Вы находитесь на странице: 1из 11

Chapter 5

Multicomponent Systems
So far we have only dealt with pure substances. Pure substances are dull, because it is our ability to transform between dierent molecules that makes chemistry interesting. In as simple a reaction as A B, there are at least two components to consider. And in systems involving phase transitions, we can exchange molecules between one phase and another (e.g. liquid gas). In this chapter, we will take up multicomponent systems.

5.1

The Gibbs Free Energy for a Mixture

In Ch. 4, we showed that dG = SdT + V dP + dn. When more than one species is present in the system, this equation must be generalized to: dG = SdT + V dP + 1 dn1 + 2 dn2 + , (5.1)

where 1, 2, etc. refer to the dierent species. There is a dierent chemical potential i for each species i. Notice that even though i is the chemical potential for a single species, it in general depends on the number of moles of all the other species in the system as well as on P and T , i.e. i = i (P, T, n1 , n2 , . . .). In Eq.(5.1), P is the total pressure of the system. In Ch. 4, we used the extensivity of G and the fact that it depends on only one extensive variable n to derive the relationship G = n for a pure substance. It is not so clear that this should remain true for a multicomponent system, because in a multicomponent system, I can vary n1 , n2 , etc. independently of each other, and it is no longer obvious that G should be equal to 1 n1+2 n2 + . But we will see in a moment that this is actually correct. To prove this, we use exactly the same argument employed in the last chapter for pure substances. Imagine we have a mixture with n1 moles of species 1, n2 moles of species 2, etc., with corresponding chemical potentials 1 , 2 , etc. The chemical potentials are dened by i = G ni (5.2)
P,T,n1 ,n2 ,...ni1 ,ni+1 ,...

5.2. PARTIAL MOLAR QUANTITIES

If we increase the number of moles of each species i by the same multiple x such that the new number of moles of ni is xni , then the Gibbs free energy of the entire system G should become xG because G is extensive. Now if x is some dierential amount d , we can apply Eq.(5.1) and get: dG = Gd = SdT + V dP + (1 n1 + 2 n2 + )d. (5.3)

Here I have decided to vary the numbers of moles all the species by the same fraction, but dT , dP and d are completely independent and I am free to choose any dierential amount for each of them as I please. But no matter what I choose for each of these, Eq.(5.3) must remain true (as long as the deviations are small enough). The only way Eq.(5.3) is true for every possible dT , dP and d is to have G(T, P, n1 , n2 , . . .) = 1 n1 + 2 n2 + =
i

i ni .

(5.4)

Eq.(5.4) deserves some explanation because it is sometimes confusing for students. While Eq.(5.4) is true, it does not imply the following. For example, if we know that for a mixture with 2 moles of A and 5 moles of B, A = 4J/mole and B = 3J/mole such that G for the mixture is 23J, it does not mean that for another mixture with 3 moles of A and 2 moles of B, G is necessarily (34J+ 23J) = 18J. This is because while Eq.(5.4) is true, it does not say anything at all about how the chemical potential i of species depends on all of the other species in the system. In particular, when the numbers of moles of the other species changes, the chemical potentials change too. Therefore, Eq.(5.4) by itself has very little practical utility at all. However, we will see later that there is a much more useful relationship among the chemical potentials of the dierent components in the system.

5.2

Partial Molar Quantities

Eq.(5.4) is an example of a general property of all extensive thermodynamic variables in an experiment with constant T , P and numbers of moles {nj }. Some other examples are: V (T, P, n1 , n2 , . . .) =
i

V i ni ,

(5.5)

where V i = (V /ni ) is called the partial molar volume of species i, and S (T, P, n1 , n2 , . . .) =
i

S i ni ,

(5.6)

where S i = (S/ni ) is called the partial molar entropy of species i. Copyright c 2009 by C.H. Mak

CHAPTER 5. MULTICOMPONENT SYSTEMS

5.3

The Gibbs-Duhem Equation

We have seen that G = i i ni for a multicomponent system. We can write down the dierential for G based on this as: dG =
i

i dni + ni di .

(5.7)

(Note that this equation does not require constant T or P and is generally true under all conditions.) But as we have seen many times before, dG is also given by dG = SdT + V dP + i i dni . Equating these, we see that it must be true that 0 = SdT + V dP ni di . (5.8)
i

This is known as the Gibbs-Duhem equation. The Gibbs-Duhem equation relates the change in the chemical potential of one component in the system to another. It demands that the changes in {i } in a multicomponent systems should not be independent but are all related to each other. The Gibbs-Duhem equation is extremely useful, because at constant T and P , if we know the chemical potential of one species, we can use Eq.(5.8) to calculate the chemical potential of another species when we vary ni of one or all of them.

5.4

Chemical Potentials for an Ideal Gas Mixture

For a mixture of ideal gases, the chemical potential of each species is easy to determine. Since the components (the gases) do not interact, the Gibbs free energy of the mixture is just G = G1 + G2 + , where Gi is the chemical potential species i would have if it was the only gas in the system. In Sect. 4.5, we saw that for a single ideal gas, lets say species j , at T , P and n, G(T, P, nj ) = G(T, P0 , nj ) + RT ln Pj , P0 (5.9)

where Pj is the pressure of species j , and P0 is some reference pressure. We know the molar Gibbs free energy G for a pure substance is just its chemical potential , so we make the replacement: j (T, P ) = j (T, P0 ) + RT ln Pj P0 (5.10)

for one species when it is by itself in the system. If we now put all the components into the same system, the pressures Pj become the partial pressures. If we dene the mole fraction of specise j as yj = nj /n, Copyright c 2009 by C.H. Mak (5.11)

5.5. CONCENTRATION-DEPENDENCE OF G AND THE ACTIVITY

where n is the total number of moles of all speices, the partial pressure Pj = yj P , where P is the total pressure. Because the components are independent of each other in an ideal gas mixture, j must not depend on the partial pressure of any other species in the system. The total Gibbs free energy is just a sum over all of these species: G=
j

j nj =
j

nj j (T, P0 ) + RT ln

yj P . P0

(ideal)

(5.12)

Even though Eq.(5.12) looks like Eq.(5.4), they are not the same. Eq.(5.4) is generally true for all systems, real or ideal, but Eq.(5.12) says that for an ideal system the chemical potential of each species is independent of the others, which is not generally ture for nonideal systems.

5.5

Concentration-Dependence of G and the Activity

In the last section, we saw that for an ideal mixture, j depends on the mole fraction of j like ln yj . Equivalently, j goes like ln nj , since yj is nj /n. Like the logarithmic dependence of G on the pressure, this logarithmic dependence of on the number of moles of each species j is quite generic and it accounts for the majority of the dependence of j on nj . Therefore, when we come to nonideal mixtures, we would like to retain this feature and dene the activity of species j as: j (P, T ) = (5.13) j (P, T ) + RT ln aj , where j (P, T ) is the chemical potential of species j at some reference concentration or mole fraction for which the activity aj is dened to be 1 ( j is often available in tables and is assumed known). The activity is unitless. For an ideal gas, aj = Pj /P0 . For a real gas, aj = fj /P0 . We will see later that aj is related to the concentration of a solute in a dilute solution. In a liquid or solid solution, aj is related to the mole fraction of each of the components.

5.6

Microscopic Model for a Solution

To illustrate some of the concepts in the last two sections, we will develop a microscopic model to study multicomponent systems, in particular, dilute solutions containing one more more solutes dissolved in a solvent. Once again, we rely on a lattice-type model like the ones used in Ch. 3. This model is shown in Fig. 5.1. We imagine the solution can be divided into cells, each one can be occupied by one of the solute molecules or a solvent particle. In the drawing, we show an example with two types of solutes, 1: red and 2: orange. If a cell is not occupied by a solute molecule, it is occupied by a solvent molecule; therefore, the rest of the cells colored yellow are understood to be lled with Copyright c 2009 by C.H. Mak

CHAPTER 5. MULTICOMPONENT SYSTEMS

solvent molecules. The total number of cells is M , and the number of particles of type 1, 2, . . .. are N1 , N2 , . . .. For the time being, we will assume the solute and solvent particles have no interactions with each other (other than excluded volume interactions), so our model will describe an ideal solution. Most solutions are not very compressible, and so pressure has little eect on their thermodynamics. So to make things a little easier on ourselves, our model will be constant V instead of constant P . For constant (T, P, n1 , n2 , ), we will need A instead of G, but the equilibrium condition derived above in Eq.(5.32) remains correct.

Figure 5.1: Lattice model for a multicomponent solution. In Ch. 3, we derived the Boltzmann distribution law for computing the Helmholtz free energy A for a constant (T, V, n) system. We will use it now to calculate A for our model. Q = eA/kB T = dE (E )eE/kB T , (5.14)

where (E ) is the total number of microstates with energy E , and the integration is carried out over all possible energies. In our case, all microstates have the same energy E = 0 because the model is noninteracting. Therefore, Q = (0). To calculate (0), we use the same argument in Sect. 3.3 to get: Q= M! , (M N1 N2 )!N1 !N2 ! (5.15)

and from this, we obtain the Helmholtz free energy: A = M kB T [(1 x1 x2 ) ln(1 x1 x2 ) + x1 ln x1 + x2 ln x2 ] , (5.16)

where x1 = N1 /M and x2 = N2 /M are the mole fractions of the solutes, and 1 x1 x2 is the mole fraction of the solvent, and we have made use of Stirlings approximation. The chemical potential for species 1, for example, is 1 = For a dilute soultion, x1 , x2 A = RT ln n1 1, so 1 RT ln (x1 ) . Copyright c 2009 by C.H. Mak (5.18) x1 1 x1 x2 . (5.17)

5.6. MICROSCOPIC MODEL FOR A SOLUTION

This justies the motivation behind our denition of the activity in Eq.(5.13), showing that the dependence of on the number of moles of species i is predominantly ln xi . In an ideal solution, ai = xi . Now, the assumption that the solutes have no interaction with the solvent molecules must not be very realistic after all. So we will extend the model by including the possibility of a potential energy term due to solute-solvent interactions. To account for short-range interactions (e.g. dipole-dipole force), we will assume that a solute molecule of species i will interact with a solvent molecule next to it by an energy i . If i < 0, the interaction is favorable; otherwise, the interaction is unfavorable. For simplicity, we will assume no solute-solute interactions are present, so two solute molecules, either of the same type or dierent, haven no interactions with each other even if they are sitting on neighboring cells. Even with the inclusion of this oversimplied solute-solvent interaction, the Q in Eq.(5.14) can no longer be calculated exactly. The reason is that sometimes a solute molecule is surrounded by 4 solvent molecules, if no other solute is next to it. But somtimes a solute is surrounded by only 3 solvent molecules, if one other solute is sitting in a neighboring cell. Other times, it could be surrounded by 2, 1 or no solvent molecules. To determine the energy of each microstate exactly, we must count the number of each of these occurances, which is a highly complicated problem. Therefore, while the model appears to be overly simplistic, no one has ever computed Q exactly for this simple interacting solutesolvent model! In the dilute limit, we can try to use an approximation to compute Q. When Ni M for all the solutes i, the chance for two solute molecules sitting next to each other would be rather small. So we will just assume that for every solute molecule there are always 4 solvent molecules around it. So the total energy of each microstate becomes: E N1 (4 1 ) + N2 (4 2 ). (5.19)

(Notice that if the model is 3-dimensional, the 4 would become a 6. But this will not change any of the essential conclusions below.) So instead of Eq.(5.15), we have Q 4M (x1 1 + x2 2 ) M! exp . (M N1 N2 )!N1 !N2 ! kB T (5.20)

From this, we can compute the Helmholtz free energy A and then the chemical potential in the dilute limit becomes: 1 = RT ln x1 1 x1 x2 + 4Na 1 , RT ln (x1 ) + 4Na 1 , (5.21)

where Na is Avogadros number. The dierence between this and the chemical potential in the non-interacting model in Eq.(5.18) is the addition of the term involving 1 . Even though this result contains the solute-solvent interactions, it is still essentially an ideal solution, because the chemical potential of species 1 Copyright c 2009 by C.H. Mak

CHAPTER 5. MULTICOMPONENT SYSTEMS

does not depend on species 2. For this solution, the activity according to its denition in Eq.(5.13) is just a1 = x1 , (5.22) if we take the reference condition for 1 to be x1 = 1 at temperature T , giving o 1 = 4Na 1 . At not-so-dilute concentrations, nonideality must enter the chemical potentials. The approximation in Eq.(5.19) is good for innitely low concentrations. For higher concentrations, a solute molecule should not be interacting fully with 4 solvent molecules anymore. Instead, a better approximation is to take E N1 (4x
1)

+ N2 (4x

2 ),

(5.23)

where x = 1 x1 x2 is the mole fraction of the solvent molecules. This kind of approximation is known as a mean eld approximation, because the interactions are approximated by the interactions between the solute and the solvent in an average way. With this, Q becomes Q 4M (1 x1 x2 )(x1 M! exp (M N1 N2 )!N1 !N2 ! kB T
1

+ x2 2 )

. (5.24)

In this approximation, the chemical potential turns out to be: 1 RT ln (x1 ) + 4Na
1

(1 2x1 x2 ) ,

(5.25)

which we see is no longer ideal because it depends on x2 , the mole fraction of the other solute. The activity becomes a1 = x1 e4
1 (2x1 x2 )/kB T

(5.26)

if we use the same reference chemical potential o 1 = 4Na 1 as before. We see that at high temperatures, the activity approaches the ideal result a1 x1 . For the second solute, the chemical potential is of course given by a similar formula: 2 RT ln (x2 ) + 4Na
2

(1 2x2 x1 ) .

(5.27)

5.7

Equilibria in Multicomponent Systems

The second law denes the spontaneous direction of natural processes. For a constant (T, P, n) process, the second law says that a spontaneous process must lead to G 0. For a multicomponent system, this is unchanged, but we of course have to include the dependence of G on all the species, so G = G(T, P, n1 , n2 , . . .). We will use the variational statement of the second law to analyze a multicomponent system and use it to determine its equilibrium position. For simplicity, we will rst look at a one-substance/two-phase system. We use the word phase here to refer to a chemically or physically distinct state of the substance, e.g. it could be in a liquid or a solid state, or it could be in a moatomic Copyright c 2009 by C.H. Mak

5.7. EQUILIBRIA IN MULTICOMPONENT SYSTEMS

form or diatomic form of the same substance. Imagine a system at constant (T, P, n). There is a substance X in the system with total number of moles n, but it is distributed in the two phases of the system with n(a) moles in phase a and n(b) moles in phase b, such that the total n = n(a) + n(b) is xed. We can think of this system as a composite of two subsystems, each with the same T and P , but one has n(a) and the other has n(b) moles of the substance, with a channel for particle exchange between them. In the form of a chemical equation, the exchange can be written as: aX(a) bX(b) , (5.28)

where a and b are the stoichiometric coecients.

Figure 5.2: A one-substance/two-phase system at constant T , P and total n.

For the composite system in Fig. 5.2, the total Gibbs free energy is G = (a) n(a) + (b) n(b) , where (a) and (b) are the chemical potentials of the substance in the two phases. Now imagine we have an internal constraint that partitions the substance X into the two phases. Moving this partition, we can make X go from the a phase to the b phase and vice versa, but subject to the constraint that the total n = n(a) + n(b) is xed. Lets say we move the partition so that the change in n(a) is n(a) . According to the chemical equation (5.28), the corresponding change in n(b) must obey the stoichiometric relationship: 1 1 n(b) = n(a) b a (5.29)

These changes will result in a G of the composite system. If G is negative, the move is spontaneous. If we allow the partition to continue to move in the direction of negative G, it will eventually stop when G reaches a minimum. That will be the equilibrium position of the system. If we consider small changes away from this equilibrium position, then dG = (a) dn(a) + (b) dn(b) = b (a) (b) dn(a) , a (5.30)

where we have used the stoichiometric relationship above. Since G is a minimum here, dG must be zero for any small uctuations dn(a) , and the only way for this to be true is a(a) = b(b) , (5.31) Copyright c 2009 by C.H. Mak

CHAPTER 5. MULTICOMPONENT SYSTEMS

which is the equilibrium condition for this system. This analysis can easily be extended to other many-substance/many-phase systems at constant T and P , and the equilibrium condition in general is that for every substance j , the chemical potential of j in each phase multiplied by its stoichiometric coecient must be equal to all other phases: aj j
(a)

= bj j = cj j = ,

(b)

(c)

(5.32)

where aj , bj , cj , . . . are the stoichiometric coecient of j in the a, b, c, . . . phases, respectively. We will make use of this result in the next two chapters to study dierent kinds of equilibria.

5.8

Activities In An Ideal Solution

We argued in Sect. 5.5 that the chemical potential of a solute in a solution should take the form of Eq.(5.13) where the activity a is related to the concentration. We also saw in Sect. 5.6 from the model for an ideal solution that the activity a is simply equal to the mole fraction x, which in the dilute limit is proportional to the concentration of the solute. In this section, we will use the equilibrium condition in the last section to show that if a solution produces a vapor that is an ideal mixture of its components, the solution itself is also ideal. This is a round-about way to show that each component of the solution has a chemical potential that goes like the log of its mole fraction, but this is in fact the only way to do it if we did not have the microscopic model above. When we have a solution of two or more volatile liquids and the solution is closed o to the outside at constant pressure by a piston, each liquid will produce a vapor above the solution. The mixture of vapors is in equilibrium with the solution below, and for every component of the system, the mole fraction of its vapor must be related to the mole fraction of the liquid in the solution via the equilibrium condition Eq.(5.31), where the exchange between the liquid and the vapor phase is simply: X( g ) , (5.33) X( l ) where the stoichiometric coecients are one. Equilibrium requires X = X . When a solution contains one solute at dilute concentration in a volatile solvent, the solvent often obeys Raoults law. We will denote the solute by 1 and the solvent by s. Raoults law says that the vapor pressure of the solvent should be given by: Ps = x s Ps , (5.34)
where Ps is the equilibrium vapor pressure of the pure liquid solvent and xs is the mole fraction of the solvent. This makes sense because the vapor pressure Ps should go to Ps when xs 1, and it should be reduced proportionately when the mole fraction of the solvent is lower than 1. (l ) (g )

Copyright c 2009 by C.H. Mak

10

5.8. ACTIVITIES IN AN IDEAL SOLUTION


(l) (g )

Now the equilibrium condition for the solvent requires that s = s . If the vapor is ideal, its chemical potential is given by:
g) (g ) ( s (T, P ) = s (T, P0 ) + RT ln

Ps . P0

(5.35)

Therefore, the chemical potential of the liquid is identical to this and using Raoults law for the solvent, we now have the chemical potential for the solvent in the liquid solution as:
l) (g ) ( s (T, P ) = s (T, P0 ) + RT ln xs Ps P0

(5.36)

We can rewrite this as


l) (l) ( + RT ln xs , s = s (g )

(5.37)

where we have grouped s (T, P0 ) together with RT ln(Ps /P0 ) to make it equal to the chemical potential of the pure solvent. So we have accomplished our goal, which is to express the chemical potential of the solvent in the liquid as a function of the liquid composition. We see that the activity of the solvent in an ideal solution is just its mole fraction, which is consistent with our conclusion from the microscopic model above. If there are other volatile components other than the solvent and they also obey Raoults law, their activies will similarly be equal to their mole fractions if the solution is ideal. In the dilute limit, the solute(s) usually do not obey Raoults law. We will see examples of this in Ch. 7. Sometimes, the solute may not even be volatile so they dont have a vapor pressure at all. To obatin the activities of the solutes in the dilute limit, we can resort to the Gibbs-Duhem equation. We already know from Eq.(5.37) how the chemical potential for the solvent depends on its mole fraction. Now the Gibbs-Duhem equation also tells us how this is related to the chemical potential of the solute. Assuming we have only one solute, the Gibbs-Duhem equation says: n1 d1 + ns ds = 0 or d1 = xs ds x1 (5.38)

where n1 , 1 and x1 are the number of moles, the chemical potential and the mole fraction for the solute, n2 , s and xs are the same for the solvent, and all of the quantities are for the liquid phase. When the mole fraction of the solvent changes by dxs , it will produce a change in its chemical potential ds , and according to Eq.(5.37): dxs . (5.39) ds = RT xs Since x1 + xs = 1, we can replaced dxs by dx1 , and Eq.(5.38) becomes: d1 = RT dx1 , x1 (5.40)

Copyright c 2009 by C.H. Mak

CHAPTER 5. MULTICOMPONENT SYSTEMS

11

Integrating, we get the relationship we want between the chemical potential of the solute and its mole fraction in the solution: 1 = 1
(l) (l)

+ RT ln x1 ,

(5.41)

where we have restored the superscript (l) to remind ourselves that this equation applies to component 1 in the liquid phase. So we see that the activities of the solute are also equal to their mole fractions in an ideal solution.

Copyright c 2009 by C.H. Mak

Вам также может понравиться