguish at this moment the real effect due to the salt addition from the variation due to a decrease in dissolved
Figure 2. pH of phosphate (a) and cacodylate (b) buffers (5 mM, initial pH 7) with increasing concentration of salts: NaCl (]); NaBr (2); NaNO3 (O); NaI (b); NaClO4 (9); calculated pH from DH limiting law (dashed line). SDmax ) ( 0.02 pH units.
CO2. The removal of carbon dioxide with a stream of nitrogen bubbled through nonbuffered NaNO3 aqueous solutions (0.5 and 1.0 M), produced an increase of 0.3 and 0.2 pH units, respectively. This suggests that, although the ionic strength is very high, some CO2 is still present, and the extended DebyeHu¨ckel theory does not predict the experimental pH value. However, since the goal of this work is the effect of background electrolytes in pH measurements of buffered solutions, no attempt to further rationalize the specific ion effect on pH measured in nonbuffered solutions was done. 3.2. Salt Effects on the pH of Buffered Solutions. Sodium phosphate and sodium cacodylate buffers 5 and 20 mM, at pH 7, were used as starting solutions, and the various sodium salts were added. Again pH values were measured at different salt concentrations. Figures 2a and b show the results of pH measurements as a function of sodium salt concentration for starting solutions of 5 mM phosphate and cacodylate buffers at pH 7. The dashed lines indicate the pH values calculated according to the extended Debye-Hu¨ckel (DH) theory (see eqs 2-4). The first remark is that, in the presence of a buffer at pH 7, the addition of salts causes a significant decrease of the measured pH values as the concentration increases. Different pH changes are recorded in the two buffers for the various sodium salts. The efficacy of anions in changing pH follows
2952 J. Phys. Chem. B, Vol. 110, No. 6, 2006
Figure 3. pH of phosphate buffer pH 7, 20 mM solution with increasing concentration of salts: NaCl (]); NaBr (2); NaI (b); NaNO3 (O); NaClO4 (9); calculated pH from DH limiting law (dashed line). SDmax ) ( 0.02 pH units.
different sequences. The effects are strongly dependent on the nature of the buffer. At high salt concentration (1 M), ∆pH values in the range of 0.60-0.85 and of 0.2-0.4 pH units are observed for phosphate and cacodylate, respectively. Phosphate buffers induce a larger decrease of pH values than cacodylate. The extended DH theory calculations show a trend qualitatively similar to those experimentally observed (see the discussion below). For the 5 mM buffers, the anion effectiveness in changing pH is ClO4- > NO3- ≈ I- > Br- > Cl- for phosphate buffer, and Br- > Cl> NO3- > ClO4- for cacodylate buffer, with an inversion in the series. It is worth recalling that, in terms of the kosmotropic/ chaotropic model, the phosphate anion itself is probably the most efficient kosmotrope, whereas the added anion having the highest chaotropic effect (ClO4-) is the most effective in changing the measured pH. Thus the anions show a decreasing effectiveness from chaotrope to kosmotrope in the Hofmeister anion sequence. For the phosphate buffer, a clear crossover between NaI and NaClO4 occurs around 0.4 M salt concentration (Figure 2a). The same effect is observed between NaNO3 and NaClO4 in cacodylate buffer (Figure 2b). This effect does not appear in the 20 mM phosphate buffer (see Figure 3). The effect of buffer concentration was further investigated for NaCl and NaClO4 salts (see section 3.4). 3.3. Reversal of Effects with Potassium Salts. The measurements shown in Figure 2a were repeated replacing sodium with potassium. Because of its low solubility, potassium perchlorate was replaced by potassium thiocyanate. The pH measurements were performed up to a 2 M concentration of salts. A smaller decrease of pH was observed with potassium than with sodium. And, considering the data measured at 1 M concentration, the effects are in the order Br> Cl- ≈ NO3- > SCN-> I-, as shown in Figure 4. That is, replacing Na+ with K+ brings about a reverse Hofmeister series effect. It seems difficult to assign this finding to a bulk effect, and presumably a mechanism involving specific surface adsorption of cations is indicated.18 3.4. Role of Buffer Concentration. Phosphate buffer solutions 0.1, 1, 5, and 20 mM were used. Results are shown in Figure 5a for NaCl and Figure 5b for NaClO4. Besides the expected major effect of NaClO4 salt, the role of buffer concentration can be clearly seen. At 0.1 mM, the buffer
Salis et al.
Figure 4. pH of potassium phosphate (buffer 5 mM, pH 7) with increasing concentration of salts: KCl (]); KBr (2); KI (b); KNO3 (O); KSCN (0); calculated pH from DH limiting law (dashed line). SDmax ) ( 0.02 pH units.
Figure 5. Effect of phosphate buffer concentration on pH of NaCl (a) and NaClO4 (b) solutions. 0.1 mM (O); 1 mM (2); 5 mM (]); 20 mM (b).SDmax ) ( 0.02 pH units for buffer concentration 5 and 20 mM and SDmax ) ( 0.05 pH units for buffer concentration 0.1 and 1 mM.
shows a fluctuating, but markedly decreasing, trend for both salts. ∆pH values of about 0.9 and 1.3 are measured for Cland ClO4- anions, respectively. As expected, the ∆pH effects decrease significantly with increasing buffer concentration. The more dilute the buffer solution, the lower the recorded pH value, regardless of the electrolyte concentration.
Effect of Electrolytes on pH Measurements
J. Phys. Chem. B, Vol. 110, No. 6, 2006 2953
TABLE 1: pH Values of Phosphate Buffer Solutions in the Presence of Different Electrolytes (0.5 and 1 M), Partial Molar Volume (νs, cm3/mol), Molar Refractivity45(Rs for 1 M Salt Solutions, cm3/mol), and Molar Surface Tension Increment (σ, mN‚L/m‚mol) Values for Each Anion pH [salt] ) 0.5 M
[salt] ) 1 M
[buffer] ) 20 mM
[buffer] ) 5 mM
[buffer] ) 20 mM
[buffer] ) 5 mM
νs
Rs
σ
7.0 6.6 6.5 6.5 6.5 6.5
7.0 6.6 6.5 6.5 6.4 6.35
7.0 6.4 6.3 6.3 6.3 6.2
7.0 6.35 6.3 6.3 6.3 6.2
16.6 23.5 27.7 35.0 43.0
8.51 12.13 11.02 13.72 19.25
1.63-2.21 1.31-1.83 0.89-2.08 1.08-1.23 0.22-0.62
H2O ClBrNO3IClO4-
Finally we remark that cations also seem to play a key role as demonstrated by the different behavior observed in Na+ and K+ phosphate/salt systems. 4. Discussion Besides any speculation about the interpretation of these results, it is clear that a delicate balance related to the different composition of buffers and salts, and to their respective concentrations, plays a subtle role in determining the measurement of pH of the solution. All these variables affect pH values significantly. At least what we measure. This seems not to have been widely recognized. Whether the measurements reflect a real or only an apparent change in pH seems to be an important open question. It is, perhaps, not too surprising that in the classical theory that underlies interpretation of the pH measurement NES forces acting on ions are missing. The observations have relevance to biological problems for two reasons. The changes in pH measured, whether real or apparent, depend on salt concentration that is ionic strength or Debye length. In a real biological situation, for example, physiological saline roughly 0.15 M, it might be expected that these effects are small. But in fact a real systemsthe cytoplasms contains a significant amount of proteins, carbohydrates, and other multivalent species. For such a system, even a very small amount of multivalent ions increases the Debye length dramatically42-44 Not only is the measured pH influenced, but dramatic changes in the enzymatic activity were found as a result of added electrolytes in different buffers.14,16,19-22 4.1. Correlation with Other Physicochemical Properties: Hofmeister Fingerprints. We summarize our results in terms of direct or reverse Hofmeister series (HS). The usual, direct Hofmeister series is as follows:
show correlations when the experimental data are compared to some physicochemical parameters that are characteristic fingerprints of Hofmeister effects. Typically some of these quantities are ion excess polarizability, partial molar volume, molar refractivity, surface tension molar increment, Gibbs free energy and entropy change of hydration, lyotropic number, Jones-Dole viscosity B-coefficients, Setschenow constants, entropy change of water, and so forth. Since they depend on the chemical nature, these quantities provide a marker for each ionic species or ion pair.30 Regular trends are found when the experimental pH values are related, at fixed concentrations, to the partial molar volume (νs) of the anions, the molar refractivity (Rs) and the surface tension molar increment (σ). For a more detailed presentation of these parameters, see refs 30, 38. Table 1 summarizes those parameters along with the pH values measured in the presence of the various salts, at 0.5 and 1 M concentration, and in 5 mM and 20 mM phosphate buffers. Figure 6 shows the change in the measured pH value as a function of the νs of each anion. Remarkably, an almost linear trend is observed. The pH decreases with increasing νs, and Rs, and with decreasing σ. In terms of the kosmotropic/chaotropic language, the pH is less affected by the harder, less polarizable kosmotropes (Cl-). It consistently decreases with the softer, more polarizable chaotropes (ClO4- and I-). These properties (νs, Rs, and σ) are directly related via (mainly) polarizability of anions. This is so for hydration or self-free energies of ions.34 It is also so for changes in Born and dispersion self-energies due to different
H2PO4- > SO42- > F- > Cl- > Br- > NO3- > I- > ClO4- > SCNthen, in terms of effectiveness in modifying pH, we observe: pure water; Na+ salts
I- > ClO4- ≈ Br- > Cl- ≈ NO3reverse HS
Na+ phosphate 5 mM; Na+ salts ClO4- > NO3- ≈ I- > Br- > Clreverse HS Na+ cacodylate 5 mM; Na+ salts Br- > Cl- > NO3- > ClO4direct HS K+ phosphate 5 mM; K+ salts
Br- > Cl- > NO3- > SCN->Idirect HS
Phenomena where significant specific ion effects occur usually
Figure 6. pH values of phosphate buffer solutions in the presence of different electrolytes, as a function of the partial molar volume. b: [buffer] ) 20 mM and [salt] ) 0.5 M; O: [buffer] ) 5 mM and [salt] ) 0.5 M; 4: [buffer] ) 20 mM and [salt] ) 1 M; 2: [buffer] ) 5 mM and [salt] ) 1 M. Partial molar volumes from ref 45.
2954 J. Phys. Chem. B, Vol. 110, No. 6, 2006
Salis et al.
ion-substrate interactions, that are given by the same theoretical formalism of Lifshitz.34,36,46,47 We can infer that, also, the same highly specific electrodynamic NES interactions must participate in interpretation of the salt- and buffer-dependent pH change. These observations, together with the reversals with buffer type and cation, suggest that the phenomenon is due to a delicate interplay of different intermolecular interactions. These interactions are related to polarizabilities and ionization potentials, and to corresponding dielectric properties of substrate and water. We now discuss how NES forces might affect pH via bulk and surface interactions between ions in the solution and between ions and glass electrode surface, respectively. 4.2. Bulk effects. A direct bulk effect of the activities (a ) γ × c) of the acid/base pairs (H2PO4-/HPO42- and (CH3)2AsOOH/(CH3)2AsOO-) must be considered. The standard analysis used to calculate the activity coefficients of the buffers in the presence of salts is here resumed by giving the most important formulas. From the equilibrium:
H2PO4- + H2O a HPO42- + H3O+ We can write the expression of the thermodynamic equilibrium constant Ka20:
Ka2 ) 0
pH ) pKa20 + log
a(HPO42-) × a(H3O+) a(H2PO4-) × a(H2O)
[HPO42-] [H2PO4-]
+ log
(2)
γ(HPO42-)
γ(H2PO4-) log a(H2O) (3)
Where γ(X-) and [X-] are the activity coefficient and the molar concentration of the generic anion X-, respectively. A similar equation can be written for the cacodylate buffer. The expected pH can be calculated from these equations, assuming that K0a2 and the molar concentrations of the acid and its conjugated base are constant (see section 3.2). With this assumption pH changes originate from the variation in activity coefficients with ionic strength I. Activity coefficients are calculated from the extended Debye-Hu¨ckel (DH) equation:
xI -log γz ) Az2 1 + BaxI
(4)
where A and B are constants that depend on temperature and the dielectric constant of the solvent , z is the ion charge, and a is an ion size parameter (unknown, subject to best fitting demands). The dashed lines reported in the Figures 2, 3, and 4 were obtained using eqs 3-4 and assuming a(H2O) ) 1. Such equations do not take into account any real ion specificity. More importantly, they do not account for the series inversion observed for phosphate and cacodylate buffers (Figure 2). Even more problems arise when we replace sodium with potassium (compare with data in Figure 2a and Figure 4). To account for the observed deviations from a purely electrostatic approach, different attempts have been made by adding to eq 4 a first-order term (bI), but b is, again, a fitting parameter.48,49 Another approach can be the introduction, in the pH calculations, of the activity coefficients of water obtained from the osmotic coefficients according to the usual equation
for an electrolyte solution:48
ln a(H2O) )
VmWH2O φ 1000
(5)
where V is the number of ions formed from one mole of electrolyte; m is the molality (mol Kg-1), WH2O is the molecular weight of water, φ is the osmotic coefficient of the electrolyte. Then, the pH values for NaCl, NaBr, and NaNO3 salts were calculated using eq 5 and according to eqs 3 and 4 (e.g., the extended DH equation with pK0a2phosphate ) 7.2055, A ) 0.51, and B ) 0.33 at 25 °C in water; a ) 4 Å for H2PO4- and HPO42-).50 Comparing these new figures with those calculated by the nonion-specific extended DH theory (see Additional Materials), the pH variations for the different salts appear in the 3rd or 4th digit in pH units. Clearly this is irrelevant for either a more reliable prediction of the experimental data or to justify the observed HS effects. Hence, the theoretical curves reported in Figures 2-4 are simply based on the extended DH law. These results suggest that some other considerations have to be involved. According to eq 3, the effect of salts on pH can be related to changes in the values of the activity of HPO42-, H2PO4-, and water. Since a detailed calculation of a(H2O) from osmotic coefficients shows that its value changes only slightly from one salt to another, its contribution cannot be responsible for the deviations detected. This evidence prompts us to ascribe the main role to the other terms, namely pK0a2 (K0a2, the dissociation constant of H2PO4-) and log[a(HPO42-)/a(H2PO4-)]. It may be proposed that NES dispersion interactions between the salt anions and the ionic species that participate in the acid/base equilibrium, can lead to a delicate but nonzero variation of the ratio K0a2 × a(H2PO4-)/a(HPO42-). The balance of this variation and the change in a(H2O) would determine the final pH. This hypothesis relies on the fact that salts can change the ionization state of a weak acid.51-53 What is new here is that NES dispersion forces can be at the origin of such phenomenon. In this context, for instance, it is explained why chloride, that can establish hydrogen bond interactions with the monovalent biprotic species H2PO4-, produces the lowest change in pH. Instead, a species less active in producing hydrogen bonds but with a larger polarizability, such as I-, will stabilize the monoprotic divalent HPO42- ion, with a consequent larger effect on pH. 4.3. Explanation of Bulk Phenomena. Let us focus now on the results obtained in 5 mM buffers. Figure 2a) and b show two main results: (i) the measured pH is not constant but decreases with increasing salt concentration; (ii) the pH decrease is ion specific and follows the Hofmeister series. Tentatively, these results may be accommodated in terms of bulk phenomena. If we consider the range of ionic strength up to 0.1 M, we can, in first approximation, neglect the terms containing the activity coefficients in eq 3. Even in this case the measured pH is lower than the expected value of 7. This can be explained in terms of ES forces acting between Na+, H+, and their counterparts H2PO4-, HPO42-. Close to neutrality, [H+] ) 10-7 mol/L and [Na+] ) 10-1 mol/L, so Na+ is 106 times more concentrated than H+. Thus, Na+ will compete with H+ for interacting with the divalent anion HPO42-. This will lead to less H+ bound to phosphate buffer and more H+ free in the solution and thus to a lower pH. This phenomenon will increase by increasing salt concentration from 0.1 to 1.0 M. A similar explanation appears to be responsible for the decrease in pH with salt concentration in the case of cacodylate buffer. The higher pH measured in
Effect of Electrolytes on pH Measurements
J. Phys. Chem. B, Vol. 110, No. 6, 2006 2955
this case, with respect to the phosphate buffer, should be produced by the weaker interactions between Na+ and (CH3)2AsOO- as compared to that between Na+ and HPO42-. Ion specificity begins to show up strongly at I > 0.1 M, and it increases with concentration. This range of concentration is that where the NES (dispersion) forces begin to dominate.36 The extended DH theory contains additional ion, ion pair, and buffer size parameters in the expression for γz (see eq 4). Since no theoretical expression for activity coefficients that considers ion specificity, or mixed electrolytes has yet been derived, the availability of such parameters, specific to each case is tantamount to the tautological restatement that pH is buffer and salt concentration dependent. In qualitative terms, dispersion potentials between ions will be different for differently polarizable anions. The effect will give different pH/salt curves for different anions. 4.4. Surface Effects. Specific ion adsorption at a metal electrode surface is a well-known phenomenon. It was treated in terms of a Langmuir isotherm by Stern in 1924,54 and the ideas were successively developed by Parsons.55 It was believed that the adsorption occurs via electronic transfer from the anion to the orbitals of the metallic electrode.56-58 More recently a new approach correlating electrochemistry with colloid science emerged. In particular, Conway found a Hofmeister series in the zero charge potential of Hg electrodes. He proposed a new explanation in terms of solvation factor.59 A succeeding work by Conway60 correlated the adsorption with the effective ionic radii and hydration volumes. The main conclusion was that electronic effects alone are insufficient to describe ion specificity. Hydration effects in anion adsorption were invoked as the substantial part of the observed specificity of ion adsorption.60 Ion adsorption specificity was also observed at air-water interface where no such electronic effects can be invoked.34,40,61 Clearly, ion adsorption may occur at the glass-water interface as well. In the past, bulk effects had been expected to account for the general decrease of pH and ion specificity of pH/salt curves. If bulk effects were the only phenomenon responsible for the observed changes, we should expect the same Hofmeister series trends with different buffers for enzyme action,16 and for reverse and direct sequences with proteins.18 Figure 2 shows that the Hofmeister series is inverted if cacodylate buffer instead of phosphate is used. In addition, Figure 2a and Figure 4 indicate a reverse Hofmeister series when sodium is replaced by potassium. These effects may be explained in terms of different competition between ionic species for the buffer and/or the glass electrode surface.31At the very least ref 31 shows that the same NES forces responsible for interfacial tensions and specific ion effects in proteins predict changes in pH of the right magnitude and cannot, therefore, be neglected. 4.5. A Comment on Junction Potential. Glass electrode functioning is based on a glass membrane that is sensible to H+ activity. When calibrated against standard buffer solutions of known pH (pHs), in principle it shows a Nernstian behavior. It is known that:1
pHmeasured ) pHs -
(Es - E′) + (E′LJ - ELJ) 2.303RT/F
(6)
where, Es is the potential of the standard buffer solution, E′ is the potential of the unknown solution, E′LJ and ELJ are the liquid junction potentials of the unknown and the standard buffer solutions, respectively. Liquid junction potentials are caused by the different distribution of cations and anions at the interface between two different electrolyte solutions. This asymmetry is
usually ascribed to different activities and diffusion velocities across the interface between the two solutions.7,8,62 In a typical pH measurement the inner solution of a glass electrode is always the same, and the different ELJ is due to the different composition between the standard and the unknown solution. Differences in ELJ are usually ascribed to differences in ionic strength1,3 or to H+ activity coefficient differences in the presence of different background salts (bulk effect).6 In the view of old and recent findings related to Hofmeister effects,30 rationalized in terms of dispersion forces, an alternative explanation can be proposed. The potential E′LJ changes with type and concentration of the electrolyte in the unknown solution; this is caused by ion-glass membrane interaction (surface effect), driven by the combination of ES and NES forces, which, we remark, are not additive.36 This will lead to ion specific (E′LJ - ELJ) differences and thus to ion specific pHmeasured.8 In a more sophisticated electrochemical treatment, the different contributions to the glass electrode response should be considered. Namely, the potential difference on the membrane, the ohmic drop, the internal and external reference electrode potentials, and the liquid junction potential arising from the migration of ions between the sample and the reference solution must be taken into account. The latter being probably the most affected by the variation of the background electrolyte composition and concentration.63 However, in the present study, we report and discuss the effect of salt solutions on the overall measured pH of a buffer solution, and do not make any attempt at separating the individual outcome on each single potential. 5. Conclusions We have shown that conventional theories are not able to predict experimental measurements on the pH of buffer solutions with increasing salt concentration. A first attempt to rationalize these effects was done by Bostro¨m et al.31 This analysis was incorrect since the experiments did not take into account CO2 dissolution.64 Nevertheless, the idea that dispersion forces acting on ions were sufficiently large to accommodate observed trends there observed and in buffers seemed to be interesting. Indeed, the dispersion forces contribution36 has explained a number of Hofmeister phenomena. NES forces certainly exist and have not been taken into account properly. However, the debate on ascribing Hofmeister effects to bulk or surface effects is in fact largely semantic. A more subtle bulk effect can be identified. Thus the interactions of an ion with water, in the presence of its neighbors, involves contributions from the Born (electrostatic interactions) plus quantifiable specific NES many body contributions that are included in the terms hydration.34 The consequently “dressed” ions, with hydration “shells” that are soft or hard, kosmotropic or chaotropic depending on one’s taste in terminology, can interact, and overlap, more or less via combined ES and NES forces, to give rise to ion specific activities. The specificity seems to occur mainly via ionic dispersion forces that, within the same (Lifshitz) formalism, give both self and interaction energies. The same route can be accomplished, in principle, via molecular simulations. But it is somewhat simpler to use measured dielectric properties of solutions as a function of frequency to deduce the required energies. The same processes go on at any interface. Our results seem to indicate that the adsorption of ions at the glass electrode/water interface is a not negligible phenomenon. This conclusion is consistent with other interfacial phenomena,15,18,40 and with some more recent papers that show that only a few layers of hydrating solvent molecules are
2956 J. Phys. Chem. B, Vol. 110, No. 6, 2006 strongly affected by the ionic electric field, while the structure of bulk water remains unperturbed.26-29 So, a contribution that involves specific surface adsorption of ions at interfaces seems to be more relevant. The wider implications, we believe, deserve note. For if, what we believe is, an experimental pH measurement relies on an interpretation based on incorrect assumptions, then the implications for extrapolating and using such measurements to infer surface pH and potential of a biomembrane, and the standard pKas of proteins and colloids, is also an open question. These effects are important if one considers that the glass electrode procedure is the most common way for measuring pH, particularly in biological systems. In the end, this paper shows how the addition of different electrolytes affects the pH of a buffer solution measured with a glass electrode. This effectswhich depends on the nature of salt and buffer solutions, and on their concentrationsis real and cannot be neglected. It is not our intention here to discuss, in detail, the particular potentials that contribute to the measured pH value, and that may be affected by the presence of the electrolytes. This investigation will be the object of a future work. Acknowledgment. MIUR-Prin 40% (Italy), and Consorzio Sistemi Grande Interfase (CSGI, Italy) are acknowledged for financial support. B.W.N. thanks MIUR for a visiting professor position within the “Rientro dei cervelli” project 2003. Supporting Information Available: Calculated pH of phosphate (5 and 20 mM) and cacodylate (5 mM) buffer solutions at initial pH 7 with increasing salt concentrations. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Hedwig, G. R.; Powell, H. K. J. Anal. Chem. 1971, 43, 1206. (2) Garcia-Mira, M. M.; Sanchez-Ruiz, J. M. Biophys. J. 2001, 81, 3489. (3) Brown, R. J. C.; Milton, M. J. T. Accredit. Qual. Assur. 2003, 8, 505. (4) Pooler, P. M.; M. L. Wahl; Rabinowitz, A. B.; Owen, C. S. Anal. Biochem. 1998, 256, 240. (5) Brandariz, I.; Barriada, J. L.; Taboada-Pan, C.; Sastre de Vicente, M. E. Electroanalysis 2001, 13, 1110. (6) Brandariz, I.; Vilarino T.; Alonso P.; Herrero, R.; Fiol, S.; Sastre de Vicente, M. E. Talanta 1998, 46, 1469. (7) Buck, R. P.; Rondinini, S.; Covington, A. K.; Baucke, F. G. K.; Brett, C. M. A.; Camoes, M. F.; Milton, M. J. T.; Mussini, T.; Naumann, R.; Pratt, K. W.; Spitzer, P.; Wilson, G. S. Pure Appl. Chem. 2002, 74, 2169. (8) Schneider, A. C.; Pasel, C.; Luckas, M.; Schmidt, G. K.; Herbell, J.-D. J. Solution Chem. 2004, 33, 257. (9) Hofmeister, F. Arch. Exp. Pathol. Pharmakol. 1888, 24, 247. (10) Kunz, W.; Henle, J.; Ninham, B. W. Curr. Opin. Colloid Interface Sci. 2004, 9, 19. (11) M. Bostro¨m; F. W. Tavares; S. Finet; F. Skouri-Panet; A. Tardieu; B. W. Ninham. Biophys. Chem. 2005, In Press. (12) Loeb, J. Science 1920, LII, 449. (13) Gustavson, K. H. Specific ion effects in the behaviour of tanning agents toward collagen treated with neutral salts. In Colloid Symposium Monograph; Weiser, H. B., Ed.; The Chemical Catalog Company Inc.: New York, 1926. (14) Finet, S.; Skouri-Panet, F.; Casselyn, M.; F.Bonnete; A. Tardieu. Curr. Opin. Colloid Interface Sci. 2004, 9, 112. (15) Lonetti, B.; Lo Nostro, P.; Ninham, B. W.; Baglioni, P. Langmuir 2005, 21, 2242. (16) Kim, H.-K.; Tuite, E.; Norde´n, B.; Ninham, B. W. Eur. Phys. J. E 2001, 4, 411. (17) Baglioni, P.; Fratini, E.; Lonetti, B.; Chen, S. H. J. Phys.: Condens. Matter 2004, 16, 5003. (18) Bostro¨m, M.; Tavares, F. W.; Finet, S.; Skouri-Panet, F.; Tardieu, A.; Ninham, B. W. Biophys. Chem. 2005, 117, 217.
Salis et al. (19) Lo Nostro, P.; Lo Nostro, A.; Ninham, B. W.; Pasavento, G.; Fratoni, L.; Baglioni, P. Curr. Opin. Colloid Interface Sci. 2004, 9, 97. (20) Lo Nostro, P.; Ninham, B. W.; Lo Nostro, A.; Pesavento, G.; Fratoni, L.; Baglioni, P. Pys. Biol. 2005, 2, 1. (21) Pinna, M. C.; Salis, A.; Monduzzi, M.; Ninham, B. W. J. Phys. Chem. B 2005, 109, 5406. (22) Pinna, M. C.; Salis, A.; Monduzzi, M.; Ninham, B. W. J. Phys. Chem. B 2005, 109, 14752. (23) Pinna, M. C.; Bauduin, P.; Touraud, D.; Monduzzi, M.; Ninham, B. W.; Kunz, W. J. Phys. Chem. B 2005, 109, 16511. (24) Bauduin, P.; Nohmie, F.; Touraud, D.; Neueder, R.; Kunz, W.; Ninham, B. W. J. Mol. Liq. 2006, 123, 14. (25) Collins, K. D. Biophys. J. 1997, 72, 65. (26) Omta, A. W.; Kropman, M. F.; Woutersen, S. J. Chem. Phys. 2003, 119, 12457. (27) Omta, A. W.; Kropman, M. F.; Woutersen, S.; Bakker, H. J. Science 2004, 301, 347. (28) Na¨slund, L. Å; Edwards, D. C.; Wernet, P.; Bergmann, U.; Ogasawara, H.; Petterson, L. G. M.; Myneni, S.; Nilsson, A. J. Phys. Chem. A 2005, 109, 5995. (29) Zhang, Y.; Furyk, S.; Bergbreiter, D. E.; Cremer, P. S. J. Am. Chem. Soc. 2005, 127, 14505. (30) Kunz, W.; Lo Nostro, P.; Ninham, B. W. Curr. Opin. Colloid Interface Sci. 2004, 9, 1. (31) Bostro¨m, M.; Craig, V. S. J.; Albion, R.; Williams, D. R. M.; Ninham, B. W. J. Phys. Chem. B 2003, 107, 2875. (32) Butler, J. N.; Cogley, D. R. Ionic Equlibrium: Solubility and pH Calculations; John Wiley & Sons: New York, 1998. (33) Mahanty, J.; Ninham, B. W. Dispersion Forces; Academic Press: London and New York, 1976. (34) Bostro¨m, M.; Ninham, B. W. J. Phys. Chem. B 2004, 108, 12593. (35) Tavares, F. W.; D. Bratko; H. W. Blanch; Prausnitz, J. M. J. Phys. Chem. B 2004, 108, 9228 (36) Ninham, B. W.; Yaminsky, V. Langmuir 1997, 13, 2097. (37) Lo Nostro, P.; B. W. Ninham; M. Ambrosi; L. Fratoni; S. Palma; D. Allemandi; Baglioni, P. Langmuir 2003, 19, 9583. (38) Lo Nostro, P.; L. Fratoni; B. W. Ninham; Baglioni, P. Biomacromolecules 2002, 3, 1217. (39) Lo Nostro, P.; J. R. Lopes; B. W. Ninham; Baglioni, P. J. Phys. Chem. B 2002, 106, 2166. (40) Bostro¨m, M.; Kunz, W.; Ninham, B. W. Langmuir 2005, 21, 2619. (41) Butler, J. N.; Cogley, D. R. Ionic Equlibrium: Solubility and pH Calculations; John Wiley & Sons: New York, 1998. (42) Ke´kicheff, P.; Ninham, B. W. Europhysics Lett. 1990, 12, 471. (43) Nylander, T.; Ke´kicheff, P.; Ninham, B. W. J. Colloid Interface Sci. 1994, 164, 136. (44) Pashley, R. M.; Ninham, B. W. J. Phys. Chem. 1987, 91, 2902. (45) Millero, F. J. Compilation of the partial molal volumes of electrolytes at infinite diluition, and the apparent molal volume concentration dependence constants, at various temperatures. In Water and Aqueous Solutions; Horne, R. A., Ed.; Wiley-Interscience: New York, 1972; p 565 (46) Bostro¨m, M.; Ninham, B. W. Langmuir 2004, 20, 7569. (47) Bostro¨m, M.; Ninham, B. W. Biophys. Chem. 2005, 114, 95. (48) Robinson, R. A.; Stokes, R. H. Electrolyte Solutions; Butterworths: London, 1959. (49) Butler, J. N.; Cogley, D. R. Ionic Equlibrium: Solubility and pH Calculations; John Wiley & Sons: New York, 1998. (50) Kielland, J. J. Am. Chem. Soc. 1937, 59, 1675. (51) Jano, I.; Jarvis, T. J. Solution Chem. 2002, 31, 317. (52) Fazary, A. E. J. Chem. Eng. Data 2005, 50, 888. (53) Partanen, J. I. Acta Chem. Scand. 1996, 50, 492. (54) Stern, O. Elektrochemistry 1924, 30, 508. (55) Parsons, R. In Modern Aspects of Electrochemistry; Bockris; J. O. M., Ed.; Butterworths: London, 1954. (56) Grahame, D. C. Chem. ReV. 1947, 47, 441. (57) Lorenz, W.; Salie´ Zeit, G. Phys. Chem. 1961, 218, 259. (58) Schultze, J. W.; Koppitz, F. D. Electrochim. Acta 1976, 21, 327. (59) Conway, B. E. Electrochim. Acta 1995, 40, 1501. (60) Conway, B. E.; Ayranci, E. J. Solution Chem. 1999, 28, 163. (61) Jarvis, N. J.; Schieman, M. A. J. Phys. Chem. 1968, 72, 74. (62) Atkins, P. W. Physical Chemistry, Third edition; Oxford University Press: Oxford, 1986. (63) Tishchenko, P. Y.; Chichkin, R. V.; Il’ina, E. M.; Wong, C. S. Oceanology 2002, 42, 32. (64) Westcott, C. C. pH measurements; Academic Press: Orlando, 1978.
