(12f) The Solvation Model Used to Predict the Salt Effect on Vapor-Liquid Equilibrium

The vapor-liquid equilibrium (VLE) of a mixture containing dissolved salt behaves quite differently from a mixture composed of volatile components alone. When, to a volatile component system in its equilibrium state, a salt is added, the composition of the vapor phase, as well as the temperature or pressure, will change to another equilibrium state. This change, called the salt effect on VLE, means that prediction requires a different approach than that for the no-salt system.

This is because the structure of the salt-containing system, or electrolyte system, is entirely different from that of the no-salt system, or non-electrolyte system. A typical model of the configuration of a non-electrolyte system is the local composition model. In contrast, a typical model of the configuration of an electrolyte system is the solvation model which we originally presented. The configuration of the non-electrolyte system is entirely different from that of the electrolyte system.

The formation of solvate in electrolyte solutions is well known. The solvation numbers for each ion of the salts have been reported by Marcus, who used Stoke's radii. We adopted Marcus's solvation numbers at infinite dilution and confirmed that the numbers determined by our model coincide with Marcus's ion solvation numbers. We have developed a prediction method that can express such a system thermodynamically based on the solvation model. Our solvation model can express the salt effect fully in terms of change of liquid phase composition and total vapor pressure.

Another approach is a modification of the local composition model, which was originally developed for non-electrolyte systems. Some methods using this approach include the Wilson and NRTL equations. This model extends the local composition model to deal with salt systems by adding electrolyte terms. Such equations have some drawbacks or deficiencies.

1. The modification is not congruent: A non-electrolyte model modified for electrolyte solutions is a misapplication, because of the difference in liquid configuration between non-electrolyte and electrolyte solutions.

2. The extended term is the same as the original term: We examined the extended ?electrolyte? NRTL and can present evidence that it is in fact the same equation as in the original ?non-electrolyte? NRTL. This means that the modified local composition equation is not relevant. Furthermore, for most of the observed data we got the same result using the original NRTL as we got using the modified NRTL. Therefore, such equations cannot represent the system completely. They can only fit the data, by optimizing the error. Consequently, they cannot predict the equilibrium.

In our model, we assume that:

1. Both volatile components make solvates with each ion of a salt.

2. The salt is perfectly ionized to anion and cation.

3. Solvates cannot contribute to vapor liquid equilibria.

4. A free volatile component can contribute to the equilibria.

5. The free volatile component is not affected by the solvate.

6. The activity coefficient is divided into two parts:

? expression of non-ideality

?vapor pressure lowering

Thus, the number of liquid molecules that determine vapor-liquid equilibria changes from that in the original composition. (We call this resulting changed composition the ?effective liquid composition?.)

Using these assumptions, our model affords a more versatile method for predicting and correlating VLE for electrolyte solutions. In our model we have two methods for determining solvation numbers. The first method uses data on the elevation of the boiling point of a pure solvent with a salt. The second works directly from the observed data.

Our model can both predict the salt effect on vapor liquid equilibrium and correlate the observed data. We call the first method the predictive method and the second the correlative method. Using the predictive method with data on the boiling point elevation of a pure solvent with salt, we can predict and confirm the breaking of the azeotropic point (for example, in an ethanol, water, and NH4Cl system at 101.3 kPa). At present, no other method can attain this result. We applied the solvation model to predict the methanol, ethanol, water and calcium chloride system, with quite satisfactory results. The solvation numbers used for each volatile component were those determined from the constituent binary volatile components of the salt system.

Our method has the following advantages:

1. The solvation number can be obtained from the vapor pressure lowering data of a pure solvent with salt, or from the salt effect data.

2. The ion solvation number reported independently can also be used.

3. We can predict VLE from the solvation numbers.

Moreover, our model expresses the salt effect more meaningfully. It can explain and demonstrate thermodynamically the mechanism of the salt effect on VLE. A further notable advantage of our model is that the visual interpretation of our parameters reflects more realistically the actual chemical structure of the liquid solution, making it easier to comprehend. We have applied our method for almost all systems reported in the literature with satisfactory results.