ثرمو 13
ثرمو 13
ثرمو 13
Thermodynamic Formulations
for Vapor/Liquid Equilibrium
The objective of this chapter is to apply the framework of solution thermodynamics devel-
oped in Chapter 10 to the specific situation of vapor/liquid equilibrium (VLE), as introduced
qualitatively in Chapter 12. Because of the practical importance of distillation as a means of
separating and purifying chemical species, VLE is the most studied type of phase equilibrium.
Approaches developed for analyzing VLE also provide the foundation for most analyses of
liquid/liquid equilibrium (LLE), vapor/liquid/liquid equilibrium (VLLE), and combined phase
and reaction equilibrium, as considered in Chapter 15.
The analysis of VLE problems in the present chapter begins by developing a general
formulation of such problems in terms of vapor-phase fugacity coefficients and liquid-phase
activity coefficients. For VLE at low pressure, where the gas phase approaches the ideal-gas state,
simplified versions of this formulation are applicable. For those conditions, activity coefficients can
be obtained directly from experimental VLE data. These can then be fit to mathematical models.
Finally, the models can be used to predict activity coefficients and VLE behavior for situations
where experiments have not been performed. Thus, the analyses presented in this chapter allow
for the efficient correlation and generalization of the observed behavior of real physical systems.
450
13.1. Excess Gibbs Energy and Activity Coefficients 451
∙ Introduce several excess Gibbs energy and activity coefficient models and the fitting of
model parameters to experimental VLE data
∙ Perform VLE calculations under conditions where the complete gamma/phi formulation
is required
∙ Show that residual properties and excess properties can also be evaluated from cubic
equations of state
∙ Demonstrate the formulation and solution of VLE problems using cubic equations of state
The foundation for VLE calculations was laid in Chapter 10, where Eq. (10.39) was
shown to apply for the equilibrium of pure species:
f iv = f i l = f isat
(10.39)
and Eq. (10.48) was shown to apply to the equilibrium of species in mixtures:
fˆ iv = fˆ il (i = 1, 2, . . . , N ) (10.48)
We recall also the definitions of fugacity coefficients, as given by Eqs. (10.34) and (10.52).
From the latter we can write for the fugacity coefficient of species i in a vapor phase:
f ˆ iv = ϕˆ i v y i P (13.1)
An analogous equation can be written for the liquid phase, but this phase is often treated differently.
By difference,
fˆ i
γ i ≡ _
(13.2)
x i fi
( RT ) RT
nG E nV E nH E
= _
d _ dP − _
2 dT + ∑ ln γ i dn i (13.4)
RT i
The generality of this equation precludes its direct practical application. Instead, use is made
of restricted forms, which are written by inspection:
[ ] [ ]
V E
___ ∂ (G E / RT ) H E ∂ (G E / RT )
= _________
(13.5) ___
= − T _________ (13.6)
RT ∂ P RT ∂ T
T, x P, x
[ ]
∂ (n G E ∕ RT )
ln γ i = __________ (13.7)
∂ n i P, T, n j
Equations (13.5) through (13.7) are analogs of Eqs. (10.57) through (10.59) for residual
properties. Whereas the fundamental residual-property relation derives its usefulness from its
direct relation to experimental PVT data and equations of state, the fundamental excess-property
relation is useful because VE, HE, and γi are all experimentally accessible. Activity coefficients
are found from vapor/liquid equilibrium data, as discussed in Sec. 13.5, while VE and HE
values come from mixing experiments, as discussed in Chap. 11.
Equation (13.7) demonstrates that ln γi is a partial property with respect to GE∕RT. It
is the analog of Eq. (10.59), which shows the same relation of ln ϕ ˆ i to GR∕RT. The partial-
property analogs of Eqs. (13.5) and (13.6) are:
V¯ i E H¯ E
( ∂ P ) ( ∂ T )
∂ ln γ i ∂ ln γ i
_____
= ___
(13.8) _____
= − ____
i 2 (13.9)
T, x
RT P, x RT
These equations allow the calculation of the effect of pressure and temperature on the activity
coefficients.
The following forms of the summability and Gibbs/Duhem equations result from the
fact that ln γi is a partial property with respect to GE∕RT:
_G E
= ∑ x i ln γ i(13.10)
RT i
∑
x i d ln γ i = 0 (const T, P)(13.11)
i
Just as the fundamental property relation of Eq. (10.54) provides complete property
information from a canonical equation of state expressing G∕RT as a function of T, P, and
composition, so the fundamental residual-property relation, Eq. (10.55) or (10.56), provides
13.2. The Gamma/Phi Formulation of VLE 453
complete residual-property information from a PVT equation of state, from PVT data, or from
generalized PVT correlations. However, obtaining complete property information requires,
in addition to PVT data, the ideal-gas-state heat capacities of the species that comprise the
system. In complete analogy, the fundamental excess-property relation, Eq. (13.4), provides
complete excess-property information, given an equation for GE∕RT as a function of its
canonical variables, T, P, and composition. However, this formulation represents less-complete
property information than does the residual-property formulation, because it tells us nothing
about the properties of the pure constituent chemical species.
Rearranging Eq. (13.2), the definition of the activity coefficient, and writing it for species i in
the liquid phase gives:
fˆ i l = x i γil f i l
Substitution in Eq. (10.48) for f ˆ il by this equation and for fˆ i v by Eq. (13.1) yields:
y i Φ i P = x i γ i Pisat
(i = 1, 2, . . . , N )
(13.13)
where
ϕi [ ]
ϕˆ i v V il (P − Pisat
)
Φ i ≡ ____
sat
exp ___________
−
RT
ϕˆ i v
Φ i ≡ ____
(13.14)
ϕisat
The vapor pressure of pure species i is most commonly given by Eq. (6.90), the Antoine equation:
B i
= Ai − _____
ln Pisat
(13.15)
T + C i
454 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
Figure 13.1 shows a vessel in which a vapor mixture and a liquid solution coexist in vapor/liquid
equilibrium. If the vapor phase is assumed to be in its ideal-gas state and the liquid phase is
assumed to be an ideal solution, both Φi and γi in Eq. (13.13) approach unity, and this equation
reduces to its simplest possible form, Raoult’s law:1
y i P = x i Pisat
(i = 1, 2, . . . , N )
(13.16)
isat
where xi is a liquid-phase mole fraction, yi is a vapor-phase mole fraction, and P is the vapor
pressure of pure species i at the temperature of the system. The product yiP is the partial pres-
sure of species i in the vapor phase. Note that the only thermodynamic function to survive here
is the vapor pressure of pure-species i, suggesting its primary importance in VLE calculations.
Vapor
Figure 13.1: Schematic representation of VLE. The temperature T and T, P, yi
pressure P are uniform throughout the vessel and can be measured
with appropriate instruments. Vapor and liquid samples may be
withdrawn for analysis, and this provides experimental values for
mole fractions in the vapor {yi} and mole fractions in the liquid {xi}. Liquid
T, P, xi
The ideal-gas-state assumption means that Raoult’s law is limited in application to low to
moderate pressures. The ideal-solution assumption implies that Raoult’s law can have approx-
imate validity only when the species that comprise the system are chemically similar. Just as
the ideal-gas state serves as a standard to which real-gas behavior may be compared, the ideal
solution represents a standard to which real-solution behavior may be compared. Liquid-phase
ideal-solution behavior is promoted when the molecular species are not too different in size
and have the same chemical nature. Thus, a mixture of isomers, such as ortho-, meta-, and
para-xylene, conforms very closely to ideal-solution behavior. So do mixtures of adjacent
members of a homologous series, as for example, n-hexane/n-heptane, ethanol/propanol,
and benzene/toluene. Other examples are acetone/acetonitrile and acetonitrile/nitromethane.
Figures 12.11 and 12.12 for the latter system are constructed to represent Raoult’s law.
The simple model of VLE represented by Eq. (13.16) provides a realistic description of
actual behavior for a relatively small class of systems. Nevertheless, it serves as a standard of
comparison for more complex systems. A limitation of Raoult’s law is that it can be applied
only to species of known vapor pressure, and this requires each species to be “subcritical,” i.e.,
at a temperature below its critical temperature. Raoult’s law cannot be applied to situations for
which the temperature exceeds the critical temperature of one or more species in the mixture.
This equation finds direct application in bubblepoint calculations, where the vapor-phase
composition is unknown. For a binary system with x2 = 1 − x1,
P = P2sat
+ (P1sat
− P2sat
) x 1(13.18)
and a plot of P vs. x1 at constant temperature is a straight line connecting P2sat at x1 = 0 with
P1sat
at x1 = 1. The Pxy diagram of Fig. 12.11 for acetonitrile(l)/nitromethane(2) shows this
linear relation.
For this system at a temperature of 75°C, the pure-species vapor pressures are
P1sat = 83.21 kPa and P2sat
= 41.98 kPa. BUBL P calculations are readily carried out by
456 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
substitution of these values in Eq. (13.18), along with values of x1. The results allow calcula-
tion of the P-x1 relation. The corresponding values of y1 are found from Eq. (13.16):
x 1P1sat
y 1 = _____
P
The following table shows the results of calculation. These are the values used to construct the
P-x1-y1 diagram of Fig. 12.11:
x1 y1 P/kPa x1 y1 P/kPa
0.0 0.0000 41.98 0.6 0.7483 66.72
0.2 0.3313 50.23 0.8 0.8880 74.96
0.4 0.5692 58.47 1.0 1.0000 83.21
When P is fixed, the temperature varies along with x1 and y1, and the temperature range
1sat
is bounded by saturation temperatures t and t2sat
, the temperatures at which the pure species
exert vapor pressures equal to P. These temperatures can be calculated from Antoine equation:
B i
= _______
tisat
− C i
Ai − ln P
y i P = x i γ i Pisat
(i = 1, 2, . . . , N )
(13.19)
This equation provides for entirely satisfactory representation of the VLE behavior of a great
variety of systems at low to moderate pressures.
Because ∑ i y i = 1, Eq. (13.19) may be summed over all species to yield:
P = ∑ x i γ i Pisat
(13.20)
i
Alternatively, Eq. (13.19) may be solved for xi, in which case summing over all species yields:
1
P = ____________
(13.21)
∑ y i∕γ i Pisat
i
Bubblepoint and dewpoint calculations made with the modified Raoult’s law are only a bit
more complex than the same calculations made with Raoult’s law. In particular, bubblepoint
pressure calculations are straightforward because the specified liquid composition allows
immediate evaluation of the activity coefficients. Dewpoint pressure calculations require an
iterative solution process because the unknown liquid-phase composition is required to eval-
uate the activity coefficients. Bubblepoint and dewpoint temperature calculations are further
complicated by the temperature dependence of the activity coefficients, along with the temper-
ature dependence of the vapor pressures, but the same iterative or trial-and-error approaches
used with Raoult’s law calculations can still be employed.
Example 13.1
For the system methanol(1)/methyl acetate(2), the following equations provide a rea-
sonable correlation for the activity coefficients:
ln γ 1 = Ax22 ln γ 2 = Ax12 where A = 2.771 − 0.00523T
In addition, the following Antoine equations provide vapor pressures:
3643.31 2665.54
= 16.59158 − _________
ln P1sat = 14.25326 − _________
ln P2sat
T − 33.424 T − 53.424
where T is in kelvins and the vapor pressures are in kPa. Assuming the validity of
Eq. (13.19), calculate:
(a) P and { yi} for T = 318.15 K and x1 = 0.25.
(b) P and {xi} for T = 318.15 K and y1 = 0.60.
(c) T and { yi} for P = 101.33 kPa and x1 = 0.85.
(d) T and {xi} for P = 101.33 kPa and y1 = 0.40.
(e) The azeotropic pressure and the azeotropic composition for T = 318.15 K.
Solution 13.1
In the dewpoint and bubblepoint calculations of parts (a) through (d), the key is the
dependence of the activity coefficients on T and x1. In (a), both values are given,
458 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
and solution is direct. In (b) only T is given, and solution is by trial with x1 varied to
reproduce the given value of y1. In (c) only x1 is given, and T is varied to reproduce
the given value of P. In (d) neither T nor x1 is given, and both are varied alternately,
T to yield P and x1 to yield y1. In parts (b) through (d), the trial and error calculations
are readily automated using Microsoft Excel’s Goal Seek function.
(a) A BUBL P calculation. For T = 318.15 K, the Antoine equations yield
P1sat 2sat
= 44.51 and P = 65.64 kPa.The activity-coefficient correlation provides
A = 1.107, γ1 = 1.864, and γ2 = 1.072. By Eq. (13.20), P = 73.50 kPa and by
Eq. (13.19), y1 = 0.282.
(b) A DEW P calculation. With T unchanged from part (a), P 1sat
, P2sat
, and A are
also unchanged. The unknown liquid-phase composition is varied in trial cal-
culations that evaluate the activity coefficients, P by Eq. (13.21), and y1 by
Eq. (13.19), with the goal of reproducing the given value y1 = 0.6. This leads
to final values:
P = 62.59 kPa x 1 = 0.8169 γ 1 = 1.0378 γ 2 = 2.0935
(c) A BUBL T calculation. Solution by trial here varies T until the given value of
P is reproduced. A reasonable starting value for T is found from the saturation
temperatures of the pure species at the known pressure. The Antoine equation,
solved for T, becomes:
B i
Tisat = _______
− C i
Ai − ln P
1sat
Application for P = 101.33 kPa leads to: T 2sat
= 337.71 and T = 330.08 K.
An average of these values serves as an initial T: Each trial value of T leads
immediately to values for the activity coefficients and to a value for P by
Eq. (13.20). The known value of P = 101.33 kPa is reproduced when:
T = 326.70 K P1sat
= 64.63 kPa P2sat
= 89.94 kPa
A = 1.0624 γ 1 = 1.3628 γ 2 = 1.2523
x 1 = 0.4602 x 2 = 0.5398
13.3. Simplifications: Raoult’s Law, Modified Raoult’s Law, and Henry’s Law 459
(e) First determine whether or not an azeotrope exists at the given temperature. This
calculation is facilitated by the definition of a quantity called the relative volatility:
y 1 ∕ x 1
α 12 ≡ _
(13.22)
y 2 ∕ x 2
P az = γ1az P1sat
= (1.657) (44.51)
The criterion for vapor/liquid equilibrium is that the fugacity of species i is the same
in both phases. If the vapor phase is in its ideal-gas state, then the fugacity equals the partial
pressure, and
The liquid-phase fugacity of species i increases from zero at infinite dilution (xi = yi → 0) to
Pisat
for pure species i. This is illustrated by the data of Table 13.1 for the methyl ethyl ketone(l)/
toluene(2) system at 50°C.3 The first three columns list experimental P-x1-y1 data, and columns
4 and 5 show fˆ 1 l = y 1 P and fˆ 2 l = y 2 P. The fugacities are plotted in Fig. 13.2 as solid lines. The
straight dashed lines represent Eq. (10.83), the Lewis/Randall rule, which expresses the compo-
sition dependence of the constituent fugacities in an ideal solution:
Although derived from a particular set of data, Fig. 13.2 illustrates the general nature of
the fˆ 1 l and fˆ 2 l vs. x1 relationships for a binary liquid solution at constant T. The equilibrium pres-
sure P varies with composition, but its influence on the liquid-phase values of fˆ 1 l and fˆ 2 l is
negligible. Thus a plot at constant T and P would look the same, as indicated in Fig. 13.3 for
species i (i = 1, 2) in a binary solution at constant T and P.
3M. Diaz Peña, A. Crespo Colin, and A. Compostizo, J. Chem. Thermodyn., vol. 10, pp. 337–341, 1978.
13.3. Simplifications: Raoult’s Law, Modified Raoult’s Law, and Henry’s Law 461
f1 = P1sat Hi
Constant T, P
30
^
f1 = y1 P
aw
20 ^
fi
fi /kPa
sl
y’
fi
nr
He
^
f2 = P 2sat
^
fi
e
10 rul
d all
^ an
f2 = y2 P s/R
wi
Le
0 1 0 1
xi
x1
The lower dashed line in Fig. 13.3, representing the Lewis/Randall rule, is c haracteristic
of ideal-solution behavior. It provides the simplest possible model for the composition
dependence of f ˆ i l , representing a standard to which actual behavior may be compared. Indeed,
the activity coefficient as defined by Eq. (13.2) formalizes this comparison:
Thus the activity coefficient of a species in solution is the ratio of its actual fugacity to the value
given by the Lewis/Randall rule at the same T, P, and composition. For the calculation of experi-
mental values of γi, both fˆ i l and f ˆ i id are eliminated in favor of measurable quantities.
y i P y i P
γ i = ____l = _____
(i = 1, 2, . . . , N ) (13.24)
x i f i x i Pisat
This is a restatement of Eq. (13.19), the modified Raoult’s law, and is adequate for present
purposes, allowing easy calculation of activity coefficients from experimental low-pressure
VLE data. Values from this equation appear in the last two columns of Table 13.1.
Figure 13.4 shows plots of ln γi based on experimental measurements for six binary
systems at 50°C, illustrating the variety of behavior that is observed. Note in every case that
as xi → 1, ln γi → 0 with zero slope. Usually (but not always) the infinite-dilution a ctivity
462 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
3 1.5 1.5
2 In γ1 In γ1 1.0 In γ1
1.0
In γ2 In γ2
In γ2
1 0.5 0.5
0 1 0 1 0 1
x1 x1 x1
(d ) (e) (f)
Figure 13.4: Logarithms of the activity coefficients at 50°C for six binary liquid systems:
(a) chloroform(l)/n-heptane(2); (b) acetone(l)/methanol(2); (c) acetone(l)/chloroform(2);
(d) ethanol(l)/n-heptane(2); (e) ethanol(l)/chloroform(2); ( f ) ethanol(l)/water(2).
c oefficient is an extreme value. Comparison of these graphs with those of Fig. 10.3 indicates
that the ln γi generally have the same sign as GE. That is, positive GE implies activity
coefficients greater than unity and negative GE implies activity coefficients less than unity, at
least over most of the composition range.
Henry’s Law
The solid lines in both Figs. 13.2 and 13.3, representing experimental values of fˆ i l , become
tangent to the Lewis/Randall-rule lines at xi = 1. This is a consequence of the Gibbs/Duhem
equation, as will be shown presently. In the other limit, xi → 0, fˆ i l becomes zero. Thus, the
ratio f ˆ i l ∕ x iis indeterminate in this limit, and application of l’Hôpital’s rule yields:
i = ( ___i )
fˆ l d fˆ l
lim __
≡ i (13.25)
x→0 x i d x i xi = 0
13.3. Simplifications: Raoult’s Law, Modified Raoult’s Law, and Henry’s Law 463
Equation (13.25) defines Henry’s constant i as the limiting slope of the fˆ i l -vs.-xi curve at
xi = 0. As shown by Fig. 13.3, this is the slope of a line drawn tangent to the curve at xi = 0.
The equation of this tangent line expresses Henry’s law.
Applicable in the limit as xi → 0, it is also of approximate validity for small values of xi.
For a system of air in equilibrium with liquid water, the mole fraction of water vapor in
the air is found from Raoult’s law applied to the water with the assumption that the water is
essentially pure. Thus, Raoult’s law for the water (species 2) becomes y 2P = P2sat
. At 25°C and
atmospheric pressure, this equation yields:
P2sat
______3.166
y 2 = ____
=
= 0.0312
P 101.33
2sat
where the pressures are in kPa, and P comes from the steam tables.
Calculation of the mole fraction of air dissolved in the water is accomplished with H
enry’s
law, applied here for a pressure low enough that the vapor phase is in its ideal-gas state. Values
of i come from experiment, and Table 13.2 lists values at 25°C for a few gases dissolved in
water. For the air/water system at 25°C and atmospheric pressure, Henry’s Law applied to air
(species 1) with y1 = 1 − 0.0312 = 0.9688 yields:
y 1 P ______________
( 0.9688)( 1.0133)
x 1 = ____
= 10 −5
= 1.35 ×
1 72,950
This result justifies the assumption that the water approaches purity.
Gas /bar
Acetylene 1,350
Air 72,950
Carbon dioxide 1,670
Carbon monoxide 54,600
Ethane 30,600
Ethylene 11,550
Helium 126,600
Hydrogen 71,600
Hydrogen sulfide 550
Methane 41,850
Nitrogen 87,650
Oxygen 44,380
464 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
Example 13.2
Assuming that carbonated water contains only CO2(1) and H2O(2), determine
the compositions of the vapor and liquid phases in a sealed can of “soda” at 25°C
if the pressure inside the can is 5 bar.
Solution 13.2
Expecting that the liquid phase will be nearly pure water and the vapor phase will
be nearly pure CO2, we apply Henry’s law for CO2 (species 1) and Raoult’s law
for water (species 2):
y 1 P P 5
x 1 = ____
≈ ____
= _____
= 0.0030
1 1 1670
x 2 P2sat
____Psat
y 2 = ______
≈ 2
P P
From the steam tables, the vapor pressure of water at 25°C is 3.166 kPa, or
0.0317 bar. Thus, y2 = 0.0317/5 = 0.0063. Consistent with our expectations, the
liquid is 99.7% water and the vapor is 99.4% CO2.
Henry’s law is related to the Lewis/Randall rule through the Gibbs/Duhem equation, expressed
¯ il = μil , it becomes:
¯ i replaced by G
by Eq. (10.14). Written for a binary liquid solution with M
Differentiation of Eq. (10.46) at constant T and P yields: d μil = RTd ln fˆ i l . The preceding equa-
tion is then
x 1 d ln fˆ 1 l + x 2 d ln fˆ 2 l = 0 (const T, P)
Upon division by dx1.
d ln fˆ 1 l d ln fˆ 2 l
x 1 ______
+ x 2 ______
= 0 (const T, P) (13.27)
d x 1 d x 1
This is a particular form of the Gibbs/Duhem equation. Substitution of −dx2 for dx1 in the
second term produces:
13.3. Simplifications: Raoult’s Law, Modified Raoult’s Law, and Henry’s Law 465
According to Eq. (13.25), the numerator and denominator on the right side of this equation are
equal, and therefore:
This equation is the exact expression of the Lewis/Randall rule as applied to real solutions.
It also implies that Eq. (10.83) provides approximately correct values of fˆ i l when xi ≈ 1:
fˆ i l ≈ fˆ i id = x i f i l.
The fugacity shown by Fig. 13.3 is for a species with positive deviations from ideality
in the sense of the Lewis/Randall rule. Negative deviations are less common but are also
observed; the fˆ i l -vs.-xi curve then lies below the Lewis/Randall line. In Fig. 13.5 the fugacity
of acetone is shown as a function of composition for two different binary liquid solutions at
50°C. When the second species is methanol, acetone exhibits positive deviations from ideal-
ity. When the second species is chloroform, the deviations are negative. The fugacity of pure
acetone facetone is of course the same regardless of the identity of the second species. However,
Henry’s constants, represented by slopes of the two dotted lines, are very different for the
two cases.
Example 13.3
A fog consists of spherical water droplets with a radius of about 10−6 m. Because of
surface tension, the internal pressure within a water droplet is greater than the exter-
nal pressure. For droplet radius r and surface tension σ, the internal pressure in a water
466 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
0.6 facetone
In methanol
0.4
f^acetone /bar
0 1
xacetone
droplet is greater than the external pressure by δP = 2σ∕r. This increases the fugacity
of the water in a droplet above that of bulk water at the same conditions. Thus, it tends
to cause the fog to evaporate. However, the fugacity increase may be countered by a
decrease in temperature or by the dissolution of atmospheric pollutants. To stabilize
the fog with respect to bulk water at the same temperature, determine:
(a) The minimum temperature decrease required.
(b) The minimum concentration of atmospheric pollutants required in the fog
droplets.
Solution 13.3
(a) At 25°C the surface tension of pure water is 0.0694 N·m−1, and
( 2)( 0.0694)
δP = __________ 10 6 Pa or 1.388 bar
= 0.1338 ×
10 −6
Two general equations apply to the isothermal change in Gibbs energy resulting
from a change in pressure. The first results from differentiation of Eq. (10.31),
the defining equation for fugacity, and the second from restriction of Eq. (6.11)
to constant temperature:
V i
d ln fi = ___
dP ( const T )
RT
V H O
δ ln f H 2O = _____
2 δP (const T )
RT
For a molar volume of water of 18 cm3·mol−1, and a pressure change of 1.388 bar,
( 18 cm 3·mol)( 1.388 bar)
______________________________
δ ln f H 2O =
= 0.00101
(83.14 cm 3·bar·mol −1·K −1) ( 298 K)
This is the amount by which the fugacity of water in a fog droplet exceeds the
fugacity of bulk water at the same temperature. Thus the fog is not stable, and
it dissipates by evaporation. However, a sufficient decrease in temperature of
the fog reduces droplet fugacity enough to stabilize the fog.
An equation giving the effect of temperature on fugacity is developed from
Eq. (10.31) by subtracting from it the ideal-gas-state form of the same equa-
tion (for which fi = P). Thus
ig
G i − Gi = RT ln fi − RT ln P
Substituting for the partial derivatives on the right by Eq. (6.39) gives:
The residual enthalpy here is approximately the negative of the latent heat of
vaporization of water at 25°C; thus
− HiR = 2442.3 J·g −1 × 18 g·mol −1 = 43,960 J⋅mol −1. Then
( ∂ T ) P (8.314 J·mol −1·K −1) (298 K)
∂ ln f H 2O 43,960 J·mol −1
_______
= _________________________
2 = 0.0595 K −1
468 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
This is the minimum temperature decrease of the fog required for stability
with respect to bulk water.
(b) The fugacity of water in fog droplets is also lowered by the dissolution of
impurities, such as atmospheric pollutants. The droplets then become solutions
with water as the major constituent, and their fugacity is fˆ H 2O. Because the
mole fraction of water is near unity, the Lewis/Randall rule provides an excel-
lent approximation, fˆ H 2O = f H 2O x H 2O.The fugacity change of the water result-
ing from the dissolution of impurities is then:
δ f H 2O = fˆ H 2O − f H 2O = f H 2O x H 2O − f H 2O = f H 2O(x H 2O − 1) = − f H 2O x impurity
or
− δ f H 2O
x impurity = _______
≈ − δ ln f H 2O
f H 2O
This quantity is again the negative of the increase caused by surface tension.
Thus,
x impurity = 0.00101
and impurities of only 0.1% stabilize the fog with respect to bulk water.
Liquid-phase activity coefficients γi play a vital role in the gamma/phi formulation of
VLE. They are directly related to the excess Gibbs energy. In general, GE∕RT is a function
of T, P, and composition, but for liquids at low to moderate pressures it is a very weak
function of P. Therefore its pressure dependence is usually neglected, and for applications at
constant T, excess Gibbs energy is treated as a function of composition alone:
G E
___
= g(x 1, x 2, . . ., x n)
(const T )
RT
G E
Y ≡ _______
= a + bx 1 + cx12 + · · ·
(const T )
x 1 x 2 RT
Because x2 = 1 − x1, mole fraction x1 serves as the single independent variable. An equivalent
power series with certain advantages is known as the Redlich/Kister expansion:4
a
Y = A 0 + ∑
An z n (13.29)
n = 1
where, by definition, z ≡ x1 − x2 = 2x1 − 1, a is the order of the power series, and parameters
An are functions of temperature.
Expressions for the activity coefficients are found from Eqs. (10.15) and (10.16) with
GE∕RT replacing ME.
d(G E ∕ RT ) dY
_________
= x 1 x 2 ____
+ Y(x 2 − x 1)
d x 1 d x 1
( d x 1) ( d x 1)
dY dY
ln γ 1 = x22 Y + x 1 _
(13.32) ln γ 2 = x12 Y − x 2 _
(13.33)
dY
____
a
= ∑
n A z n−1
(13.34)
d x 1 n = 1 n
In application, different truncations of these series are appropriate, and truncations with a ≤ 5
are frequent in the literature.
4O. Redlich, A. T. Kister, and C. E. Turnquist, Chem. Eng. Progr. Symp. Ser. No. 2, vol. 48, pp. 49–61, 1952.
470 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
When all parameters are zero, ln γ1 = 0, ln γ2 = 0, γ1 = γ2 = 1. These are the values for
an ideal solution, and they represent a limiting case where the excess Gibbs energy is zero.
If all parameters except A0 are zero, Y = A0, and Eqs. (13.32) and (13.33) reduce to:
The symmetrical nature of these relations is evident. Infinite-dilution values of the activity
1∞ = ln γ2∞ = A 0.
coefficients are ln γ
Most widely found is the two-parameter truncation:
Y = A 0 + A 1(x 1 − x 2) = A 0 + A 1(2 x 1 − 1)
in which Y is linear in x1. An alternate form of this equation results from the definitions A0 + A1 =
A21 and A0 − A1 = A12. Eliminating parameters A0 and A1 in favor of A21 and A12, we obtain:
G E
___
= (A 21 x 1 + A 12 x 2) x 1 x 2 (13.39)
RT
ln γ 1 = x22 [ A 12 + 2(A 21 − A 12) x 1] (13.40)
These are known as the Margules5 equations. For the limiting conditions of infinite
dilution, they imply:
( ′ x 2) ( ′ x 1)
′ x 1 −2
A 12 ′ x 2 −2
A 21
′ 1 + ________
ln γ 1 = A 12
′ 1 + ________
(13.43) ln γ 2 = A 21
(13.44)
A 21 A 12
Local-Composition Models
Theoretical developments in the molecular thermodynamics of liquid-solution behavior are often
based on the concept of local composition. Within a liquid solution, local compositions, different
from the overall mixture composition, are presumed to account for the short-range order and
nonrandom molecular orientations that result from differences in molecular size and intermo-
lecular forces. The concept was introduced by G. M. Wilson in 1964 with the publication of a
model of solution behavior since known as the Wilson equation.8 The success of this equation
in the correlation of VLE data prompted the development of alternative local-composition mod-
els, most notably the NRTL (Non-Random-Two-Liquid) equation of Renon and Prausnitz9 and
the UNIQUAC (UNIversal QUAsi-Chemical) equation of Abrams and Prausnitz.10 A further
significant development, based on the UNIQUAC equation, is the UNIFAC method,11 in which
activity coefficients are calculated from contributions of the various groups making up the mol-
ecules of a solution.
Wilson Equation. Like the Margules and van Laar equations, the Wilson equation con-
tains just two parameters for a binary system (Λ12 and Λ21). It is written:
G E
___
= − x 1ln ( x 1 + x 2 Λ 12) − x 2ln ( x 2 + x 1 Λ 21)
(13.45)
RT
6Johannes Jacobus van Laar (1860–1938), Dutch physical chemist; see http://en.wikipedia.org/wiki/
Johannes_van_Laar.
7H. C. Van Ness and M. M. Abbott, Classical Thermodynamics of Nonelectrolyte Solutions: With Applications to
nitz, AIChE J., vol. 21, p. 1086–1099, 1975; given detailed treatment in the monograph Aa. Fredenslund, J. Gmehling,
and P. Rasmussen, Vapor-Liquid Equilibrium Using UNIFAC, Elsevier, Amsterdam, 1977.
472 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
( x 1 + x 2 Λ 12 x 2 + x 1 Λ 21)
Λ 12 Λ 21
ln γ 1 = − ln ( x 1 + x 2 Λ 12) + x 2 _
− _
(13.46)
( x 1 + x 2 Λ 12 x 2 + x 1 Λ 21)
Λ 12 Λ 21
ln γ 2 = − ln (x 2 + x 1 Λ 21) − x 1 _
− _
(13.47)
G 12 = exp (− α τ 12) G 21 = exp (− α τ 21)
Here,
b1 2 b2 1
τ 12 = ___
and τ 21 = ___
RT RT
where α, b12, and b21, parameters specific to a particular pair of species, are independent of
composition and temperature. The infinite-dilution values of the activity coefficients are given
by the equations:
The UNIQUAC equation and the UNIFAC method are models of greater complexity
and are treated in App. G.
Multicomponent Systems
The local-composition models have limited flexibility in the fitting of data, but they are adequate
for most engineering purposes. Moreover, they are implicitly generalizable to multicomponent
systems without the introduction of any parameters beyond those required to describe the
constituent binary systems. For example, the Wilson equation for multicomponent systems is:
x j Λ ij
(j )
G E
___
= − ∑ x
i ln ∑ (13.51)
RT i
where Λij = 1 for i = j, etc. All indices refer to the same species, and summations are over all
species. For each ij pair there are two parameters, because Λij ≠ Λji. For a ternary system the
three ij pairs are associated with the parameters Λ12, Λ21; Λ13, Λ31; and Λ23, Λ32,
The temperature dependence of the parameters is given by:
V j − a ij
Λ ij = ___
exp ____
(i ≠ j) (13.53)
V i RT
where Vj and Vi are the molar volumes at temperature T of pure liquids j and i, and aij is a
constant independent of composition and temperature. Thus the Wilson equation, like all other
local-composition models, has built into it an approximate temperature dependence for the
parameters. Moreover, all parameters are found from data for binary (in contrast to multicom-
ponent) systems. This makes parameter determination for the local-composition models a task
of manageable proportions.
In Table 13.3 the first three columns repeat the P-x1-y1 data of Table 13.1 for the system
methyl ethyl ketone(l)/toluene(2). These data points are also shown as circles on Fig. 13.6(a).
Values of ln γ1 and ln γ2 are listed in columns 4 and 5, and are shown by the open squares and
triangles of Fig. 13.6(b). They are combined for a binary system in accord with Eq. (13.10):
_G E
= x 1 ln γ 1 + x 2 ln γ 2 (13.54)
RT
The values of GE∕RT so calculated are then divided by x1x2 to provide values of GE∕(x1x2 RT );
the two sets of numbers are listed in columns 6 and 7 of Table 13.3 and appear as solid circles
on Fig. 13.6(b).
The four thermodynamic functions ln γ1, ln γ2, GE∕RT, and GE∕(x1x2RT ), are properties
of the liquid phase. Figure 13.6(b) shows how their experimental values vary with composi-
tion for a particular binary system at a specified temperature. This figure is characteristic of
systems for which:
γ i ≥ 1 and ln γ i ≥ 0 (i = 1, 2)
In such cases the liquid phase shows positive deviations from Raoult’s-law behavior. This
is seen also in Fig. 13.6(a), where the P-x1 data points all lie above the dashed straight line,
which represents Raoult’s law.
Because the activity coefficient of a species in solution becomes unity as the species
becomes pure, each ln γi (i = 1, 2) tends to zero as xi → 1. This is evident in Fig. 13.6(b). At
the other limit, where xi → 0 and species i becomes infinitely dilute, ln γi approaches a finite
limit, namely, ln γi∞
. In the limit as x1 → 0, the dimensionless excess Gibbs energy GE∕RT as
given by Eq. (13.54) becomes:
G E
lim ___
= (0) ln γ1∞ + (1)(0) = 0
x 1→0 RT
474 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
35
0.4
ln γ∞
1
30
GE/x1 x2RT
P-x1 0.3
25
P/kPa
P-y1 ln γ1
20 0.2
ln γ2∞
ln γ2
15 0.1
GE/RT
10
(a) ( b)
Figure 13.6: The methyl ethyl ketone(1)/toluene(2) system at 50°C. (a) Pxy data and their correlation.
(b) Liquid-phase properties and their correlation.
The same result is obtained for x2 → 0 (x1 → 1). The value of GE∕RT (and GE) is there-
fore zero at both x1 = 0 and x1 = 1.
The quantity GE∕(x1x2RT ) becomes indeterminate both at x1 = 0 and x1 = 1, because GE
is zero in both limits, as is the product x1x2. For x1 → 0, l’Hôpital’s rule yields:
G E G E ∕ RT d(G E ∕ RT )
lim _______
= lim ______
= lim _________
(A)
x 1→0 x 1 x 2 RT x 1→0 x 1 x 1→0 d x 1
13.5. Fitting Activity Coefficient Models to VLE Data 475
Differentiation of Eq. (13.54) with respect to x1 provides the derivative of the final member:
d(G E ∕ RT ) d ln γ 1 d ln γ 2
_________
= x 1 ______
+ ln γ 1 + x 2 ______
− ln γ 2 (B)
d x 1 d x 1 d x 1
The minus sign preceding the last term comes from dx2∕dx1 = −1, a consequence of the
equation, x1 + x2 = 1. The Gibbs/Duhem equation, Eq. (13.11), written for a binary system, is
divided by dx1 to give:
d ln γ 1 d ln γ 2
x 1 _
+ x 2 _
= 0 (const T, P) (13.55)
d x 1 d x 1
d(G E ∕ RT ) γ 1
lim _________
= lim ln __
= ln γ1∞
x 1→0 d x
1 x 1→0 γ 2
G E G E
By Eq. (A), lim _______
= ln γ1∞ Similarly, lim _______
= ln γ2∞
x 1→0
x
1
x
2
RT x 1→1
x
1
x
2
RT
Thus the limiting values of GE∕(x1x2RT ) are equal to the infinite-dilution limits of ln γ1 and ln
γ2. This result is illustrated in Fig. 13.6(b).
These results depend on Eq. (13.11), which is valid for constant T and P. Although the
data of Table 13.3 are for constant T, but variable P, negligible error is introduced through
Eq. (13.11) because liquid-phase activity coefficients are very nearly independent of P for
systems at low to moderate pressures.
Equation (13.11) has further influence on the nature of Fig. 13.6(b). Rewritten as,
d ln γ 1
______ x 2d ln γ 2
= − ___
______
d x 1 x 1 d x 1
it requires the slope of the ln γ1 curve to be everywhere of opposite sign to the slope of the
ln γ2 curve. Furthermore, when x2 → 0 (and x1 → 1), the slope of the ln γ1 curve is zero.
Similarly, when x1 → 0, the slope of the ln γ2 curve is zero. Thus, each ln γi (i = 1, 2) curve
terminates at zero with zero slope at xi = 1.
Data Reduction
Of the sets of points shown in Fig. 13.6(b), those for GE∕(x1x2RT ) most closely conform to
a simple mathematical relation. Thus a straight line provides a reasonable approximation to
this set of points, and mathematical expression is given to this linear relation by the equation:
_______ G E
= A 21 x 1 + A 12 x 2
x 1 x 2 RT
476 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
where A21 and A12 are constants in any particular application. This is the Margules equation
as given by Eq. (13.39), with corresponding expressions for the activity coefficients given by
Eqs. (13.40) and (13.41).
For the methyl ethyl ketone/toluene system considered here, the curves of Fig. 13.6(b)
for GE∕RT, ln γ1, and ln γ2 represent Eqs. (13.39), (13.40), and (13.41) with:
A 12 = 0.372 and A 21 = 0.198
These are values of the intercepts at x1 = 0 and x1 = 1 of the straight line drawn to represent
the GE∕(x1x2RT ) data points.
A set of VLE data has here been reduced to a simple mathematical equation for the
dimensionless excess Gibbs energy:
G E
___
= (0.198 x 1 + 0.372 x 2) x 1 x 2
RT
This equation concisely stores the information of the data set. Indeed, the Margules e quations
for ln γ1 and ln γ2 allow construction of a correlation of the original P-x1-y1 data set.
Equation (13.19) is written for species 1 and 2 of a binary system as:
y 1 P = x 1 γ 1 P1sat
and y 2 P = x 2 γ 2 P2sat
x 1 γ 1 P1sat
Whence, y 1 = _________________
(13.58)
x 1 γ 1 P1 + x 2 γ 2 P2sat
sat
Values of γ1 and γ2 from Eqs. (13.40) and (13.41) with A12 and A21 as determined for
the methyl ethyl ketone(l)/toluene(2) system are combined with the experimental values
of P1sat
and P2sat
to calculate P and y1 by Eqs. (13.57) and (13.58) at various values of x1.
The results are shown by the solid lines of Fig. 13.6(a), which represent the calculated
P-x1 and P-y1 relations. They clearly provide an adequate correlation of the experimental
data points.
A second set of P-x1-y1 data, for chloroform(1)/1,4-dioxane(2) at 50°C,12 is given in
Table 13.4, along with values of pertinent thermodynamic functions. Figures 13.7(a) and
13.7(b) display all of the experimental values as points. This system shows negative deviations
from Raoult’s-law behavior; γ1 and γ2 are less than unity, and values of ln γ1, ln γ2, GE∕RT, and
GE∕(x1x2RT ) are negative. Moreover, the P-x1 data points in Fig. 13.7(a) all lie below the
dashed line representing Raoult’s-law behavior. Again the data points for GE∕(x1x2RT ) are
reasonably well correlated by the Margules equation, in this case with parameters:
A 12 = − 0.72 and A 21 = − 1.27
Values of GE∕RT, ln γ1, ln γ2, P, and y1 calculated by Eqs. (13.39), (13.40), (13.41), (13.57),
and (13.58) provide the curves shown for these quantities in Figs. 13.7(a) and 13.7(b). Again,
the experimental P-x1-y1 data are adequately correlated. Although the correlations provided by
12M. L. McGlashan and R. P. Rastogi, Trans. Faraday Soc, vol. 54, p. 496, 1958.
13.5. Fitting Activity Coefficient Models to VLE Data 477
x1
70 0.2 0.4 0.6 0.8 1. 0
0
60
0.2
G E/RT
50
0.4
ln γ1
40 P-x1 ln γ2
ln γ1∞
P/kPa
0.6
30
P-y1 0.8
G E/x1 x2RT
20
1. 0
10
1. 2
ln 2
0 0.2 0.4 0.6 0.8 1. 0
x1 , y1
(a) (b)
Figure 13.7: The chloroform(1)/1,4-dioxane(2) system at 50°C. (a) Pxy data and their correlation.
(b) Liquid-phase properties and their correlation.
478 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
the Margules equations for the two sets of VLE data presented here are satisfactory, they are
not perfect. The two possible reasons are, first, that the Margules equations are not precisely
suited to the data set; second, that the P-x1-y1 data themselves are systematically in error such
that they do not conform to the requirements of the Gibbs/Duhem equation.
We have presumed in applying the Margules equations that the deviations of the
experimental points for GE∕(x1x2RT ) from the straight lines drawn to represent them result
from random error in the data. Indeed, the straight lines do provide excellent correlations
of all but a few data points. Only toward edges of a diagram are there significant deviations,
and these have been discounted because the error bounds widen rapidly as the edges of
such a diagram are approached. In the limits as x1 → 0 and x1 → 1, GE∕(x1x2RT ) becomes
indeterminate; experimentally this means that the values are subject to unlimited error and
are not measurable. However, the possibility exists that the correlation would be improved
were the GE∕(x1x2RT ) points represented by an appropriate curve, rather than a straight line.
Finding the correlation that best represents the data is a trial procedure.
Thermodynamic Consistency
The Gibbs/Duhem equation, Eq. (13.55), imposes a constraint on activity coefficients
that may not be satisfied by a set of experimental values derived from P-x1-y1 data. The
experimental values of ln γ1 and ln γ2 combine by Eq. (13.54) to give values of GE∕RT.
This addition process is independent of the Gibbs/Duhem equation. On the other hand, the
Gibbs/Duhem equation is implicit in Eq. (13.7), and activity coefficients derived from this
equation necessarily obey the Gibbs/Duhem equation. These derived activity coefficients
cannot possibly be consistent with the experimental values unless the experimental
values also satisfy the Gibbs/Duhem equation. Nor can a P-x1-y1 correlation calculated by
Eqs. (13.57) and (13.58) be consistent with such experimental values. If the experimental
data are inconsistent with the Gibbs/Duhem equation, they are necessarily incorrect as the
result of systematic error in the data. Because correlating equations for GE∕RT impose
consistency on derived activity coefficients, no correlation exists that can precisely
reproduce P-x1-y1 data that are inconsistent.
Our purpose now is to develop a simple test for the consistency with respect to the
Gibbs/Duhem equation of a P-x1-y1 data set. Equation (13.54) is written with experimental
values, calculated by Eq. (13.24), and denoted by an asterisk:
( RT )
G E *
___
= x 1ln γ1* + x 2ln γ2*
Differentiation gives:
This equation is subtracted from Eq. (13.56), written for derived property values, i.e., those
given by a correlation, such as the Margules equations:
d(G E ∕ RT ) __________
γ2* ( 1 dx 1 dx 1 )
d(G E ∕ RT ) * γ 1 γ1* d ln γ1* d ln γ2*
__________ −
= ln __
− ln ___ − x ______
+ x 2 ______
dx 1 dx 1 γ 2
The differences between like terms are residuals, which may be represented by a δ notation.
The preceding equation then becomes:
d δ(G E ∕ RT )
γ 2 ( d x 1 )
γ 1 d ln γ1* d ln γ2*
___________
= δ ln
__
− x 1 ______
x 2 ______
+
dx 1 d x 1
If a data set is reduced so as to make the residuals in GE∕RT scatter about zero, then the
derivative d δ(GE∕RT )∕dx1 is effectively zero, reducing the preceding equation to:
The right side of this equation is exactly the quantity that Eq. (13.55),
the Gibbs/Duhem equation, requires to be zero for consistent data. The
residual on the left therefore provides a direct measure of deviation from
the Gibbs/Duhem equation. The extent to which a data set departs from
consistency is measured by the degree to which these residuals fail to
scatter about zero.13
Example 13.4
VLE data for diethyl ketone(1)/n-hexane(2) at 65°C as reported by Maripuri and
Ratcliff14 are given in the first three columns of Table 13.5. Reduce this set of data.
Solution 13.4
The last three columns of Table 13.5 present the experimental values, ln γ1* , ln γ 2* ,
and (GE∕(x1x2RT ))*, calculated from the data by Eqs. (13.24) and (13.54). All
values are shown as points on Figs. 13.8(a) and 13.8(b). The object here is to find
an equation for GE∕RT that provides a suitable correlation of the data. Although
13This test and other aspects of VLE data reduction are treated by H. C. Van Ness, J. Chem. Thermodyn., vol. 27,
pp. 113–134, 1995; Pure & Appl. Chem., vol. 67, pp. 859–872, 1995. See also, P. T. Eubank, B. G. Lamonte, and
J. F. Javier Alvarado, J. Chem. Eng. Data, vol. 45, pp. 1040–1048, 2000.
14V. C. Maripuri and G. A. Ratcliff, J. Appl. Chem. Biotechnol., vol. 22, pp. 899–903, 1972.
480 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
( x 1 x 2 RT )
G E *
P/kPa x1 y1 ln γ1* ln γ2* ________
90.15 (P2sat
) 0.000 0.000 0.000
91.78 0.063 0.049 0.901 0.033 1.481
88.01 0.248 0.131 0.472 0.121 1.114
81.67 0.372 0.182 0.321 0.166 0.955
78.89 0.443 0.215 0.278 0.210 0.972
76.82 0.508 0.248 0.257 0.264 1.043
73.39 0.561 0.268 0.190 0.306 0.977
66.45 0.640 0.316 0.123 0.337 0.869
62.95 0.702 0.368 0.129 0.393 0.993
57.70 0.763 0.412 0.072 0.462 0.909
50.16 0.834 0.490 0.016 0.536 0.740
45.70 0.874 0.570 0.027 0.548 0.844
29.00 (P1sat ) 1.000 1.000 0.000
the data points of Fig. 13.8(b) for (GE∕(x1x2RT ))* show scatter, they are adequate
to define a straight line:
_______ G E
= 0.70x 1 + 1.35x 2
x 1 x 2 RT
This is the Margules equation with A21 = 0.70 and A12 = 1.35. Derived values of
ln γ1 and ln γ2 are calculated by Eqs. (13.40) and Eqs. (13.41), and derived values
of P and y1 all come from Eqs. (13.57) and (13.58). These results, plotted as solid
lines of Figs. 13.8(a) and 13.8(b), clearly do not represent a good correlation of
the data.
The difficulty is that the data are not consistent with the Gibbs/Duhem equa-
tion. That is, the sets of experimental values, ln γ1* and ln γ2* ,shown in Table 13.5
are not in accord with Eq. (13.55). However, the values of ln γ1 and ln γ2 derived from
the correlation necessarily obey this equation; the experimental and derived val-
ues therefore cannot possibly agree, and the resulting correlation cannot provide a
precise representation of the complete set of P-x1-y1 data.
Application of the test for consistency represented by Eq. (13.59) requires cal-
culation of the residuals δ(GE∕RT ) and δ ln(γ1∕γ2), values of which are plotted vs.
x1 in Fig. 13.9. The residuals δ(GE∕RT ) distribute themselves about zero,15 as is
required by the test, but the residuals δ ln(γ1∕γ2), which show the extent to which
the data fail to satisfy the Gibbs/Duhem equation, clearly do not. Average absolute
values of this residual less than 0.03 indicate data of a high degree of consistency;
average absolute values of less than 0.10 are probably acceptable. The data set
15The simple procedure used here to find a correlation for GE/RT would be slightly improved by a regression
procedure that determines the values of A21 and A12 that minimize the sum of squares of the residuals δ(GE/RT ) .
13.5. Fitting Activity Coefficient Models to VLE Data 481
1.4
90
GE/x1 x2RT
1.2
80
P-x1
1.0
70
P-y1
0.8 In γ1
P/kPa
60
0.6 In γ2
50
0.4
40
0.2
30
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
x1 , y1 x1
(a) (b)
Figure 13.8: The diethyl ketone(l)/n-hexane(2) system at 65°C. (a) Pxy data and their correlations.
(b) Liquid-phase properties and their correlation.
considered here shows an average absolute deviation of about 0.15 and must there-
fore contain significant error. Although one cannot be certain where the error lies,
the values of y1 are usually most suspect.
The method just described produces a correlation that is unnecessarily diver-
gent from the experimental values. An alternative is to process just the P-x1 data;
this is possible because the P-x1-y1 data set includes more information than neces-
sary. The procedure requires a computer but in principle is simple enough. Assum-
ing that the Margules equation is appropriate to the data, one merely searches for
values of the parameters A12 and A21 that yield pressures by Eq. (13.57) that are as
close as possible to the measured values. The method is applicable regardless of
the correlating equation assumed and is known as Barker’s method.16 Applied to
the present data set, it yields the parameters:
A 21 = 0.596 and A 12 = 1.153
Use of these parameters in Eqs. (13.39), (13.40), (13.41), (13.57), and (13.58) pro-
duces the results described by the dashed lines of Figs. 13.8(a) and 13.8(b). The
correlation cannot be precise, but it clearly provides a better overall representation
of the experimental P-x1-y1 data. Note, however, that it necessarily provides a
0.2
0.1
GE
RT
0.2
0 0.5 1.0
x1
worse fit to the experimentally derived ln γ1, ln γ2, and (GE∕(x1x2RT )). This fitting
procedure ignores the vapor-phase composition data from which those experimen-
tally derived activity coefficients were determined.
RT [ ]
P 1
ˆ v = exp ___
ϕ i ∑ ∑ y j y k(2δj i − δ jk) (13.60)
B ii + _
2 j k
δj i ≡ 2 B ji − B jj − B ii δ jk ≡ 2 B jk − B jj − B kk
where
with δii = 0, δjj = 0, etc., and δij = δji, etc. Values of the virial coefficients come from a gen-
eralized correlation, as represented for example by Eqs. (10.69) through (10.74). The fugacity
coefficient for pure i as a saturated vapor ϕisat
is obtained from Eq. (13.60) with δji and δjk set
equal to zero:
B ii Pisat
= exp ______
ϕ isat
(13.61)
RT
13.5. Fitting Activity Coefficient Models to VLE Data 483
B 11(P − P1sat
) +P y22 δ 12
Φ 1 = exp ___________________
(13.63)
RT
B 22(P − P2sat
) +P y12 δ 12
Φ 2 = exp ___________________
(13.64)
RT
The inclusion of Φ1 and Φ2 evaluated by Eqs. (13.63) and (13.64) in the reduction of a set of
moderate-pressure VLE data is straightforward, because in this situation the values of T, P,
and y1 at each data point are known. After evaluating Φi, at each data point, the activity coef-
ficients are calculated from:
y i Φ i P
γ i = ______
(13.65)
x i Pisat
These experimentally derived activity coefficients are then combined via Eq. (13.54), and
fitting to an excess Gibbs energy model proceeds as usual.
On the other hand, inclusion of the vapor-phase fugacity coefficients introduces significant
new complications to bubblepoint calculations and flash calculations because in those cases the
vapor-phase composition is unknown. When vapor-phase fugacity coefficients were not included,
bubblepoint pressure calculations could be made directly, without requiring any iterative solution
procedure. However, with the inclusion of the vapor-phase fugacity coefficients, all types of
bubblepoint, dewpoint, and flash calculations require iterative solution.
( RT )
G E H E
= − ____
d _ 2 dT (const P, x)
R T
and the excess-property analog of Eq. (2.20):
dH E = CPE dT (const P, x)
Integration of the first of these equations from T0 to T gives:
∫ T 0 R T 2
___
= ___
dT (13.66)
T 0
484 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
∫ T 1
( ∂ T ) P, x
∂ CE
In addition. d CPE = ____
P dT
Combining this equation with Eqs. (13.66) and (13.67) leads to:
RT ( RT ) ( RT ) ( T 0 ) T
G E
___ E
G H E T 1
T
= ___
− ___ − 1 ___
_
T 0 T 1
(13.68)
R [ T 0 ( T 0 ) T ]
E
C P 2
____ _T _T T 1
_
− ln − − 1 − J
This general equation makes use of excess Gibbs-energy data at temperature T0, excess
enthalpy (heat-of-mixing) data at T1, and excess heat-capacity data at T2.
Evaluation of integral J requires information with respect to the temperature dependence
of C PE . Because of the relative paucity of excess-heat-capacity data, the usual assumption is
that this property is constant, independent of T. In this event, integral J is zero, and the closer
T0 and T1 are to T, the less the influence of this assumption. When no information is available
with respect to C PE , and excess enthalpy data are available at only a single temperature, the
excess heat capacity must be assumed to be zero. In this case only the first two terms on the
right side of Eq. (13.68) are retained, and it more rapidly becomes imprecise as T increases.
Because the parameters of two-parameter correlations of GE data are directly related
to infinite-dilution values of the activity coefficients, our primary interest in Eq. (13.68) is
its application to binary systems at infinite dilution of one of the constituent species. For this
purpose, we divide Eq. (13.68) by the product x1x2. For C PE independent of T (and thus with
J = 0), it then becomes:
The preceding equation applied at infinite dilution of species i may therefore be written:
Data for the ethanol(1)/water(2) binary system provide a specific illustration. At a base tem-
perature T0 of 363.15 K (90°C), the VLE data of Pemberton and Mash17 yield accurate values
for infinite-dilution activity coefficients:
Correlation of the excess enthalpy data of J. A. Larkin18 at 110°C yields the values:
(_______
x 1 x 2 RT ) ( x 1 x 2 RT )
H E H E
= − 0.0598 and _______
= 0.6735
T 1, x 1 = 0 T 1, x 2 = 0
Correlations of the excess enthalpy for the temperature range from 50 to 110°C lead to
infinite-dilution values of CPE ∕ (x 1 x 2 R), which are nearly constant and equal to
Equation (13.69) may be directly applied with these data to estimate ln γ1∞ and ln γ2∞ for tem-
peratures greater than 90°C. The van Laar equations [Eqs. (13.43) and (13.44)] are appropriate
here, with parameters directly related to the infinite-dilution activity coefficients:
A 12 γ1∞ and A 21
′ = ln γ2∞
′ = ln
These data allow prediction of VLE by an equation of state at 90°C and at two higher tem-
peratures, 423.15 and 473.15 K (150 and 200°C), for which measured VLE data are given by
Barr-David and Dodge.19 Pemberton and Mash report pure-species vapor pressures at 90°C
for both ethanol and water, but the data of Barr-David and Dodge (at 150 and 200°C) do not
include these values. They are therefore calculated from reliable correlations. Results of cal-
culations based on the Peng/Robinson equation of state are given in Table 13.6. Shown for
the three temperatures are values of the van Laar parameters A 12 ′ and A 21
′ , the pure-species
sat sat
vapor pressures P
1 and P
2 , the equation-of-state parameters bi and qi, and root-mean-square
(RMS) deviations between computed and experimental values for P and y1.
17R. C. Pemberton and C. J. Mash, Int. DATA Series, Ser. B, vol. 1, p. 66, 1978.
18As reported in Heats of Mixing Data Collection, Chemistry Data Series, vol. Ill, part 1, pp. 457–459, DECHEMA,
Frankfurt/Main, 1984.
19F. H. Barr-David and B. F. Dodge, J. Chem. Eng. Data, vol. 4, pp. 107–121, 1959.
486 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
T °C ′
A 12 ′
A 21 P1sat
P2sat
q1 q2 RMS RMS
bar bar % δP δy1
90 1.7720 0.9042 1.5789 0.7012 12.0364 15.4551 0.29 *****
150 1.7356 0.7796 9.825 4.760 8.8905 12.2158 2.54 0.005
200 1.5204 0.6001 29.861 15.547 7.0268 10.2080 1.40 0.005
***** Vapor-phase compositions not measured.
30
200ºC
25
20
P/bar
15
10
150ºC
Figure 13.10: Pxy diagram for ethanol(1)/water(2). The lines represent predicted values; the points are
experimental values.
The small value of RMS % δP shown for 90°C indicates both the suitability of the
van Laar equation for correlation of the VLE data and the capability of the equation of state
to reproduce the data. A direct fit of these data with the van Laar equation by the gamma/phi
Rev. Confirming Pages
procedure yields RMS % δP = 0.19.20 The results at 150 and 200°C are based only on
vapor-pressure data for the pure species and on mixture data at lower temperatures. The quality
of prediction is indicated by the P-x-y diagram of Fig. 13.13, which reflects the uncertainty of
the data as well.
In Sec. 6.3 we treated the calculation of residual properties by the virial equations of state, as
well as from generalized correlations, but we did not extend the treatment to cubic equations
of state at that time. The key differentiating feature of cubic equations of state is their ability to
treat both vapor and liquid phase properties. This capability is most valuable in the context of
VLE calculations, as considered in the present chapter. Thus, in this section we first treat the
computation of residual properties from cubic equations of state and then, in Sec. 13.7, show
how these can be used to carry out phase equilibrium calculations.
Results of some generality follow from the generic cubic equation of state:
RT a(T )
P = _____
− _____________
(3.41)
V − b (V + εb)(V + σb)
Equations (6.58) through (6.60) are compatible with pressure-explicit equations of state.
We need only recast Eq. (3.41) to yield Z with density ρ as the independent variable. We
therefore divide Eq. (3.41) by ρRT and substitute V = 1∕ρ. With q given by Eq. (3.47),
q ≡ a(T )∕bRT, the result after algebraic reduction is:
1 ρb
Z = _____
− q ______________
1 − ρb ( 1 + ερb)( 1 + σρb)
The two quantities needed for evaluation of the integrals in Eqs. (6.58) through (6.60), Z − 1
and (∂Z∕∂T )ρ, are readily obtained from this equation:
ρb ρb
Z − 1 = _____ − q _______________
1 − ρb (1 + ερb)(1 + σρb)
(13.70)
( ∂ T ) ρ ( dT ) (1 + ερb)(1 + σρb)
∂ Z
___ dq _______________
_ ρb
= −
The integrals of Eqs. (6.58) through (6.60) are now evaluated as follows:
(Z − 1) ___
= _____ − q ______________
ρ ρ ρ
∫0
dρ
∫ 0 1 − ρb ρb
ρb _____ d(ρb)
∫ 0 (1 + ερb) (1 + σρb)
d(ρb)
ρ
___ ______________
∫ 0 ( ∂ T ) ρ ρ
ρ
∂ Z dρ dq ρ
dT ∫ 0 (1 + ερb) (1 + σρb)
d(ρb)
___
= − ___
20As
reported in Vapor-Liquid Equilibrium Data Collection, Chemistry Data Series, vol. 1, part 1a, p. 145,
DECHEMA, Frankfurt/Main, 1981.
The generic equation of state presents two cases for the evaluation of this integral:
σ − ε ( 1 + ερb )
1 1 + σρb
I = ____
Case I: ε ≠ σ ln _
(13.71)
σ − ε ( Z + εβ )
1 Z + σβ
I = ____
ln _
(13.72)
ρb β
I = ______
Case II: ε = σ = _____
1 + ερb Z + εβ
The van der Waals equation is the only one considered here to which Case II applies, and this
equation, with ε = 0, reduces to I = β∕Z.
With evaluation of the integrals, Eqs. (6.58) through (6.60) reduce to:
G R
___
= Z − 1 − ln [ (1 − ρb) Z ] − qI (13.73)
RT
_G R
or
= Z − 1 − ln (Z − β) − qI (13.74)
RT
( dT ) ( d Tr )
H R
___ dq dq
= Z − 1 + T _
I = Z − 1 + Tr _
I
RT
( d Tr )
S R
___ dq
and = ln (Z − β) + q + Tr _
I
R
dq
The quantity Tr ____
is readily found from Eq. (3.51):
d Tr
d Tr [ d ln Tr ]
dq d ln α(Tr )
Tr ____
= _
− 1 q
[ ]
H R d ln α(Tr )
_
= Z − 1 + _
− 1 qI (13.75)
RT d ln Tr
13.6. Residual Properties by Cubic Equations of State 489
_S R d ln α(Tr )
= ln (Z − β) + _
qI (13.76)
R d ln Tr
Before applying these equations one must find Z by solution of the equation of state itself, typ-
ically written in the form of Eq. (3.48) or Eq. (3.49) for a vapor or liquid phase, respectively.
Example 13.5
Find values for the residual enthalpy HR and the residual entropy SR of n-butane gas at
500 K and 50 bar as given by the Redlich/Kwong equation.
Solution 13.5
For the given conditions:
500 50
Tr = _____
= 1.176 Pr = _____
= 1.317
425.1 37.96
By Eq. (3.50), with Ω for the Redlich/Kwong equation from Table 3.1,
Pr (0.08664)(1.317)
β = Ω ___
= ______________
= 0.09703
Tr 1.176
With values for Ψ and Ω, and with the expression α r−1/2
(Tr ) = T from Table 3.1,
Eq. (3.51) yields:
Ψα( Tr ) _______________
0.42748
q = ______
=
= 3.8689
Ω Tr (0.08664) (1.176) 1.5
Substitution of these values of β and q, along with ε = 0, and σ = 1 into Eq. (3.48)
reduces it to:
Z − 0.09703
Z = 1 + 0.09703 − (3.8689) (0.09703) ____________
Z(Z + 0.09703)
Iterative solution of this equation yields Z = 0.6850. Then:
Z + β
I = ln ____
= 0.13247
Z
1 d ln α( Tr ) 1
Given that ln α(Tr ) = − __ , we have ________
ln Tr = − __
, and Eqs. (13.75)
2 d ln Tr 2
and (13.76) become:
H
___
R
= 0.6850 − 1 + (− 0.5 − 1) (3.8689) (0.13247) = − 1.0838
RT
S
___
R
= ln (0.6850 − 0.09703) − (0.5) (3.8689) (0.13247) = − 0.78735
R
490 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
H R = (8.314)(500)(− 1.0838) = − 4505 J⋅mol −1
Whence,
S R = (8.314)(− 0.78735) = − 6.546 J⋅mol −1⋅K −1
These results are compared with those of other calculations in Table 13.7.
Table 13.7: Values for Z, HR, and SR for n-butane at 500 K and 50 Bar
As shown in Sec. 10.6, phases at the same T and P are in equilibrium when the fugacity of
each species is the same in all phases. For VLE, this requirement is written:
fˆ i v = fˆ i l (i = 1, 2, . . . , N )
(10.48)
An alternative form results from the introduction of the fugacity coefficient, defined by Eq. (10.52):
y i ϕˆ i v P = x i ϕˆ i l P
Applications of this equation with fugacity coefficients evaluated using cubic equations of
state are presented in the following subsections.
approaches the constant b [Eq. 3.41]. The segment on the right (tu) with gentle downward
slope is characteristic of vapors; in the limit as P → 0 molar volume V approaches infinity.
The middle segment (st), containing both a minimum (note here that P < 0) and a maximum,
provides a smooth transition from liquid to vapor, but has no physical meaning. The actual
transition from liquid to vapor occurs at the vapor pressure along a horizontal line like that
which connects points M and W.
P
r
t
P'
M W Figure 13.11: Isotherm for T < Tc
u on a PV diagram for a pure fluid.
0
V
For pure species i, Eq. (13.77) reduces to Eq. (10.41), ϕiv = ϕil , which may be written:
The fugacity coefficient of a pure liquid or vapor is a function of its temperature and pressure.
For a saturated liquid or vapor, the equilibrium pressure is Pisat
, and Eq. (13.78) implicitly
expresses the functional relation,
g(T, Pisat
) = 0 or Pisat
= f (T )
An isotherm generated by a cubic equation of state, as represented in Fig. 13.11, has
three volume roots for a specified pressure between P = 0 and P = P′. The smallest root lies
on the left line segment and is a liquid-like volume, e.g., at point M. The largest root lies on the
right line segment and is a vapor-like volume, e.g., at point W.
If these points lie at the vapor pressure, then M represents saturated
liquid, W represents saturated vapor, and they exist in phase equilibrium.
The root lying on the middle line segment has no physical significance.
Two widely used cubic equations of state, developed specifically for VLE calculations,
are the Soave/Redlich/Kwong (SRK) equation21 and the Peng/Robinson (PR) equation.22
Both are special cases of the generic cubic equation of state, Eq. (3.41). Equation-of-state
21G. Soave, Chem. Eng. Sci., vol. 27, pp. 1197–1203, 1972.
22D.-Y. Peng and D. B. Robinson, Ind. Eng. Chem. Fundam., vol. 15, pp. 59–64, 1976.
492 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
parameters are independent of phase, and in accord with Eqs. (3.44) through (3.47) they are
given by:
bi P a i(T )
βi ≡ ___
(13.81) q i ≡ _____ (13.82)
RT bi RT
( q i βi )
1 + βi − Zil
Zil = βi + (Zil + ε βi )(Zil + σ βi ) __________
(13.84)
The pure numbers ε, σ, Ψ, and Ω and expressions for α(Tri) are specific to the equation of state
and are given in Table 3.1 for several prototypical cubic equations of state.
In Sec. 13.6 and Sec. 10.5, we developed the following two relationships:
G iR
___ = Z i − 1 − ln(Z i − βi ) − q i I i (13.74)
RT
G iR
___ = ln ϕi (10.33)
RT
bi Pisat
βi ≡ ______
RT
( q i βi )
1 + βi − Zil
Zil = βi + (Zil + ε βi )(Zil + σ βi ) __________
This equation yields I iv with Z iv from Eq. (13.83), and Iil with Zil from Eq. (13.79). However,
iv and Zil contain qi, the quantity sought. Thus, a solution procedure is
the equations for Z
required that yields values for eight unknowns given eight equations. An initial value of qi is
provided by the generalized correlation of Eqs. (13.79), (13.80), and (13.82).
Mixture VLE
The fundamental assumption when an equation of state is written for mixtures is that
it has exactly the same form as when written for pure species. Thus for mixtures,
Eqs. (13.83) and (13.84), written without subscripts, become:
Z v − β v
Z v = 1 + β v − q v β v ________________
Vapor:
(13.87)
( Z + ε β v) ( Z v + σ β v)
v
( q l β l )
1 + β l − Z l
Z l = β l + (Z l + ε β l)(Z l + σ β l) ________
Liquid: (13.88)
494 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
Here, βl, βv, ql, and qv are for mixtures, with definitions:
b p P a p
β p ≡ ____
(p = l, v) (13.89) q p ≡ _____
p (p = l, v) (13.90)
RT b RT
The complication is that mixture parameters ap and bp, and therefore βp and qp, are functions
of composition. Systems in vapor/liquid equilibrium consist in general of two phases with
different compositions. The PV isotherms generated by an equation of state for these two
fixed compositions are represented in Fig. 13.12 by two similar lines: the solid line for the
liquid-phase composition and the dashed line for the vapor-phase composition. They are dis-
placed from one another because the equation-of-state parameters are different for the two
compositions. However, each line includes three segments as described in connection with
the isotherm of Fig. 13.11. Thus, we distinguish between the composition that characterizes
a complete line and the phases, all of the same composition, that are associated with the seg-
ments of an isotherm.
Each line contains a bubblepoint on its left segment representing saturated liquid and a
dewpoint of the same composition on its right segment representing saturated vapor23.
Because these points for a given line are for the same composition, they do not represent
phases in equilibrium and do not lie at the same pressure. (See Fig. 12.3, where for a given
constant-composition loop and a given T, saturated liquid and saturated vapor are at different
pressures.)
For a BUBL P calculation, the temperature and liquid composition are known, and this
fixes the location of the PV isotherm for the composition of the liquid phase (solid line).
The BUBL P calculation then finds the composition for a second (dashed) line that con-
tains a dewpoint D on its vapor segment that lies at the pressure of the bubblepoint B on
the liquid segment of the solid line. This pressure is then the phase-equilibrium pressure,
23Note that bubblepoint B and dewpoint D in Fig. 13.12 are on different lines.
13.7. VLE from Cubic Equations of State 495
and the composition for the dashed line is that of the equilibrium vapor. This equilibrium
condition is shown by Fig. 13.12, where bubblepoint B and dewpoint D lie at the same P on
isotherms for the same T but representing the different compositions of liquid and vapor in
equilibrium.
Because no established theory prescribes the form of the composition dependence of
the equation-of-state parameters, empirical mixing rules have been proposed to relate mixture
parameters to pure-species parameters. The simplest realistic expressions are a linear mixing
rule for parameter b and a quadratic mixing rule for parameter a:
with aij = aji. The general mole-fraction variable xi is used here because these mixing
rules are applied to both liquid and vapor mixtures. The aij are of two types: pure-species
parameters (repeated subscripts, e.g., a11) and interaction parameters (unlike subscripts, e.g.,
a12). Parameter bi is for pure species i. The interaction parameters aij are often evaluated from
pure-species parameters by combining rules, e.g., a geometric-mean rule:
a ij = (a i a j) 1/2 (13.93)
These equations, known as van der Waals prescriptions, provide for the evaluation of mixture
parameters solely from parameters for the pure constituent species. Although they are satis-
factory only for mixtures comprised of simple and chemically similar molecules, they allow
straightforward calculations that illustrate how complex VLE problems can be solved.
Also useful for application of equations of state to mixtures are partial equation-of-state
parameters, defined by:
suited to these variables. The derivation of such an equation starts with Eq. (10.56), written for
a mixture with VR replaced by Eq. (6.40), VR = RT(Z − 1)∕P:
( RT )
n G R n(Z − 1) n H R
d _
= _______
dP − ____
2 dT + ∑ ln ϕˆ i d n i
P R T i
Division by dni and restriction to constant T, n∕ρ (= nV ), and nj ( j ≠ i) leads to:
[ ] ( ∂ n i) T, n∕ρ, n j
∂ (n G R ∕ RT ) n(Z − 1) ___ ∂ P
ln ϕˆ i = __________
− _______
(13.97)
∂ n i P
T, n∕ρ, n j
For simplicity of notation, the partial derivatives in the following development are w
ritten
without subscripts, and they are understood to be at constant T, n∕ρ, and nj. Thus, with
P = (nZ)RT∕(n∕ρ),
∂ P
___ RT ∂ (nZ) ___ P ∂ (nZ)
= ___
_____
= _____
(13.98)
∂ n i n ∕ ρ ∂ n i nZ ∂ n i
( Z ) ∂ n i Z ( ∂ n i )
∂ (n G R ∕ RT ) ∂ (nZ) __________
Z − 1 _____ ∂ (n G R ∕ RT ) _____
∂ (nZ) __ 1 ∂ Z
ln ϕˆ i = __________
− _
=
− + n _
+ Z
∂ n i ∂ n i ∂ n i
Equation (13.73), written for the mixture and multiplied by n, is differentiated to give the first
term on the right:
n G
R
____ = nZ − n − n ln [ (1 − ρb) Z ] − (nq) I
RT
[ ∂ n i ]
∂ (n G R ∕ RT ) _____
__________ ∂ (nZ) ∂ ln (1 − ρb) _
___________ ∂ ln Z ∂ I
___
= − 1 − ln [(1 − ρb) Z ] − n
+ − nq − I q¯ i
∂ n i ∂ n i ∂ n i ∂ n i
Z ( ∂ n i )
n ___
__ ∂ Z ∂ I
___ ∂ (nZ) __
_____ 1 _ ∂ Z
− − nq − I q¯ i −
+ n + Z
Z ∂ n i ∂ n i ∂ n i
∂ (ρb)
n _____ ∂ I
This reduces to: ln ϕˆ i = _____
− nq ___
− ln [ (1 − ρb) Z ] − q¯ i I
1 − ρb ∂ n i ∂ n i
All that remains is evaluation of the two partial derivatives. The first is:
( n ∕ ρ ) ρ
nb
∂ _
∂ (
ρ b )
_____
= ______
b¯ i
= __
∂ n i ∂ n i n
13.7. VLE from Cubic Equations of State 497
The second follows from differentiation of Eq. (13.71). After algebraic reduction this yields:
∂ I ∂ ( ρb)______________
___ 1 b¯ i ρb
= _____
= ___ ______________
∂ n i ∂ n i ( 1 + σρb)(1 + ερb) nb (1 + σρb)(1 + ερb)
ˆ i reduces it to:
Substitution of these derivatives in the preceding equation for ln ϕ
b¯ i
b [ 1 − ρb (1 + ερb)(1 + σρb)]
ρb ρb
ln ϕˆ i = __ _
− q ________________
− ln [ (1 − ρb) Z ] − q¯ i I
Reference to Eq. (13.70) shows that the term in the first set of square brackets is Z − 1. Therefore,
b¯ i
ln ϕˆ i = __ (Z − 1) − ln [(1 − ρb)Z] − q¯ i I
b
bP P β
β ≡ ___
Moreover, and Z ≡ ____ ; whence ρb = __
RT ρRT Z
b ¯ i
Thus, ln ϕˆ i = __ (Z − 1) − ln (Z − β) − q¯ i I
b
Because experience has shown that Eq. (13.91) is an acceptable mixing rule for parameter b,
it is adopted here. Whence,
nb = ∑ n i bi
i
where I is evaluated by Eq. (13.72). For the special case of pure species i, this becomes:
ln ϕi = Z i − 1 − ln( Z i − βi ) − q i I i (13.100)
( )
a¯ i b¯ i
[ ∂ n i ] ( b)
∂ (nq) a¯ i bi
¯ i ≡ _____
Whence, q = q 1 + __ − __ = q 1 + __ − _
(13.101)
T, n
a b a
j
Any two of the three partial parameters form an independent pair, and any one of them can be
found from the other two.24
24 ¯ i ≠ a¯ i ∕b¯ i RT
Because q, a, and b are not linearly related, q
498 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
Example 13.6
A vapor mixture of N2(1) and CH4(2) at 200 K and 30 bar contains 40 mol-% N2. Deter-
mine the fugacity coefficients of nitrogen and methane in the mixture by Eq. (13.99)
and the Redlich/Kwong equation of state.
Solution 13.6
For the Redlich/Kwong equation, ε = 0 and σ = 1, and Eq. (13.87) becomes:
Z − β
Z = 1 + β − qβ _______
(A)
Z(Z + β)
where β and q are given by Eqs. (13.89) and (13.90). Superscripts are here omitted
because all calculations are for a vapor phase. The mixing rules most commonly
used with the Redlich/Kwong equation for parameters a(T ) and b are given by
Eqs. (13.91) through (13.93). For a binary mixture they become:
____
a = y12 a 1 + 2 y 1 y 2 √ a 1 a 2 + y22 a 2(B)
b = y 1 b1 + y 2 b2 (C)
In Eq. (B), a1 and a2 are pure-species parameters given by Eq. (13.79) written for
the Redlich/Kwong equation:
Tr−1/2 2 2
i (83.14) Tc i
a i = 0.42748 _______________
bar·cm 6·mol −2(D)
Pc i
Critical constants for nitrogen and methane from Table B.1 of App. B and
calculated values for bi and ai from Eqs. (D) and (E) are:
β(Z − β)
Z = 1 + β − 3.6916 _______
with β = 0.051612
Z(Z + β)
13.7. VLE from Cubic Equations of State 499
( b)
2 y 1 a 1 + 2 y 2 √ a 1 a 2 _ b1
¯ 1 = q _________________
q −
(F)
a
_
( b)
2 y 2 a 2 + 2 y 1 √ a 1 a 2 _ b2
¯ 2 = q _________________
q
− (G)
a
Substitution of numerical values into these equations and into Eq. (13.99) leads to
the following results:
Equations (13.77) and (13.99) provide the basis for VLE calculations for mixtures, but
they incorporate a number of mixture parameters (e.g., al, bv) and thermodynamic functions
(e.g., Zl, Zv) that may initially be unknown. The calculation therefore becomes one of solving
simultaneous equations equal in number to the number of unknowns. The equations available
fall into several classes: mixing- and combining-rule and parameter equations for the liquid
phase, the same equations for the vapor phase, and equilibrium and related equations. All
of these parameters and equations have already been presented, but they are classified for
convenience in Table 13.9. The presumption is that all pure-species parameters (e.g., ai and bi)
are known and that either the temperature T or pressure P and either the liquid-phase or vapor-
phase composition is specified. In addition to the N primary unknowns (T or P and either
liquid or vapor phase composition), Table 13.9 enumerates 12 + 4N auxiliary variables and a
total of 12 + 5N equations.
Rev. Confirming Pages
β l = b l P ∕ (RT ) q l = a l ∕ (b l RT )
( a b l)
a¯ l bi
q¯ il = q l 1 + __il − __ (i = 1, 2, . . . , N)
β v = b vP ∕ (RT ) q v = a v ∕ ( b v RT )
( a b v)
a¯ v bi
q¯ vi = q v 1 + ___iv − _ (i = 1, 2, . . . , N)
4 + 3N equations; 4 + 2N variables: { ϕˆ i v }, { ϕˆ i l }, I l, Z l, I v, Z v
A direct and somewhat intuitive solution procedure makes use of Eq. (13.77), rewritten
as yi = Ki xi. Because ∑ i y i = 1,
Example 13.7
Develop the Pxy diagram at 37.78°C for the methane(1)/n-butane(2) binary system.
Base calculations on the Soave/Redlich/Kwong equation with mixing rules given by
Eqs. (13.91) through (13.93). Experimental data at this temperature for comparison are
given by Sage et al.25
Solution 13.7
The procedure here is to do a BUBL P calculation for each experimental data
point. For each calculation, estimated values of P and y1 are required to initiate
iteration. These estimates are here provided by the experimental data. Where no
such data are available, several trials may be required to find values for which the
iterative procedure converges.
Pure-species parameters ai and bi are found from Eqs. (13.79) and (13.80) with
constants and an expression for α(Tr) from Table 3.1. For a temperature of 310.93 K
[37.78°C] and with critical constants and ωi from Table B.1, calculations provide
the following pure-species values:
c i ∕K
T Tr i ω i α( Tr ) P
c i ∕bar bi 10–6ai
CH4(1) 190.6 1.6313 0.012 0.7425 45.99 29.853 1.7331
n-C4H10(2) 425.1 0.7314 0.200 1.2411 37.96 80.667 17.458
25B. H. Sage, B. L. Hicks, and W. N. Lacey, Industrial and Engineering Chemistry, vol. 32, pp. 1085–1092, 1940.
502 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
The mixing rules adopted here are the same as in Ex. 13.6, where Eqs. (B), (C),
(F), and (G) give mixture parameters for the vapor phase. When applied to the
liquid phase, xi replaces yi as the mole-fraction variable:
____
a l = x12 a 1 + 2x 1x 2 √ a 1a 2 + x22 a 2 b l = x 1b1 + x 2b2
_ _
( b ) ( b )
2x 1a 1 + 2x 2 √ a 1a 2 _ b 1 2x 2a 2 + 2x 1√ a 1a 2 _ b2
l l _________________
q¯ 1 = q
l
− l
q l l _________________
¯ 2 = q
l
− l
a a
( q β )
1 + β l − Z l Z v − β v
Z l = β l + Z l(Z l + β l) _ Z v = 1 + β v − q v β v _________
v v
l l Z (Z + β v)
where βl, βv, ql, and qv are given by Eqs. (13.89) and (13.90). The first set of BUBL P
calculations is made for the assumed pressure. With the given liquid-phase com-
position and assumed vapor-phase composition, values for Zl and Zv are deter-
mined by the preceding equations, and fugacity coefficients ϕ ˆ l and ϕˆ v then follow
i i
from Eq. (13.99). Values of K1 and K2 come from Eq. (13.103). The constraint
y1 + y2 = 1 has not been imposed, and Eq. (13.102) is unlikely to be satisfied. In
this event, K1x1+ K2x2 ≠ 1, and a new vapor composition for the next iteration is
given by the normalizing equation:
K 1 x 1
y 1 = __________
with y 2 = 1 − y 1
K 1 x 1 + K 2 x 2
This new vapor composition allows reevaluation of { ϕˆ i v }, {Ki}, and {Kixi}. If the
sum K1x1 + K2x2 has changed, a new vapor composition is found and the sequence
of calculations is repeated. Continued iteration leads to stable values of all quan-
tities. If the sum K1x1 + K2x2 is not unity, the assumed pressure is incorrect, and
it must be adjusted according by some rational scheme. When ∑ i K i x i > 1, P is
too low; when ∑ i K i x i < 1, P is too high. The entire iterative procedure is then
repeated with a new pressure P. The last calculated values of yi are used for the
initial estimate of {yi}. The process continues until K1x1 + K2x2 = 1. Of course
the same result can be obtained by any other method capable of solving the set of
equations given in Table 13.9, with P, y1, and y2 as the unknowns.
The results of all calculations are shown by the solid lines of Fig. 13.13. Exper-
imental values appear as points. The root-mean-square percentage difference
between experimental and calculated pressures is 3.9%, and the root-mean-square
deviation between experimental and calculated y values is 0.013. These results,
based on the simple mixing rules of Eqs. (13.91) and (13.92), are representative
for systems that exhibit modest and well-behaved deviations from ideal-solution
behavior, e.g., for systems comprised of hydrocarbons and cryogenic fluids.
13.8. Flash Calculations 503
140
Critical Point
120
100
80
P/bar
60
40
20
0
0 0.2 0.4 0.6 0.8 1
x1, y1
Figure 13.13: Pxy diagram at 37.8°C (100°F) for methane(1)/n-butane(2). Lines represent values from
BUBL P calculations with the SRK equation; points are experimental values.
In previous sections, we have focused on bubblepoint and dewpoint calculations, which are
common calculations in practice, and which provide a basis for constructing phase diagrams
for VLE. Perhaps an even more important application of VLE is the flash calculation. The
name originates from the fact that a liquid at a pressure equal to or greater than its bubblepoint
pressure “flashes” or partially evaporates when the pressure is reduced below its bubblepoint
pressure, producing a two-phase system of vapor and liquid in equilibrium. We consider here
only the P, T-flash, which refers to any calculation of the quantities and compositions of the
504 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
vapor and liquid phases making up a two-phase system in equilibrium at known T, P, and
overall composition. This poses a problem known to be determinate on the basis of Duhem’s
theorem, because two independent variables (T and P) are specified for a system of fixed over-
all composition, that is, a system formed from given masses of nonreacting chemical species.
Consider a system containing one mole of nonreacting chemical species with an overall
composition represented by the set of mole fractions {zi}. Let be the moles of liquid, with
mole fractions {xi}, and let be the moles of vapor, with mole fractions {yi}. The material
balance equations are:
+ = 1
z i = x i + y i (i = 1, 2, . . . , N )
Combining these equations to eliminate gives:
z i = x i(1 − ) + y i (13.104)
The K-value, as defined in the previous section (Ki ≡ yi∕xi), is a convenient construct for use in
flash calculations. Substituting xi = yi∕Ki into Eq. (13.104) and solving for yi yields:
z i K i
y i = ___________
(i = 1, 2, . . . , N) (13.105)
1 + (K i − 1)
Because ∑ i y i = 1, summing this expression over all species gives us a single equation in
which, for known K-values, the only unknown is .
z i K i
∑ ___________
= 1 (13.106)
i 1 + (K i − 1)
One general approach to solving a P, T-flash problem is to find the value of , between 0 and 1,
that satisfies this equation. Note that = 1 is always a trivial solution to this e quation. Having
done so, the vapor-phase mole fractions are then obtained from Eq. (13.105), the liquid-phase
mole fractions are obtained from xi = yi∕Ki, and is given by = 1 − . When Raoult’s law
can be applied, the K-values are constant and this is straightforward, as shown in the following
example. Historically, K-values for light hydrocarbons were often taken from a set of charts
constructed by DePriester, hence called DePriester charts. These similarly provided a set of
constant K-values for use in the preceding calculations.26
Example 13.8
The system acetone(1)/acetonitrile(2)/nitromethane(3) at 80°C and 110 kPa has the
overall composition z1 = 0.45, z2 = 0.35, z3 = 0.20. Assuming that Raoult’s law is
appropriate to this system, determine , , {xi}, and {yi}. The vapor pressures of the
pure species at 80°C are:
P1sat
= 195.75 P2sat
= 97.84 P3sat
= 50.32 kPa
26C. L. DePriester, Chem. Eng. Progr. Symp. Ser. No. 7, vol. 49, p. 42, 1953.
13.8. Flash Calculations 505
Solution 13.8
First, do a BUBL P calculation with {zi} = {xi} to determine Pbubl:
P bubl = x 1 P1sat
+ x 2 P2sat
+ x 3 P3sat
P bubl = (0.45)(195.75) + (0.35)(97.84) + (0.20)(50.32) = 132.40 kPa
Because the given pressure lies between Pbubl and Pdew, the system is in the two-
phase region, and a flash calculation is possible.
From Raoult’s law, Eq. (13.16), we have K i = y i ∕ x i = Pisat
∕ P, from which:
K 1 = 1.7795 K 2 = 0.8895 K 3 = 0.4575
Substituting known values into Eq. (13.106) gives:
( 0.45)( 0.7795) ( 0.35)( 0.8895) ____________
____________ ( 0.20)( 0.4575)
+ ____________
+ = 1
1 + 0.7795 1 − 0.1105 1 − 0.5425
Trial-and-error or iterative solution for followed by evaluation of the other
unknowns yields:
= 0.7364 mol = 1 − = 0.2636 mol
( 0.45)( 1.7795)
_________________
y 1 =
= 0.5087 y 2 = 0.3389 y 3 = 0.1524
1 + (0.7795)( 0.7634)
y 1 0.5087
x 1 = ___
= ______
= 0.2859 x 2 = 0.3810 x 3 = 0.3331
K 1 1.7795
Reassuringly, ∑ i x i = ∑ i y i = 1.
The procedure of the preceding example is applicable regardless of the number of species
present. However, for the simple case of a binary flash calculation using Raoult’s law, explicit
solution is possible. In that case, we have:
P = x 1 P1sat
+ x 2 P2sat
= x 1 P1sat
+ (1 − x 1) P2sat
Solving for x1 gives:
P − P2sat
x 1 = _________
P1 − P2sat
sat
With x1 known, the remaining variables follow immediately from Raoult’s law [Eq. (13.16)]
and the overall mass balances [Eq. (13.104)].
When the K-values are not constant, the general approach remains the same as in
Ex. 13.8, but an additional level of iterative solution is required. In the preceding treatment,
506 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
we solved for yi to obtain Eq. (13.105). If, instead, we eliminate yi using yi = Kixi, we obtain
an alternative expression:
z i
x i = ___________
(i = 1, 2, . . . , N)(13.107)
1 + (K i − 1)
Because both sets of mole fractions must sum to unity, ∑ i x i = ∑ i y i = 1. Thus, if we sum
Eq. (13.105) over all species and subtract unity from this sum, the difference Fy is zero:
z i K i
Fy = ∑ ___________
− 1 = 0(13.108)
i 1 + (K i − 1)
Similar treatment of Eq. (13.107) yields the difference Fx, which is also zero:
z i
Fx = ∑ ___________
− 1 = 0(13.109)
i 1 + (K i − 1)
A P, T-flash problem can be solved by finding a value of that makes either Fy or Fx equal to
zero for known T, P, and overall composition. A more convenient function for applying a
general solution procedure27 is the difference F ≡ Fy −Fx:
z i( K i − 1)
F = ∑ ___________
= 0 (13.110)
i 1 + (K i − 1)
Because dF∕d is always negative, the F vs. relation is monotonic. This, in turn, makes this
form of the equation exceptionally well-suited for solution by Newton’s method (App. H).
Equation H.1 for the nth iteration of Newton’s method becomes:
(d )
dF
F + ____
Δ = 0 (13.112)
where Δ ≡ n+1 − n, and F and (dF∕d ) are found by Eq. (13.110) and (13.111). In these
equations, for the general gamma/phi formulation of VLE, the K-values come from
Eq. (13.13), written:
y i γ i Pisat
K i = __
= _____ (i = 1, 2, . . . , N) (13.113)
x i Φ i P
with the Φi given by Eq. (13.14). The K-values contain all of the thermodynamic information
and are related in a complex way to T, P, {yi}, and {xi}. Because solution is for {yi} and {xi},
the P, T-flash calculation inevitably requires iterative solution. This remains the case, even at
low pressure where we can assume Φi = 1, because the activity coefficients still depend upon
the unknown {xi}.
One typically proceeds by performing a BUBL P calculation and a DEW P calculation
prior to the flash calculation. If the given pressure is below Pdew for the specified T and {zi},
27H. H. Rachford, Jr., and J. D. Rice, J. Petrol. Technol., vol. 4(10), sec. 1, p. 19 and sec. 2, p. 3, October 1952.
13.9. Synopsis 507
then the system exists as superheated vapor, and no flash calculation is possible. Similarly, if
the given pressure is above Pbubl for the specified T and {zi}, then the system is a subcooled
liquid, and no flash calculation is possible. If the specified P falls between Pdew and Pbubl
for the specified T and {zi}, then the system exists as an equilibrium mixture of vapor and
liquid, and we can proceed with the flash calculation. The results of the preliminary DEW
P and BUBL P calculations then provide useful initial estimates of {γi}, {ϕˆ i}, and . For the
dewpoint, = 1, with calculated values of Pdew, γi, dew, and ϕ ˆ i, dew; for the bubblepoint, = 0,
ˆ i, bubl. The simplest procedure is to interpolate
with calculated values of Pbubl, γi, bubl, and ϕ
linearly between dewpoint and bubblepoint with respect to P:
P bubl − P
= __________
P bubl − P dew
P − P dew
γ i = γ i, dew + (γ i, bubl − γ i, dew) __________
P bubl − P dew
P − P dew
ˆ i = ϕˆ i , dew + (ϕˆ i , bubl − ϕˆ i , dew) __________
ϕ
P bubl − P dew
With these initial values of the γi and ϕ ˆ i , one can now calculate initial values of Ki from
Eq. (13.113). Using these values with Eqs. (13.110) and (13.111) one applies Newton’s
method, iterating on Eq. (13.112) to obtain a solution for . One then proceeds as in Ex. 13.8
to compute , {xi}, and {yi}. The computed compositions are used to obtain new estimates
of {γi} and {ϕˆ i }, from which new K-values are computed. The procedure is repeated until the
change in {xi} and {yi} from one iteration to the next is negligible. The same basic procedure
can be applied with K-values computed by application of a cubic equation of state to both
phases, as exemplified by Eq. (13.103).
13.9 SYNOPSIS
After studying this chapter, including the end-of-chapter problems, one should be able to:
∙ Understand the relationship between excess Gibbs energy and activity coefficients
∙ Explain and interpret each of the following five types of VLE calculations:
∙ Bubblepoint pressure (BUBL P) calculations
∙ Dewpoint pressure (DEW P) calculations
∙ Bubblepoint temperature (BUBL T ) calculations
∙ Dewpoint temperature (DEW T ) calculations
∙ P, T-flash calculations
∙ Carry out each of the five types of VLE calculations using each of the following VLE
formulations:
∙ Raoult’s law
∙ Modified Raoult’s law, with activity coefficients
508 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
13.10 PROBLEMS
Solutions to some of the problems of this chapter require vapor pressures as a function of tem-
perature. Table B.2, Appendix B, lists parameter values for the Antoine equation, from which
these can be computed.
13.1 Assuming the validity of Raoult’s law, do the following calculations for the ben-
zene(1)/toluene(2) system:
13.2. Assuming Raoult’s law to be valid, prepare a Pxy diagram for a temperature of 90°C
and a txy diagram for a pressure of 90 kPa for one of the following systems:
(a) Benzene(1)/ethylbenzene(2).
(b) 1-Chlorobutane(1)/chlorobenzene(2).
13.10. Problems 509
(a) What are the values of x1 and y1 at t = 55°C and P = _12 (P 1sat
+ P 2sat
)? For these
conditions plot the fraction of system that is vapor vs. overall composition z1.
(b) For t = 55°C and z1 = 0.5, plot P, x1, and y1 vs. .
13.5. Prove: An equilibrium liquid/vapor system described by Raoult’s law cannot exhibit
an azeotrope.
13.6. Of the following binary liquid/vapor systems, which can be approximately modeled by
Raoult’s law? For those that cannot, why not? Table B.1 (App. B) may be useful.
13.8. Do all four parts of Prob. 13.7, and compare the results. The required temperatures
and pressures vary significantly. Discuss possible processing implications of the vari-
ous temperature and pressure levels.
13.10. Do all four parts of Prob. 13.9, and compare the results. Discuss any trends that appear.
510 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
13.11. A binary mixture of mole fraction z1 is flashed to conditions T and P. For one of the
following, determine the equilibrium mole fractions x1 and y1 of the liquid and vapor
phases formed, the molar fraction of the vapor formed, and the fractional recovery
of species 1 in the vapor phase (defined as the ratio for species 1 of moles in the
vapor to moles in the feed). Assume that Raoult’s law applies.
13.12. Humidity, relating to the quantity of moisture in atmospheric air, is accurately given
by equations derived from the ideal-gas law and Raoult’s law for H2O.
(a) The absolute humidity h is defined as the mass of water vapor in a unit mass of dry
air. Show that it is given by:
ℳ H O_______
p H 2O
h = _____
2
ℳ air P − p H 2O
where ℳ represents a molar mass and p H 2O is the partial pressure of the water
vapor, i.e., p H 2O = y H 2O P.
(b) The saturation humidity hsat is defined as the value of h when air is in equilibrium
with a large body of pure water. Show that it is given by:
pHsat 2O
ℳ H O________
h sat = _____
2
ℳ air P − pHsat O
2
where p Hsat 2O is the vapor pressure of water at the ambient temperature.
(c) The percentage humidity is defined as the ratio of h to its saturation value,
expressed as a percentage. On the other hand, the relative humidity is defined
as the ratio of the partial pressure of water vapor in air to its vapor pressure,
expressed as a percentage. What is the relation between these two quantities?
13.13. A concentrated binary solution containing mostly species 2 (but x2 ≠ 1) is in equilib-
rium with a vapor phase containing both species 1 and 2. The pressure of this two-
phase system is 1 bar; the temperature is 25°C. At this temperature, 1 = 200 bar and
P2sat
= 0.10 bar. Determine good estimates of x1 and y1. State and justify all assumptions.
13.14. Air, even more than carbon dioxide, is inexpensive and nontoxic. Why is it not the gas
of choice for making soda water and (cheap) champagne effervescent? Table 13.2 may
provide useful data.
13.15. Helium-laced gases are used as breathing media for deep-sea divers. Why? Table 13.2
may provide useful data.
13.16. A binary system of species 1 and 2 consists of vapor and liquid phases in equilibrium
at temperature T. The overall mole fraction of species 1 in the system is z1 = 0.65. At
temperature T, ln γ 1 = 0.67 x22 ; ln γ 2 = 0.67 x12 ; P1sat
= 32.27 kPa; and P2sat
= 73.14 kPa.
Assuming the validity of Eq. (13.19),
13.10. Problems 511
(a) Over what range of pressures can this system exist as two phases at the given T and z1?
(b) For a liquid-phase mole fraction x1 = 0.75, what is the pressure P and what molar
fraction of the system is vapor?
(c) Show whether or not the system exhibits an azeotrope.
13.17. For the system ethyl ethanoate(1)/n-heptane(2) at 343.15 K, l n γ 1 = 0.95 x22 ; l n γ 2 = 0.95x12 ;
P1sat 2sat
= 79.80 kPa; and P = 40.50 kPa. Assuming the validity of Eq. (13.19),
13.19. A binary system of species 1 and 2 consists of vapor and liquid phases in equilib-
1sat
rium at temperature T, for which ln γ 1 = 1.8 x22 , l n γ 2 = 1.8 x12 , P = 1.24 bar, and
sat
P2 = 40.50 kPa. Assuming the validity of Eq. (13.19),
(a) For what range of values of the overall mole fraction z1 can this two-phase system
exist with a liquid mole fraction x1 = 0.65?
(b) What are the pressure P and vapor mole fraction y1 within this range?
(c) What are the pressure and composition of the azeotrope at temperature T?
13.20. For the acetone(1)/methanol(2) system, a vapor mixture for which z1 = 0.25 and
z2 = 0.75 is cooled to temperature T in the two-phase region and flows into a separa-
tion chamber at a pressure of 1 bar. If the composition of the liquid product is to be
x1 = 0.175, what is the required value of T, and what is the value of y1? For liquid
mixtures of this system, to a good approximation, ln γ 1 = 0.64 x22 and l n γ 2 = 0.64 x12 .
13.21. The following is a rule of thumb: For a binary system in VLE at low pressure, the
equilibrium vapor-phase mole fraction y1 corresponding to an equimolar liquid mix-
ture is approximately
P1sat
y 1 = _________
P1 + P2sat
sat
where Pisat is a pure-species vapor pressure. Clearly, this equation is valid if Raoult’s
law applies. Prove that it is also valid for VLE described by Eq. (13.19), with ln γ 1 =
A x22 and l n γ 2 = A x12 .
13.22. A process stream contains light species 1 and heavy species 2. A relatively pure liquid
stream containing mostly 2 is desired, obtained by a single-stage liquid/vapor sepa-
ration. Specifications of the equilibrium composition are: x1 = 0.002 and y1 = 0.950.
Use data given below to determine T (K) and P (bar) for the separator. Assume that
512 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
Eq. (13.19) applies; the calculated P should be used to validate this assumption. Data:
For the liquid phase,
ln γ 1 = 0.93 x22
ln γ 2 = 0.93 x12
B i
∕ bar = Ai − ____
ln Pisat
T ∕ K
A1 = 10.08, B1 = 2572.0, A2 = 11.63, B2 = 6254.0
13.23. If a system exhibits VLE, at least one of the K-values must be greater than 1.0 and at
least one must be less than 1.0. Offer a proof of this observation.
13.24. Flash calculations are simpler for binary systems than for the general multicomponent
case because the equilibrium compositions for a binary are independent of the overall
composition. Show that, for a binary system in VLE,
1 − K 2 K 1(1 − K 2)
x 1 = _______
y 1 = _________
K 1 − K 2 K 1 − K 2
z 1( K 1 − K 2) − (1 − K 2)
= _________________
(K 1 − 1)(1 − K 2)
13.25. The NIST Chemistry WebBook reports critically evaluated Henry’s constants for
selected chemicals in water at 25°C. Henry’s constants from this source, denoted here
by kHi, appear in the VLE equation written for the solute in the form:
mi = kHiyiP
where mi is the liquid-phase molality of solute species i, expressed as mol i∕kg solvent.
(a) Determine an algebraic relation connecting kHi to i, Henry’s constant in
Eq. (13.26). Assume that xi is “small.”
(b) The NIST Chemistry WebBook provides a value of 0.034 mol·kg−1·bar−1 for
kHi of CO2 in H2O at 25°C. What is the implied value of i in bar? Compare this
with the value given in Table 13.2, which came from a different source.
13.26. (a) A feed containing equimolar amounts of acetone(1) and acetonitrile(2) is throttled
to pressure P and temperature T. For what pressure range (atm) will two phases
(liquid and vapor) be formed for T = 50°C? Assume that Raoult’s law applies.
(b) A feed containing equimolar amounts of acetone(1) and acetonitrile(2) is throttled
to pressure P and temperature T. For what temperature range (°C) will two phases
(liquid and vapor) be formed for P = 0.5(atm)? Assume that Raoult’s law applies.
13.27. A binary mixture of benzene(1) and toluene(2) is flashed to 75 kPa and 90°C. Analy-
sis of the effluent liquid and vapor streams from the separator yields: x1 = 0.1604 and
y1 = 0.2919. An operator remarks that the product streams are “off-spec,” and you are
asked to diagnose the problem.
(a) Verify that the exiting streams are not in binary equilibrium.
(b) Verify that an air leak into the separator could be the cause.
13.10. Problems 513
13.28. Ten (10) kmol·hr−1 of hydrogen sulfide gas is burned with the stoichiometric amount
of pure oxygen in a special unit. Reactants enter as gases at 25°C and 1(atm). Products
leave as two streams in equilibrium at 70°C and 1(atm): a phase of pure liquid water,
and a saturated vapor stream containing H2O and SO2.
(a) What is the composition (mole fractions) of the product vapor stream?
(b) What are the rates (kmol·hr−1) of the two product streams?
13.29. Physiological studies show the neutral comfort level (NCL) of moist air corresponds
to an absolute humidity of about 0.01 kg H2O per kg of dry air.
13.30. An industrial dehumidifier accepts 50 kmol·hr−1 of moist air with a dewpoint of 20°C.
Conditioned air leaving the dehumidifier has a dewpoint temperature of 10°C. At
what rate (kg·hr−1) is liquid water removed in this steady-flow process? Assume P is
constant at 1(atm).
13.32. The following is a set of VLE data for the system methanol(1)/water(2) at 333.15 K:
P/kPa x1 y1 P/kPa x1 y1
19.953 0.0000 0.0000 60.614 0.5282 0.8085
39.223 0.1686 0.5714 63.998 0.6044 0.8383
42.984 0.2167 0.6268 67.924 0.6804 0.8733
48.852 0.3039 0.6943 70.229 0.7255 0.8922
52.784 0.3681 0.7345 72.832 0.7776 0.9141
56.652 0.4461 0.7742 84.562 1.0000 1.0000
Extracted from K. Kurihara et al., J. Chem. Eng. Data, vol. 40, pp. 679–684, 1995.
(a) Basing calculations on Eq. (13.24), find parameter values for the Margules equa-
tion that provide the best fit of GE∕RT to the data, and prepare a Pxy diagram that
compares the experimental points with curves determined from the correlation.
(b) Repeat (a) for the van Laar equation.
(c) Repeat (a) for the Wilson equation.
514 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
(d) Using Barker’s method, find parameter values for the Margules equation that pro-
vide the best fit of the P–x1 data. Prepare a diagram showing the residuals δP and
δy1 plotted vs. x1.
(e) Repeat (d) for the van Laar equation.
(f) Repeat (d) for the Wilson equation.
13.33. If Eq. (13.24) is valid for isothermal VLE in a binary system, show that:
13.34. The following is a set of VLE data for the system acetone(1)/methanol(2) at 55°C:
P/kPa x1 y1 P/kPa x1 y1
68.728 0.0000 0.0000 97.646 0.5052 0.5844
72.278 0.0287 0.0647 98.462 0.5432 0.6174
75.279 0.0570 0.1295 99.811 0.6332 0.6772
77.524 0.0858 0.1848 99.950 0.6605 0.6926
78.951 0.1046 0.2190 100.278 0.6945 0.7124
82.528 0.1452 0.2694 100.467 0.7327 0.7383
86.762 0.2173 0.3633 100.999 0.7752 0.7729
90.088 0.2787 0.4184 101.059 0.7922 0.7876
93.206 0.3579 0.4779 99.877 0.9080 0.8959
95.017 0.4050 0.5135 99.799 0.9448 0.9336
96.365 0.4480 0.5512 96.885 1.0000 1.0000
Extracted from D. C. Freshwater and K. A. Pike, J. Chem. Eng. Data, vol. 12,
pp. 179–183, 1967.
(a) Basing calculations on Eq. (13.24), find parameter values for the Margules equa-
tion that provide the best fit of GE∕RT to the data, and prepare a Pxy diagram that
compares the experimental points with curves determined from the correlation.
(b) Repeat (a) for the van Laar equation.
(c) Repeat (a) for the Wilson equation.
(d) Using Barker’s method, find parameter values for the Margules equation that pro-
vide the best fit of the P–x1 data. Prepare a diagram showing the residuals δP and
δy1 plotted vs. x1.
(e) Repeat (d) for the van Laar equation.
(f) Repeat (d) for the Wilson equation.
13.35. The excess Gibbs energy for binary systems consisting of liquids not too dissimilar in
chemical nature is represented to a reasonable approximation by the equation:
G E ∕ RT = A x 1 x 2
where A is a function of temperature only. For such systems, it is often observed
that the ratio of the vapor pressures of the pure species is nearly constant over a
13.10. Problems 515
considerable temperature range. Let this ratio be r, and determine the range of values
of A, expressed as a function of r, for which no azeotrope can exist. Assume the vapor
phase to be an ideal gas.
13.36. For the ethanol(1)/chloroform(2) system at 50°C, the activity coefficients show inte-
rior extrema with respect to composition [see Fig. 13.4(e)].
(a) Prove that the van Laar equation cannot represent such behavior.
(b) The two-parameter Margules equation can represent this behavior, but only for
particular ranges of the ratio A21∕A12. What are they?
13.37. VLE data for methyl tert-butyl ether(1)/dichloromethane(2) at 308.15 K are as follows:
P/kPa x1 y1 P/kPa x1 y1
85.265 0.0000 0.0000 59.651 0.5036 0.3686
83.402 0.0330 0.0141 56.833 0.5749 0.4564
82.202 0.0579 0.0253 53.689 0.6736 0.5882
80.481 0.0924 0.0416 51.620 0.7676 0.7176
76.719 0.1665 0.0804 50.455 0.8476 0.8238
72.422 0.2482 0.1314 49.926 0.9093 0.9002
68.005 0.3322 0.1975 49.720 0.9529 0.9502
65.096 0.3880 0.2457 49.624 1.0000 1.0000
The data are well correlated by the three-parameter Margules equation [an extension
of Eq. (13.39)]:
G E
___
= (A 21 x 1 + A 12 x 2 − C x 1 x 2)x 1 x 2
RT
Implied by this equation are the expressions:
(a) Basing calculations on Eq. (13.24), find the values of parameters A12, A21, and C
that provide the best fit of GE∕RT to the data.
(b) Prepare a plot of ln γ1, ln γ2, and GE∕(x1x2RT ) vs. x1 showing both the correlation
and experimental values.
(c) Prepare a Pxy diagram [see Fig. 13.8(a)] that compares the experimental data with
the correlation determined in (a).
(d) Prepare a consistency-test diagram like Fig. 13.9.
(e) Using Barker’s method, find the values of parameters A12, A21, and C that provide
the best fit of the P–x1 data. Prepare a diagram showing the residuals δP and δy1
plotted vs. x1.
516 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
13.38. Equations analogous to Eqs. (10.15) and (10.16) apply for excess properties. Because
ln γi is a partial property with respect to GE∕RT, these analogous equations can be
written for ln γ1 and ln γ2 in a binary system.
(a) Write these equations, and apply them to Eq. (13.42) to show that Eqs. (13.43) and
(13.44) are indeed obtained.
(b) The alternative procedure is to apply Eq. (13.7). Show that by doing so
Eqs. (13.43) and (13.44) are again reproduced.
13.39. The following is a set of activity-coefficient data for a binary liquid system as deter-
mined from VLE data:
x1 γ1 γ2 x1 γ1 γ2
0.0523 1.202 1.002 0.5637 1.120 1.102
0.1299 1.307 1.004 0.6469 1.076 1.170
0.2233 1.295 1.006 0.7832 1.032 1.298
0.2764 1.228 1.024 0.8576 1.016 1.393
0.3482 1.234 1.022 0.9388 1.001 1.600
0.4187 1.180 1.049 0.9813 1.003 1.404
0.5001 1.129 1.092
Inspection of these experimental values suggests that they are noisy, but the question
is whether they are consistent, and therefore possibly on average correct.
(a) Find experimental values for GE∕RT and plot them along with the experimental
values of ln γ1 and ln γ2 on a single graph.
(b) Develop a valid correlation for the composition dependence of GE∕RT and show
lines on the graph of part (a) that represent this correlation for all three of the
quantities plotted there.
(c) Apply the consistency test described in Ex. 13.4 to these data, and draw a conclu-
sion with respect to this test.
13.40. Following are VLE data for the system acetonitrile(1)/benzene(2) at 45°C:
P/kPa x1 y1 P/kPa x1 y1
29.819 0.0000 0.0000 36.978 0.5458 0.5098
31.957 0.0455 0.1056 36.778 0.5946 0.5375
33.553 0.0940 0.1818 35.792 0.7206 0.6157
35.285 0.1829 0.2783 34.372 0.8145 0.6913
36.457 0.2909 0.3607 32.331 0.8972 0.7869
36.996 0.3980 0.4274 30.038 0.9573 0.8916
37.068 0.5069 0.4885 27.778 1.0000 1.0000
Extracted from I. Brown and F. Smith, Austral. J. Chem., vol. 8, p. 62, 1955.
13.10. Problems 517
The data are well correlated by the three-parameter Margules equation (see Prob. 13.37).
(a) Basing calculations on Eq. (13.24), find the values of parameters A12, A21, and C
that provide the best fit of GE∕RT to the data.
(b) Prepare a plot of ln γ1, ln γ2, and GE∕x1x2 RT vs. x1 showing both the correlation
and experimental values.
(c) Prepare a Pxy diagram [see Fig. 13.8(a)] that compares the experimental data with
the correlation determined in (a).
(d) Prepare a consistency-test diagram like Fig. 13.9.
(e) Using Barker’s method, find the values of parameters A12, A21, and C that provide
the best fit of the P–x1 data. Prepare a diagram showing the residuals δP and δy1
plotted vs. x1.
13.41. An unusual type of low-pressure VLE behavior is that of double azeotropy, in which
the dew and bubble curves are S-shaped, thus yielding at different compositions both
a minimum-pressure and a maximum-pressure azeotrope. Assuming that Eq. (13.57)
applies, determine under what circumstances double azeotropy is likely to occur.
13.42. Rationalize the following rule of thumb, appropriate for an equimolar binary liquid mixture:
G E
___ 1
(equimolar) ≈ __
ln (γ1∞ γ2∞ )
RT 8
Problems 13.43 through 13.54 require parameter values for the Wilson or NRTL equation
for liquid-phase activity coefficients. Table 13.10 gives parameter values for both equations.
Antoine equations for vapor pressure are given in Table B.2, Appendix B.
13.43. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the
Wilson equation, prepare a Pxy diagram for t = 60°C.
13.44. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the
Wilson equation, prepare a txy diagram for P = 101.33 kPa.
13.45. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the
NRTL equation, prepare a Pxy diagram for t = 60°C.
13.46. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the
NRTL equation, prepare a txy diagram for P = 101.33 kPa.
13.47. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the Wil-
son equation, make the following calculations:
(a) BUBL P : t = 60° C, x 1 = 0.3.
(b) DEW P : t = 60° C, y 1 = 0.3.
(c) P, T − flash: t = 60° C, P = _12 ( P bubble + P dew), z 1 = 0.3.
(d) If an azeotrope exists at t = 60° C, find P az and x1az = y1az .
Table 13.10: Parameter Values for the Wilson and NRTL Equations
Parameters a12, a21, b12, and b21 have units of cal·mol−1, and V1 and V2 have units of cm3·mol−1.
13.49. For one of the binary systems listed in Table 13.10, based on Eq. (13.19) and the
Wilson equation, make the following calculations:
(a)
BUBL T : P = 101.33 kPa, x 1 = 0.3.
(b)
DEW T : P = 101.33 kPa, y 1 = 0.3.
P, T − flash : P = 101.33 kPa, T = _12 ( T bubble + T dew), z 1 = 0.3.
(c)
If an azeotrope exists at P = 101.33 kPa, find T az and x1az = y1az .
(d)
13.51. For the acetone(1)/methanol(2)/water(3) system, based on Eq. (13.19) and the Wilson
equation, make the following calculations:
(a) BUBL P : t = 65° C, x 1 = 0.3, x 2 = 0.4.
(b) DEW P : t = 65° C, y 1 = 0.3, y 2 = 0.4.
(c) P, T − flash : t = 65° C, P = _12 (P bubble + P dew), z 1 = 0.3, z 2 = 0.4.
13.10. Problems 519
13.53. For the acetone(1)/methanol(2)/water(3) system, based on Eq. (13.19) and the Wilson
equation, make the following calculations:
(a) BUBL T : P = 101.33 kPa, x 1 = 0.3, x 2 = 0.4.
(b)
DEW T : P = 101.33 kPa, y 1 = 0.3, y 2 = 0.4.
(c) P, T − flash : P = 101.33 kPa, T = _ 12 ( T bubble + T dew), z 1 = 0.3, z 2 = 0.2.
13.55. The following expressions have been reported for the activity coefficients of species
1 and 2 in a binary liquid mixture at given T and P:
ln γ 1 = x22 (0.273 + 0.096 x 1) ln γ 2 = x12 (0 . 273 − 0.096 x 2)
13.56. Possible correlating equations for ln γ1 in a binary liquid system are given here.
For one of these cases, determine by integration of the Gibbs/Duhem equation
[Eq. (13.11)] the corresponding equation for ln γ2. What is the corresponding equation
for GE∕RT? Note that by its definition, γi = 1 for xi = 1.
13.57. A storage tank contains a heavy organic liquid. Chemical analysis shows the liquid to
contain 600 ppm (molar basis) of water. It is proposed to reduce the water concentra-
tion to 50 ppm by boiling the contents of the tank at constant atmospheric pressure.
Because the water is lighter than the organic, the vapor will be rich in water; contin-
uous removal of the vapor serves to reduce the water content of the system. Estimate
the percentage loss of organic (molar basis) in the boil-off process. Comment on the
reasonableness of the proposal.
13.58. Binary VLE data are commonly measured at constant T or at constant P. Isothermal data
are much preferred for determination of a correlation for GE for the liquid phase. Why?
520 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
13.60. A breathalyzer measures volume-% ethanol in gases exhaled from the lungs. Calibra-
tion relates it to volume-% ethanol in the bloodstream. Use VLE concepts to develop
an approximate relation between the two quantities. Numerous assumptions are
required; state and justify them where possible.
13.61. Table 13.10 gives values of parameters for the Wilson equation for the acetone(1)/
methanol(2) system. Estimate values of ln γ1∞ and l nγ2∞ at 50°C. Compare with the
values suggested by Fig. 13.4(b). Repeat the exercise with the NRTL equation.
13.62. For a binary system derive the expression for HE implied by the Wilson equation for
PE is necessarily positive. Recall
GE∕RT. Show that the implied excess heat capacity C
that the Wilson parameters depend on T, in accord with Eq. (13.53).
13.63. A single P-x1-y1 data point is available for a binary system at 25°C. Estimate from
the data:
(a) The total pressure and vapor-phase composition at 25°C for an equimolar liquid
mixture.
(b) Whether azeotropy is likely at 25°C.
1sat
Data: At 25°C, P = 183.4 and P2sat
= 96.7 kPa
For x1 = 0.253, y1 = 0.456 and P = 139.1 kPa
13.64. A single P–x1 data point is available for a binary system at 35°C. Estimate from
the data:
13.65. The excess Gibbs energy for the system chloroform(1)/ethanol(2) at 55°C is well rep-
resented by the Margules equation, GE∕RT = (1.42 x1 + 0.59 x2)x1x2. The vapor pres-
sures of chloroform and ethanol at 55°C are P1sat
= 82.37 and P2sat
= 37.31 kPa.
13.10. Problems 521
(a) Assuming the validity of Eq. (13.19), make BUBL P calculations at 55°C for
liquid-phase mole fractions of 0.25, 0.50, and 0.75.
(b) For comparison, repeat the calculations using Eqs. (13.13) and (13.14) with virial
coefficients: B11 = −963, B22 = −1523, and B12 = 52 cm3·mol−1.
ˆ 1 and ϕˆ 2for a binary gas mixture described by Eq. (3.38). The
13.66. Find expressions for ϕ
mixing rule for B is given by Eq. (10.62). The mixing rule for C is given by the general
equation:
C = ∑ ∑ ∑ y i y j y k C ijk
i j k
where Cs with the same subscripts, regardless of order, are equal. For a binary mix-
ture, this becomes:
13.67. A system formed of methane(1) and a light oil(2) at 200 K and 30 bar consists of
a vapor phase containing 95 mol-% methane and a liquid phase containing oil and
dissolved methane. The fugacity of the methane is given by Henry’s law, and at the
temperature of interest Henry’s constant is 1 = 200 bar. Stating any assumptions,
estimate the equilibrium mole fraction of methane in the liquid phase. The second
virial coefficient of pure methane at 200 K is −105 cm3·mol−1.
13.68. Use Eq. (13.13) to reduce one of the following isothermal data sets, and compare the
result with that obtained by application of Eq. (13.19). Recall that reduction means
developing a numerical expression for GE∕RT as a function of composition.
13.69. For one of the following substances, determine Psat∕bar from the Redlich/Kwong
equation at two temperatures: T = Tn (the normal boiling point), and T = 0.85Tc. For
the second temperature, compare your result with a value from the literature (e.g.,
Perry’s Chemical Engineers’ Handbook). Discuss your results.
(a) Acetylene; (b) Argon; (c) Benzene; (d) n-Butane; (e) Carbon monoxide;
(f) n-Decane; (g) Ethylene; (h) n-Heptane; (i) Methane; (j) Nitrogen
522 CHAPTER 13. Thermodynamic Formulations for Vapor/Liquid Equilibrium
13.70. Work Prob. 13.69 for one of the following: (a) The Soave/Redlich/Kwong equation;
(b) the Peng/Robinson equation.
13.71. Departures from Raoult’s law are primarily from liquid-phase nonidealities (γi ≠ 1).
But vapor-phase nonidealities (ϕ ˆ i≠ 1) also contribute. Consider the special case
where the liquid phase is an ideal solution, and the vapor phase a nonideal gas mixture
described by Eq. (3.36). Show that departures from Raoult’s law at constant tempera-
ture are likely to be negative. State clearly any assumptions and approximations.
13.72. Determine a numerical value for the acentric factor ω implied by:
(a) The van der Waals equation;
(b) The Redlich/Kwong equation.
13.73. The relative volatility α12 is commonly used in applications involving binary VLE. In
particular (see Ex. 13.1), it serves as a basis for assessing the possibility of binary azeot-
ropy. (a) Develop an expression for α12 based on Eqs. (13.13) and (13.14). (b) Specialize
the expression to the composition limits x1 = y1 = 0 and x1 =y1 = 1. Compare with the
result obtained from modified Raoult’s law, Eq. (13.19). The difference between the
results reflects the effects of vapor-phase nonidealities. (c) Further specialize the results
of part (b) to the case where the vapor phase is an ideal solution of real gases.
13.74. Although isothermal VLE data are preferred for extraction of activity coefficients, a
large body of good isobaric data exists in the literature. For a binary isobaric T-x1-y1
data set, one can extract point values of γi via Eq. (13.13):
y i Φ i( T k, P, y)P
γ i(x, T k) = ____________
x i Pisat
( T k)
Here, the variable list for γi recognizes a primary dependence on x and T; pressure
dependence is normally negligible. The notation Tk emphasizes that temperature var-
ies with data point across the composition range, and the calculated activity coeffi-
cients are at different temperatures. However, the usual goal of VLE data reduction
and correlation is to develop an appropriate expression for GE∕RT at a single temper-
ature T. A procedure is needed to correct each activity coefficient to such a T chosen
near the average for the data set. If a correlation for HE(x) is available at or near this T,
show that the values of γi corrected to T can be estimated by the expression:
− H¯ E ___
RT ( T k )
γ i(x, T ) = γ i(x, T k) exp[_____ − 1 ]
T
i
13.75. What are the relative contributions of the various terms in the gamma/phi expression
for VLE? One way to address the question is through calculation of the activity coef-
ficients for a single binary VLE data point via Eq. (13.19):
⏟
sat sat
x i P i ϕ fi
⏟ ⏟
i
(A) (C)
(B)
13.10. Problems 523
Term (A) is the value that would follow from modified Raoult’s law; term (B) accounts
for vapor-phase nonidealities; term (C) is the Poynting factor [see Eq. (10.44)]. Use
the following single-point data for the butanenitrile(1)/benzene(2) system at 318.15 K
to evaluate all terms for i = 1 and i = 2. Discuss the results.
VLE data: P = 0.20941 bar, x1 = 0.4819, y1 = 0.1813.
Ancillary data: P1sat
= 0 . 07287 and P2sat
= 0.29871 bar
B
11 = −7993, B22 = −1247, B cm3·mol−1
12 = −2089
V1l = 90, V2l = 92 cm3·mol−1
13.76. Generate P-x1-y1 diagrams at 100°C for one of the systems identified below. Base
activity coefficients on the Wilson equation, Eqs. (13.45) to (13.47). Use two p rocedures:
(i) modified Raoult’s law, Eq. (13.19), and (ii) the gamma/phi approach, Eq. (13.13),
with Φi given by Eq. (13.14). Plot the results for both procedures on the same graph.
Compare and discuss them.
isat
Data sources: For P use Table B.2. For vapor-phase nonidealities, use material from
Chap. 3; assume that the vapor phase is an (approximately) ideal solution. Estimated
parameters for the Wilson equation are given for each system.