Statistics of Adsorption - Some Special Topics: Surfaces of Liquids and Solutions - Liquid-State Physical Chemistry: Fundamentals, Modeling, and Applications (2013)

Liquid-State Physical Chemistry: Fundamentals, Modeling, and Applications (2013)

15. Some Special Topics: Surfaces of Liquids and Solutions

15.5. Statistics of Adsorption

Apart from a thermodynamic description of liquid surfaces, another route is to use a statistical mechanical approach. We do so briefly in this section, limiting ourselves again to surfaces, and to that purpose consider a binary system for which we denote the bulk composition with A1−xBx and the surface composition A1−yBy. We again use a lattice-type model, and assume that only the first molecular layer has a composition different from the bulk. As before, the components are indicated by subscript i, while the surface and bulk phase are indicated by superscripts (σ) and (α), respectively. We start with an ideal solution and thereafter consider the influence of nonideality.

We recall that the chemical potential of component i is given by

(15.70) c15-math-0070

where c15-math-5036 and c15-math-5037 are the standard chemical potentials of the pure components in the bulk and the surface, respectively, while x and y denote the mole fractions. Further, at equilibrium we have c15-math-5038. From these equations we easily obtain

(15.71) c15-math-0071

The Helmholtz energies are then given by

(15.72) c15-math-0072

For a one-component system the surface tension is given by c15-math-5039, so that

(15.73) c15-math-0073

where ai = A/Ni denotes the area per molecule. If we subtract Eq. (15.71-2) from Eq. (15.71-1) and Eq. (15.72-2) from Eq. (15.72-1) and insert Eq. (15.73) for both components, we obtain

(15.74) c15-math-0074

This is the adsorption equation for ideal solutions.

We now introduce interactions and “change gear” to statistics, employing the grand (canonical) partition function Ξ in connection with the lattice model. Note that Ξ is the proper partition function to use, since we keep μi, N and Vconstant (see Chapter 5). The configurational partition function of the surface layer Q(σ) is given by

(15.75) c15-math-0075

where

(15.76) c15-math-0076

The energy U(σ) is evaluated as for a regular solution (in the zeroth approximation) and reads, using as before the interaction energy 2w = 2wABwAAwBB,

(15.77) c15-math-0077

Here, c15-math-5040 is the coordination number for “bonds” in the surface layer with composition A1−yBy, and c15-math-5041 is the coordination number for “bonds” between the surface layer and the first bulk layer, so that the total surface coordination number reads c15-math-5042 (e.g., for a FCC (111) plane c15-math-5043 and c15-math-5044). The grand partition function becomes

(15.78) c15-math-0078

We now use the maximum-term method (see Justification 5.4), replacing the sum by its largest term and obtain

(15.79) c15-math-0079

The equilibrium condition is

(15.80) c15-math-0080

and leads to

(15.81) c15-math-0081

From c15-math-5045 and Eq. (15.71) we obtain

(15.82) c15-math-0082

so that combining Eq. (15.81) and (15.82) gives

(15.83) c15-math-0083

This is the equivalent of Eq. (15.74) in the zeroth approximation. We see that the surface tension term −(aBγBaAγA) is replaced by ½(z z(σ))(wBBwAA), while the interaction term with w is obviously absent for the ideal solution.

Estimating wii from the heat of vaporization Li in the nearest-neighbor approximation Li = −½zNAwii (with NA = Avogadro's constant) and w from the heat of solution LAB of component A in component B (or vice versa from LBA), we have

(15.84) c15-math-0084

Note that the heat of vaporization is counted positive, while the bond energy is counted negative. In the zeroth approximation of the regular solution theory ΔLAB = ΔLBA. Experimentally, this condition is only exceptionally fulfilled, which indicates a need for further improvement. However, the approach does provide a clear basis for a picture in which the surface tension – that is, the surface Helmholtz energy – is estimated as, with again NA as Avogadro's number,

(15.85) c15-math-0085

Using this approximation we obviously neglect entropy terms and have identified ½zwii with c15-math-5076 and ½z(σ)wii with c15-math-5046. The discussion in Section 15.2 showed that, numerically, the difference is far from negligible. The fraction of missing “bonds” (zz(σ))/z, as also indicated in Section 15.2, is ∼1/4.