Communications 92 (2018) 56–59

Contents lists available at ScienceDirect

Electrochemistry Communications

journal homepage: www.elsevier.com/locate/elecom

On the dissociation degree of ionic considering effects T M. Landstorfer

Weierstrass Institute for Applied Analysis and Stochastics (WIAS), Mohrenstraße 39, Berlin 10117, Germany

ARTICLE INFO ABSTRACT

Keywords: In this work the impact of solvation effects on the dissociation degree of strong and salts is discussed. pair The investigation is based on a thermodynamic mixture theory which explicitly accounts for the solvation effect. Dissociation degree Based on a remarkable relationship between differential capacity maxima and partial molar volume of in Double layer the solvation number of specific ions in solution is determined. A subsequent investigation of the Solvation shell − electrolytic space charge layer shows that a saturated solution of 1 mol L 1 solvated ions forms near the elec- Mixture theory trode, and we point out some fundamental similarities of this state to a saturated bulk solution. This finding challenges the assumption of complete dissociation, even for moderate concentrations, whereby we introduce an undissociated ion-pair in solution. We re-derive the equilibrium conditions for a two-step dis- sociation reaction, including solvation effects, which leads to a new relation to determine the dissociation de- gree. A comparison to Ostwald's dilution law clearly shows the shortcomings when solvation effects are ne- − glected and we emphasize that complete dissociation is questionable beyond 0.5 mol L 1 for aqueous, mono- valent electrolytes.

1. Introduction double layer structure in the potential region beyond the capacity maximum shows the formation of an ionic saturation layer [19,20,21], Strong electrolytes and salts are frequently assumed to completely which has some fundamental and remarkable similarities to a saturated dissociate into their respective ionic species, for all concentrations up to bulk solution. saturation [1-6]. After Arrhenius introduced the idea of dissociation This is then the starting point for our reflections on the dissociation (and also incomplete dissociation) at the end of the 19th century [7], degree, and it is shown that even for simple salts at concentrations of − Debye and Hückel concluded in 1923 [8] that strong electrolytes al- (0.5–1) mol L 1 the assumption of complete dissociation is question- ways completely dissociate in their respective ionic species [9]. From a able. thermodynamical point of view, this is a very strong a priori assumption and we show within this letter that solvation effects challenge this as- 2. Theory − sumption, especially for concentrations beyond 0.5 mol L 1. The con- cept of incomplete dissociation, ion association, or formation of ion- Consider exemplarily a mono-valent AC of concentration c − pairs in strong electrolytes was re-introduced several times [10-13] and which completely dissociates in solvated anions A and cations C+. − + is again of great scientific interest [14], especially investigated by MD Each ion A and C is assumed to bind κA and κC solvent molecules S in simulations [15,16]. its solvation shell, whereby the number density of free solvent mole- Our investigation presented here is based on a thermodynamic cules S in solution is model [17] which is capable to predict qualitatively and quantitatively nn=−⋅−⋅R κnκn. (1) the double layer capacity of various electrolytes (see Fig. 1). It turns out S S AA C C R that the capacity maxima are determined by the partial molar volume The parameter nS corresponds to the mole density of the liquid R −1 of the anion and the cation, respectively, whereby capacity measure- solvent, i.e. for water nS = 55.4 mol L . The number of mixing parti- ments can be consulted to determine explicit values for different ions. cles is then n = nS + nA + nC, and not the total number of molecules in T R For mono-valent ions in water we find that the partial molar volume of solution, which is nn=++S nnA C. the ionic species is about 45 times larger than the solvent [17]. This suggests that the ionic species are strongly solvated, and based on a 2.1. Entropy of mixing simple relation for the molar volume we can determine the solvation number from a single capacity measurement. An investigation of the For the entropy of mixing this is extremely important. In a solvation

E-mail address: [email protected]. https://doi.org/10.1016/j.elecom.2018.05.011 Received 24 January 2018; Received in revised form 4 May 2018; Accepted 9 May 2018 Available online 01 June 2018 1388-2481/ © 2018 Elsevier B.V. All rights reserved. M. Landstorfer Electrochemistry Communications 92 (2018) 56–59

Fig. 3. Entropy of mixing various solvation numbers κ = κA = κC and an ideal mixture.

nS nA nC skn=− B ⎛ S ln⎛ ⎞ + nA ln⎛ ⎞ + nC ln ⎛ ⎞⎞ . ⎝ ⎝ n ⎠ ⎝ n ⎠ ⎝ n ⎠⎠ (2) For an ideal solution, however, the mixing entities are all solvent molecules in addition to the dissolved ions, which gives for the entropy of mixing

R ideal ⎛ R ⎛nS ⎞ nA nC ⎞ skn=− ln ⎜⎟+ n ln ⎛ ⎞ + n ln⎛ ⎞ . B ⎜⎟S TTTA C ⎝ ⎝ n ⎠ ⎝ n ⎠ ⎝ n ⎠⎠ (3) Fig. 3 displays the difference between the models and shows that the impact of the solvation effect is enormous, even for small solvation numbers.

Fig. 1. Comparison between measured and computed double layer capacity for 2.2. a non-adsorbing and completely dissociated salt KPF6. Source: Top: Fig. 2.a from [18], reprinted with permission of Elsevier. Based on the entropy of mixing (2) it is possible to derive the che- mical potential of the constituents in the liquid, incompressible elec- trolyte [17]. The chemical potential of the free solvent, the solvated anion and the solvated cation is

E μα =+gkTyα B ln()α +⋅− vppα ( ) αSAC ∈ {,,}, (4) where y = nα denotes the mole fraction with respect to the number α n density n = nS + nA + nC of mixing particles, gα the constant molar E Gibbs energy, vα the partial molar volume, and p the bulk pressure.

2.3. Partial molar volume of solvated ions

While the solvation effect decreases the number of free solvent molecules in the mixture, it actually increases the and the partial molar volume of the solvated ions. The molar mass of a solvated ∼ ∼ ion is clearly mmκmAC,,,=+⋅ AC AC S, where mA,C be the mass of the central ion and mS the molar mass of the solvent. A quite similar rela- tion holds for the partial molar volume vA,C of a solvated ion. However, while the molecular mass is conserved during the solvation process, the volume is not necessarily. However, for the sake of this work it is sufficient to assume that the Fig. 2. Sketch of solvation effect in a liquid mixture and the consequence on the partial molar volume of a solvated ion is entities for the entropy of mixing. R −1 vvκvAC,,=+⋅͠ AC AC , S with vnS = (S ) , (5) mixture the mixing entities are now the free solvent molecules, the where v͠A,C is the molar volume of the central ion, vS the molar volume solvated anions and the solvated cations (see Fig. 2), leading to an of the solvent and κA,C the solvation number. This relation allows us entropy of mixing then to deduce the solvation number from a measured value of the partial molar volume.

57 M. Landstorfer Electrochemistry Communications 92 (2018) 56–59

2.4. Determination of the partial molar volume and the solvation number

In [17] it was found that the maxima of the differential capacity C of an electrolyte in contact to a metal are determined by the partial molar volume of the solvated anion and cation, respectively. The investigation is based on the chemical potential function (4), together with diffu- sional equilibrium, incompressibility and the Poisson-momentum equation system [17,22,23]. This yields an expression for C which can be compared to experimental data (see Fig. 1). The continuum model shows an exceptional agreement to experimental data, and we refer to [17] for the full derivation and validation. Surprisingly, the capacity is symmetric for many non-adsorbing salts [18,24,25], whereby the partial molar volumes of the anion and the cation are equal, i.e. vA ≈ vC. The ionic volume for mono-valent salts was found to be (40–50) times larger than the molar volume of the solvent, vA,C ≈(40–50) ⋅ vS, which gives a radius of 3 rv=≈3 (6.6–7.1) Å for solvated ions. AC, 4π AC, Fig. 4. Computed structure of the double layer for a completely dissociated salt Based on the relation (5), the measured value of vA,C suggest with AC. The solid lines display the molar concentration of solvated anions, solvated κ ≈ that each (mono-valent) ion solvates about A,C 45 solvent molecules. cations and the solvent near a metal electrode for an applied voltage of 0.5 V. This value captures all solvent molecules from the first and second The dashed line represents the electrostatic potential in the electrolytic space solvation shell around the central ion. The first solvation shell is well charge layer. understood and agglomerates (3–8) solvent molecules [26]. However, the second shell is rather poor understood but could cover far more mixture. However, from an elementary perspective, the process actually solvent molecules simply from its geometrical arrangement (see Fig. 2). occurs in two steps, initially the dissolution reaction1 Bound solvent molecules may have a slightly smaller volume than bulk R molecules due to microscopic charge-dipole interaction which de- AC|⇌ AC (7) creases their thermal motion. Ab initio methods could probably predict and subsequent the dissociation reaction precise relations between κA,C and vA,C based on a microscopic structure −+ AC(++κκAC )SA ⇌+ C, (8) model. But the goal of this work is not to predict a precise value of κA,C, ff but rather show the general impact of the solvation e ect on the dis- which accounts for the solvation effect. The reaction (8) implies that the κ sociation degree, whereby we proceed the discussion with A,C = 45. constituent CE is actually a species of the liquid mixture and thus has a chemical potential in solution. Whether to term the constituent CE an 2.5. Saturation in the electrochemical double layer ion pair, associated ion, Bjerrium pair or undissociated salt molecule in solution, is thermodynamically insignificant. What is significant, how- With the thermodynamic model of [17] it is possible to compute the ever, is the equilibrium condition the reaction (8) implies, namely structure of the double layer for arbitrary bulk concentrations and ap- plied voltages. It is to emphasize that this structure is obtained from the μAC ++()κκμμμACSAC =+. (9) very same model which predicts the validated capacity data (Fig. 1). With the representation (4) of the chemical potentials μα, this con- Fig. 4 shows a representative computation for and applied voltage of − dition actually rewrites as 0.5 V and a bulk salt concentration of c = 0.0025 mol L 1. ΔgD It turns out that a saturation layer of solvated ions forms near the yyAC· − = ekTB , 1 κκAC+ electrode surface with a concentration of about 1 mol L . In this sa- yyAC · S (10) turation layer the free water molecules are pushed out of the space Δ D κ κ ⋅ − − charge layer in order to ensure the incompressibility of the liquid. What with g = gAC +( A + C) gS gA gC. Introducing the dissociation − δ does this imply for an electrolytic solution of (1–2) mol L 1 bulk con- degree via centration ? nnδcAC==⋅and nAC =− (1 δc ) , (11) − A completely dissociated solution of c = 0.5 mol L 1 AC requires −1 where c is the molar concentration of the salt, leads to about (κA + κC) ⋅ c = 45 mol L solvent molecules, which is almost the R −1 bulk value of water, nS = 55.4 mol L . In this state there are not much δc⋅ yy more free solvent molecules left, which is a similar state as the sa- AC==R , nκδcδcS +−⋅+−⋅2(1 ) (1 ) (12) turation layer of the electrolytic space charge layer. Increasing the salt concentration further would even imply a negative value of the free (1−⋅δc ) y = , solvent molecules, which certainly does not occur. In consequence one AC R nκδcδcS +−⋅+−⋅2(1 ) (1 ) (13) has to requisition the concept of complete dissociation, even at mod- − erate concentrations of (0.1–1) mol L 1. Note that this effect occurs also nκδcR −⋅2 and y = S . when the solvation number is smaller, but then at little higher con- S R nκδcδcS +−⋅+−⋅2(1 ) (1 ) (14) centrations. The equilibrium condition (10) is thus an algebraic constraint

3. Discussion δ2 c ⋅ R (1− δ ) nκδcδcS +−⋅+−⋅2(1 ) (1 ) Requisitioning the concept of complete dissociation requires to state 2κ R ΔgD nκδcδcS +−⋅+−⋅2(1 ) (1 ) the actual bulk reactions occurring during the dissociation process. It is ⎛ ⎞ kTB ⋅ ⎜⎟R −=e0 quite common to write the dissociation reaction of AC as ⎝ nκδcS −⋅2 ⎠ (15) AC|R ⇌+ A−+ C (6) − where R refers to the solid phase and A ,C+ are parts of the electrolytic 1 Note that this process could also require solvent molecules.

58 M. Landstorfer Electrochemistry Communications 92 (2018) 56–59

− or an electrolyte beyond a bulk concentration of 0.5 mol L 1 is ques- tionable, and the degree of dissociation is determined by Eq. (15). Surprisingly, the double layer capacity maxima are correlated to the saturation maximum (or the degree of dissociation) and represent a well defined, experimentally accessible quantity to determine the cru- cial parameters of the dissociation degree. Due to the solvation effect incomplete dissociation (or the formation of ion pairs) is a necessary feature for a thermodynamic consistent theory of electrolytes.

References

[1] R. Haase, Elektrochemie I: Thermodynamik Elektrochemischer Systeme, Springer, 1972. [2] P.W. Atkins, J.D. Paula, Physical , Oxford University Press, 2006. [3] C. Hamann, W. Vielstich, Elektrochemie, Wiley-VCH, 2005. [4] J. Newman, Electrochemical Systems, Prentice Hall, 1973. [5] K. Vetter, S. Bruckenstein, B. Howard, S. Technica, Electrochemical Kinetics, Academic Press, 1967. [6] I. Prigogine, R. Defay, Chemical Thermodynamics, Longmans, 1954. [7] S. Arrhenius, Über die dissociation der in Wasser gelösten stoffe, Z. Phys. Chem. 1 (1887) 631. Fig. 5. Comparison of the computed dissociation degree according to the sol- [8] P. Debye, E. Hückel, Zur Theorie der Elektrolyte. I. Gefrierpunktserniedrigung und vation mixture model (15) with κ = 45 and Ostwald's dilution law (16). verwandte Erscheinungen, Phys. Z. 24 (1923) 185–206. [9] O. Redlich, The dissociation of strong electrolytes, Chem. Rev. 39 (1946) 333–356. [10] Y. Marcus, G. Hefter, Ion pairing, Chem. Rev. 106 (2006) 4585–4621. between the dissociation degree δ and the salt concentration c. The only [11] C.W. Davies, Incomplete dissociation. Introductory paper, Discuss. Faraday Soc. 24 parameters of this relation are the dissociation energy ΔgD and the (1957) 83–86. κ κ [12] R. Heyrovska, Physical electrochemistry of strong electrolytes based on partial solvation number = A,C. dissociation and hydrationa, J. Electrochem. Soc. 143 (1996) 1789–1793. Since we know the solvation number κA,C from the capacity max- [13] H. Bian, X. Wen, J. Li, H. Chen, S. Han, X. Sun, J. Song, W. Zhuang, J. Zheng, imum, we can numerically solve Eq. (15) for various values of ΔgD in Ion clustering in aqueous solutions probed with vibrational energy transfer, PNAS – δ δ 108 (2011) 4737 4742. order to determine the dissociation degree = (c). Fig. 5 shows [14] A. Chen, R. Pappu, Quantitative characterization of ion pairing and cluster for- computations of the dissociation degree from very dilute solutions up to mation in strong 1:1 electrolytes, J. Phys. Chem. B 111 (2007) 6469–6478. high concentrations. Note that Ostwald's dilution law, which is fre- [15] C.J. Fennell, A. Bizjak, V. Vlachy, K.A. Dill, Ion pairing in molecular simulations of aqueous alkali halide solutions, J. Phys. Chem. B 113 (2009) 6782–6791. quently used to compute or approximate the dissociation degree of [16] Y. Luo, W. Jiang, H. Yu, A.D. MacKerell, B. Roux, Simulation study of ion pairing , completely ignores the solvation effect and the concept of free in concentrated aqueous salt solutions with a polarizable force field, Faraday solvent molecules. The corresponding constraint of Ostwald's dilution Discuss. 160 (2013) 135–224. law reads [17] M. Landstorfer, C. Guhlke, W. Dreyer, Theory and structure of the metal-elec- trolyte interface incorporating adsorption and solvation effects, Electrochim. Acta – δ2 c ΔgD 201 (2016) 187 219. kTB [18] G. Valette, Double layer on silver single-crystal electrodes in contact with electro- ·eR −= 0 1−δ nS (16) lytes having anions which present a slight specific adsorption: part I. The (110) face, J. Electroanal. Chem. 122 (1981) 285–297. and significantly underestimates the dissociation degree, especially for [19] M. Kilic, M. Bazant, A. Ajdari, Steric effects in the dynamics of electrolytes at large higher concentrations (see comparison in Fig. 5). applied voltages. II. Modified Poisson-Nernst-Planck equations, Phys. Rev. E 75 (2007) 021503. [20] I. Borukhov, D. Andelman, H. Orland, Steric effects in electrolytes: a modified 4. Summary Poisson-Boltzmann equation, Phys. Rev. Lett. 79 (1997) 435–438. [21] J. Bikerman, XXXIX. Structure and capacity of electrical double layer, Philos. Mag. – For a large dissociation energy, i.e. ΔgD > 0.1 eV, which corresponds 7 (33) (1942) 384 397. [22] W. Dreyer, C. Guhlke, R. Müller, Overcoming the shortcomings of the Nernst- to a salt or a very strong electrolyte, Ostwald's dilution law predicts Planck model, Phys. Chem. Chem. Phys. 15 (2013) 7075–7086. complete dissociation for arbitrary concentrations. In contrast, the [23] W. Dreyer, C. Guhlke, M. Landstorfer, A mixture theory of electrolytes containing ff – solvation effect and its consistent incorporation in the thermodynamic solvation e ects, Electrochem. Commun. 43 (2014) 75 78. −1 [24] G. Valette, Double layer on silver single crystal electrodes in contact with electro- theory requires incomplete dissociation beyond 0.5 mol L (see lytes having anions which are slightly specifically adsorbed: part II. The (100) face, Fig. 5). Origin of this effect is simply that not enough free solvent J. Electroanal. Chem. 138 (1982) 37–54. molecules are present anymore to shift the reaction equilibrium of Eq. [25] G. Valette, Double layer on silver single crystal electrodes in contact with electro- fi κ lytes having anions which are slightly speci cally adsorbed: part III. The (111) face, (8) towards the ions in solution. For a solvation number of A,C =45 J. Electroanal. Chem. 269 (1989) 191–203. D and a dissociation energy of Δg = 0.1 eV we find that dissociation [26] J.F. Hinton, E.S. Amis, Solvation numbers of ions, Chem. Rev. 71 (1971) 627–674. − − degree is δ = 0.99 for 0.5 mol L 1, δ = 0.44 for 1 mol L 1 and δ = 0.19 − for 2 mol L 1. Hence the assumption of complete dissociation of a salt

59