The exposition begins with a virial expansion of the excess
Gibbs free energy :\frac{G^{ex}}{W_wRT} = f(I) +\sum_i \sum_j b_ib_j\lambda_{ij}(I)+\sum_i \sum_j \sum_kb_ib_jb_k\mu_{ijk}+\cdots
Ww is the mass of the water in kilograms,
bi, bj ... are the
molalities of the ions and I = \tfrac{1}{2} \sum_i b_i {z_i}^2 is the molal ionic strength. The first term,
f(I) represents a Debye–Hückel extended law (see below). The quantities
λij(I) represent the short-range interactions in the presence of solvent between solute particles
i and
j. This binary interaction parameter or second virial coefficient depends on ionic strength, on the particular species
i and
j and the temperature and pressure. The quantities
μijk represent the interactions between three particles. Higher terms may also be included in the virial expansion. Next, the free energy is expressed as the sum of
chemical potentials, or partial molal free energy, :G= \sum_i \mu_i\cdot N_i = \sum_i \left (\mu^0_i +RT \ln b_i\gamma_i \right )\cdot N_i and an expression for the activity coefficient is obtained by differentiating the virial expansion with respect to a molality b. :\ln \gamma_i = \frac{\partial(\frac{G^{ex}}{W_wRT})}{\partial b_i} =\frac{z_i^2}{2}f' +2\sum_j \lambda_{ij}b_j +\frac{z_i^2}{2}\sum_j\sum_k \lambda'_{jk} b_jb_k + 3\sum_j\sum_k \mu_{ijk} b_jb_k+ \cdots And molal
osmotic coefficient: :\phi-1=\left(\sum_ib_i\right)^{-1}\left[If'-f + \sum_i\sum_j\left(\lambda_{ij}+I\lambda'_{ij} \right)b_ib_j +2\sum_i\sum_j\sum_k \mu_{ijk} b_ib_jb_k + \cdots\right] However, these forms are not used directly, because it is not possible to determine the coefficients independently due to charge neutrality constraints. Instead the sums are re-worked in terms of observable (charge-neutral) combinations and some choices are made about their functional forms, which we will see below: • The \lambda_{ij}-related terms for salt pairs are gathered together into observable combinations (B_{ca} values) and then a specific exponential functional form is imposed on them in terms of \beta parameters, see below. • The \mu_{ijk}-related terms for salt pairs are likewise gathered into observable combinations (C_{ca} values). • All ternary interactions \mu_{ijk} involving three ions of same sign are set to 0. • For mixed electrolytes, new combinations appear: \Phi_{cc'}, \Phi_{aa'}, and ternary interactions \psi_{cc'a}, \psi_{caa'}. A special functional form is applied to \Phi in cases of unsymmetrical mixing (two ions present with same sign of charge but different magnitude). • For electrolytes combined with neutral solutes, \lambda_{ij} remain as-is, but they are assumed to be independent of ionic strength. These observable combinations then provide a set of free parameters which are then empirically fit to experimental data.
Pure electrolyte case Consider a simple electrolyte ''M'
p'X''
q with molal concentration
m, dissolved to ions
Mz+ and
Xz−, with ionic molal concentrations b_M = pm and b_X = qm. The Pitzer parameters f^\phi, B^\phi_{MX} and C^\phi_{MX} are defined as :f^\phi=\frac{f'-\frac{f}{I}}{2} :B^\phi_{MX}=\lambda_{MX}+I\lambda'_{MX} +\left(\frac{p}{2q}\right)\left(\lambda_{MM}+I\lambda'_{MM}\right)+\left(\frac{q}{2p}\right)\left(\lambda_{XX}+I\lambda'_{XX}\right) :C^\phi_{MX} =\left[\frac{3}{\sqrt{pq}}\right] \left(p\mu_{MMX}+q\mu_{MXX}\right). (Terms involving \mu_{MMM} and \mu_{XXX} are not included as interactions between three ions of the same charge are unlikely to occur except in very concentrated solutions.) With these definitions, the expression for the (molal-basis)
osmotic coefficient becomes :\phi-1=|z^+z^-|f^\phi+m\left(\frac{2pq}{p+q}\right)B^\phi_{MX} +m^2\left[2\frac{(pq)^{3/2}}{p+q}\right]C^\phi_{MX}. A similar expression is obtained for the (molal-basis)
mean activity coefficient: :\ln \gamma_\pm =|z^+z^-|f^\gamma+m\left(\frac{2pq}{p+q}\right)B^\gamma_{MX} +m^2\left[2\frac{(pq)^{3/2}}{p+q}\right]C^\gamma_{MX}, where f^\gamma, B^\gamma_{MX} and C^\gamma_{MX} are related to f^\phi, B^\phi_{MX} and C^\phi_{MX}, but distinct. Finally, some forms are imposed on the coefficients based on a mixture of theoretical and empirical observations: The term
fφ is defined to be an extended Debye–Hückel term: :f^\phi = -A_\phi \frac{I^{1/2}}{1 + b I^{1/2}} with A_{\phi} being calculated in terms of the solvent dielectric constant, and b=1.2~\mathrm{kg}^{1/2}\mathrm{mol}^{-1/2} is defined as a universal empirical parameter (note this b should not be confused with the molality b_i). The
B parameter was found empirically to show an ionic strength dependence which could be expressed as :B^\phi_{MX}=\beta^{(0)}_{MX} + \beta^{(1)}_{MX} e^{-\alpha \sqrt I}, or sometimes with a second term which can often capture ion pairing effects without requiring explicit ion association accounting: :B^\phi_{MX}=\beta^{(0)}_{MX} + \beta^{(1)}_{MX} e^{-\alpha_1 \sqrt I} + \beta^{(2)}_{MX} e^{-\alpha_2 \sqrt I}. (with specific values of \alpha_1 and \alpha_2 being chosen depending on the ion charges). The empirical Pitzer data tables therefore list \beta^{(0)}_{MX}, \beta^{(1)}_{MX}, and sometimes \beta^{(2)}_{MX} (usually 0), whereas C^\phi_{MX} is directly tabulated.
Note on ion association: If ion pairing is included as an
ion association equilibrium with an explicit separate solute species (with its own separately-accounted concentration), then the empirical values \beta^{(n)} and C^\phi will change completely. Moreover, this choice fundamentally redefines the meaning and numerical values of ionic molalities, ionic strength, mean activity coefficients, and even the osmotic coefficient. Ionic mean activities and solvent activity are, however, thermodynamically independent of this accounting choice.
General case: mixed electrolytes, neutral solutes, and single-ion activities Pitzer defines the above pure electrolyte case (yielding osmotic coefficient and mean activity) to be mathematically simple, then 'works backwards' to deduce the general case in a way that is consistent with the pure electrolyte case. Pitzer thus arrives at the following Gibbs energy: \begin{aligned} \frac{G^{ex}}{W_w RT} &= f(I) \\ &\quad + 2 \sum_{c} \sum_{a} b_c b_a \left[ B_{ca} + \left( \sum_{c} b_c z_c \right) C_{ca} \right] \\ &\quad + \mathop{\sum\sum}_{c where c indices are positive ions (cations), a are negative ions (anions), and n are neutral solutes. Note that B_{ca}, \Phi_{cc'}, \Phi_{aa'} are also functions of ionic strength. This form (with the \dots truncated) is then the actual master thermodynamic equation underlying Pitzer theory, and by differentiation it yields all other observable quantities (activity coefficients, osmotic coefficients). This expression omits various terms proportional to the total charge \sum_{i} z_i b_i, and therefore it yields different single-ion activities compared to the original G^{ex}, but only in an unobservable way. The raw Pitzer single-ion activities are not experimentally observable on their own, and in practice they are combined into observables (like the mean activity above), or they are at least transformed to obey well-known activity conventions (such as the MacInnes convention). The various terms are all consistent with the pure electrolyte case. For example, the full Debye-Huckel term is chosen to be f(I) = -(4 I A_{\phi}/b)\ln(1 + b I^{1/2}) which is consistent with the f^\phi above. Expressions for the interaction coefficients B_{ca}, C_{ca}, \Phi, and \psi can be found in standard references. It is crucial to note that the parameters appearing in the Gibbs energy differ from the tabulated parameters often labeled with \phi or \gamma superscripts. For example, B_{ca} is the
fundamental interaction parameter, while B_{ca}^\phi and B_{ca}^\gamma are derivatives used for osmotic and activity coefficient calculations respectively. == Commentary ==