Activity Coefficient Models
- Activity coefficient models consist of two experimental terms (fi and I) and several parameters.
- Given a model, its parameters, and one of its experimental terms, this tool solves for the missing one.
- To use the tool, one of its experimental term fields must be left empty.
- To empty a field, just double click it.
- If no field is left empty, the tool will randomly empty one and recompute its value based on the values of the other fields.
- This tool supports the most frequently used activity coefficient models:
- Debye-Hückel Limiting Law
- Debye-Hückel Extended Limiting Law
- Truesdell-Jones (aka, B-dot equation)
- Before using the tool please, read the "What is computed?" section thoroughly.
- If working with Truesdell-Jones Model, called B-dot equation in EQ3/EQ6 program (Pearson & Berner, 1991; Helgenson, 1969; Bethke, 1996; Helgenson, Kirkham, & Glowers, 1981), you may want to gather parameters information from USER'S MANUAL FOR WATEQ4F, a standard reference program. A current version of the program is available for download from USGS.GOV. An early report on WATEQF with FORTRAN source code is also available. Our online tool, written in PHP, was inspired in this program. It uses a light, compact source code for easily cross-mapping I and fi.
- This tool supports the most frequently used activity coefficient models:
- Activity coefficient models consist of two experimental terms (fi and I) and several fitting parameters, where
fi is the activity coefficient of ion i in a solution.
I is the solution ionic strength, typically given in molar (M) or molal (m) units.
A is a fitting parameter equal to 1.82x106·(εrT)-3/2.
B is a fitting parameter equal to 50.3·(εrT)-1/2.
C is a model fitting parameter, in L/mol so molar ionic strength units cancel out. It is usually set between 0.2 and 0.5. Our tool uses 0.3 by default.
ai is the ion size (not radius) in nanometers, nm (1 nm = 10 angstrom).
a°i is a model fitting parameter, in nm.
bi is a model fitting parameter, in L/mol, and usually set between 0.1 and 0.3. Our tool uses 0.2 by default.
εr is the solvent relative permittivity (dielectric constant), which is the ratio of the electric field strength in vacuum to that in a given medium. For pure water at 25° C (298.15 kelvins) Malmberg & Maryoff (1956) reported a value of 78.3, which is not that far from the IUPAC accepted value of 78.36 ±0.05. (Barthel, Krienke, & Werner). NIST reports a value of 78.4 (Fernandez, Mulev, Goodwin, & Levelt Sengers, 1995), which is our tool default value.
T is the temperature in kelvins at standard conditions.
zi is the valence (charge) of ion i, which can be treated as an integer parameter because computing a fractional zi has no chemical meaning in the context of activity coefficient models.
The molar ionic strength of a solution, I, is a function of the sum of products of all ion concentrations, ci, present in solution times the square of their ionic charge (valence). The sum is divided by 2 as both cations and anions are included; i.e.
I = 1/2*Σcizi2
I can be molar (mol/L) or molal (mol/kg water). To avoid confusion the units should be stated explicitly, or be used unitless (Solomon, 2001). In this work we explicitly assume that I is in mol/L.
Debye-Hückel theory of electrolytes was initially based on several assumptions: among others, that ions in a diluted solution dissociate completely, are symmetric, have a negligible ionic radius, and do not interact with other ions or the solvent. Several extended models have been proposed since then, each applicable up to an upper bound of ionic strength. These are given in Figure 1.
Figure 1. Activity coefficient models with estimated upper bound values of ionic strength, I, where activity coefficients, fi, steadily decrease.
According to Figure 1, Debye-Hückel Limiting Law model is applicable to ions in highly diluted solutions (I < 0.005 M). By contrast, Debye-Hückel Extended Limiting Law and Güntelberg are used for ions in moderately diluted solutions (I < 0.1 M), while Davis's applies to solutions with relatively high ionic strength (I < 0.5 M) and Truesdell-Jones's to even more concentrated solutions (I < 1 M); for instance, sea water whose major components are Na+, K+, Mg2+, Ca2+, Cl-, and SO42- (Ryan, 2004).
Finally, for highly saline solutions (1 < I < 20 M), the Pitzer Equation, a more sophisticated model is recommended. This model, however, requires a lot of additional parameters (virial coefficients) and is not included in our tool.
From the above discussion we may conclude, as Henry Freiser (1992) asserted that
"The limiting law is truly limiting and not a very useful law, however. Some chemists say it is only useful in 'slightly dirty water'."
In general as I decreases, the models converge to Debye-Hückel Limiting Law. See Figure 2, inspired by the one given at Aqion.
Figure 2. Parameterized relationships between several activity coefficient models. For highly diluted solutions (i.e., at low I) they all converge to the Debye-Hückel Limiting Law. I is assumed in mol/L.
Thus, arbitrarily applying Debye-Hückel Limiting Law, or any ionic strength model, is contraindicated as one may end computing meaningless activity coefficients, or the wrong ones.
- Interpretation of results
A ≃ 0.51 M-1/2 and B ≃ 3.3 M-1/2nm-1 for diluted aqueous solutions at 25° C; i.e., T = 298.15 in the Kelvin scale. C doesn't have a generally accepted value and is usually set between 0.2 < C < 0.5 (Chembuddy.com, 2017). Actually Davis used 0.2 and 0.3 (Davis, 1962). MINEQL program uses 0.2 and there is an option for its use in PHREEQE with C = 0.3. (Pearson & Berner, 1991).
The ion size parameter, ai is not the ionic radius, but a measure of the diameter of a hydrated ion and has to be known in order to calculate its activity coefficient. By contrast, a°i and bi are ion-specific parameters fitted from mean salt activity coefficient data.
In many applications, ai = 0.3 nm; therefore, Bai ≃ 1 so we can simplify the several models by setting a°i = ai = 1/B, as can be seen from Figure 2. One can also simplify the models by setting ai = bi = 0. Notice that the prevailing factor affecting activity coefficients and the selection of a model is the solution ionic strength.
Another reasons for using simplified activity coefficient models comes from the fact that the ion size parameter is not well known for all ions, particularly the complex ones, and because for the most commonly known ions this parameter is between 0.3 to 0.9 nm, as can be seen from Table 1, derived from Kielland (1937).
Table 1. Common ai values in water ai (nm) ion 0.9 H3O+, Al3+, Fe3+, La3+, Ce3+ 0.8 Mg2+, Be2+ 0.7 COO- 0.6 Ca2+, Zn2+, Cu2+, Fe2+, Sn2+, Mn2+ 0.5 Ba2+, Sr2+, Pb2+, CO32- 0.4 Na+, HCO3-, H2PO4-, HPO42-, PO43-, SO42-, CH3COO-, 0.3 K+, Ag+, NH4+, OH-, NO3-, Cl-, Br-, I-, HS-,
If using Truesdell-Jones model, one would also need to know the ion-specific parameters, determine these experimentally, or assume some values. Table 2 lists these for some common ions (Hellevang, 2006).
Table 2. Truesdell-Jones a°i and bi values for some ions in water ion a°i (nm) bi (L/mol) Na+ 0.43 0.06 K+ 0.37 0.01 H+ 0.48 0.24 Mg2+ 0.55 0.22 Fe2+ 0.51 0.16 Ca2+ 0.49 0.15 Al3+ 0.67 0.19 Cl- 0.37 0.01 OH- 1.07 0.21 HCO3- 0.54 0 CO32- 0.54 0
If we define x = I1/2, it can be demonstrated that Davis and Truesdell-Jones models lead to third degree polynomial expressions in x so the nonlinearity of these models can be understood on mathematical terms.
This explains why mapping I to fi with these models is faster than mapping fi to I. Since a close form function for I does not exist for these models, one would need to solve for I using successive approximations or residual analysis.
An important drawback, common to most activity models, is that at high I they all break down on chemical and mathematical grounds. Let see why.
At high I, ions can no longer be treated as point charges as they can interact with the solvent or other ions (e.g. by forming ion pairs). The ionic radius, negligible in diluted solutions, can then be comparable to the radius of the ionic atmosphere. As a result of this, fi increases for relatively high ionic strengths and the models predict that some values of fi can be mapped to more than one I. Thus, the models break down completely.
- Lab techs as well as chemistry teachers and their students.
- Consider an aqueous solution 0.050 M in Na2SO4 and 0.020 M in KCl at 25° C.
- Calculate its ionic strength.
- Calculate the activity coefficients of all ions in solution using Davis and Truesdell-Jones models.
- Construct a table listing these two models and the computed I and fi values for all the ions in solution. Explain the results.
- Can the other models be useful in this case? Why?
- Using all appropriate activity coefficient models, calculate the activity coefficient of Pb2+ in
- an aqueous solution 0.005 M in Pb(NO3)2 at 25° C. Compare the results from each model. Which models produce worst results and why?
- an aqueous solution 0.005 M in Pb(NO3)2 and 0.040 M in KNO3 at 25° C. Which models can(not) be used and why?
- Consider the following parameters pertaining to Mg2+ in an aqueous solution at 25° C:
- aMg2+ = 0.8 nm
- a°Mg2+ = 0.55 nm
- bMg2+ = 0.2 L/mol
- C = 0.3 L/mol
- zMg2+ = 2
- Barthel, J. M. G., Krienke, H., & Kunz, W. (1998). Physical Chemistry of Electrolyte Solutions: Modern Aspects. Springer, New York. Chapter 7. Appendix, p 373.
- Bethke, C. M. (1996). Geochemical Reaction Modeling: Concepts and Applications. Oxford University Press.
- Chembuddy.com (2017). Buffer Calculation.
- Davis, C. W. (1962). Ion Association. Butterworth, pp 190. Washington, D.C. See also Book Review. Science Magazine.
- Fernandez, D. P., Mulev, Y. Goodwin, A. R. H., & Levelt Senger, J. M. H.(1995). Database for the Static Dielectric Constant of Water and Steam. J. Phys. Chem. Ref. Data, Vol 24, No. 1.
- Freiser, H. (1992). Concepts & Calculations in Analytical Chemistry. Chapter 3, p. 41. CRC Press, Boca Raton, FL.
- Helgenson, H. C. (1969). Thermodynamics of hydrothermal systelTIS at elevated temperatures and pressures. American Journal of Science, v. 267, p. 729-804.
- Helgenson, H. C., Kirkham, D. H., & Glowers, G. C. (1981). Theoretical Prediction of the Thermodynamic Behavior of Aqueous Electrolytes at High Pressures and Temperatures. American Journal of Science, v. 281, p. 1249-1516.
- Hellevang, H. (2006). Appendix A: Reactions and equations. Bora, UiB.
- Kielland, J. (1937). Individual Activity Coefficients of Ions in Aqueous Solutions. J. Am. Chem. Soc., 1937, 59 (9), p 1675-1678.
See also http://pubs.acs.org/doi/abs/10.1021/ja01288a032.
- Malmberg, C. G. and Maryoff, A. A. (1956). Dielectric Constant of Water from 0° to 100° C. Journal of Research of the National Bureau of Standards, Vol 56, No. 1, pp 1-8.
- Pearson, F. J. & Berner, U. (1991). Nagra Thermochemical Database. Nagra Thermochemical Data Base, I. Core Data. Nagra Technical Report NTB 91-17.
- Ryan, D. K. (2004). Chemical Osceanography. Lecture 3. University of Massachusetts, Lowell.
- Solomon, T. (2001). The Definition and Unit of Ionic Strength. J. Chem. Educ., 78 (12), p 1691.
Contact us for any suggestion or question regarding this tool.