Chapter 6.

Equilibrium Based Separation Processes

6.1 Study objectives

Aim: The expected learning outcomes include:
• Knowledge and understanding of the ideal gas, the Gibbs phase equilibrium, the phase
rule, Duhem’s theorem, the ideal solution and the origin of deviations from ideal behaviour;
• The ability to construct suitable P-T; P-x-y; T-x-y; and x-y diagrams to illustrate the
qualitative behaviour of binary vapour-liquid equilibrium behaviour
• Understanding of the physical basis of equilibrium-based separation processes, and
• The ability to apply this knowledge to basic thermodynamic-based separation processes.

At the end of this syllabus theme a student will be able to:

• Calculate partition coefficients and construct y-x, T-y-x and P-y-x equilibrium diagrams for
vapour liquid equilibrium from first principles.
• State and explain Raoult’s and Henry’s laws, the γ-Φ equilibrium formulation, and the
concepts: relative volatility azeotrope and eutectic
• Solve flash and liquid-liquid separation problems
• Explain the origin of deviations from ideal gas and ideal solution behaviour.
• Screen a system to determine whether an azeotrope exists and to calculate the effect of
pressure and temperature on the azeotrope composition
• Perform steam distillation-type calculations
• Perform multi-component flash calculations using the K-value method
• Construct liquid-liquid and solid-solid-liquid phase diagrams.
• Perform binary liquid-liquid and ternary liquid-liquid extraction calculations
• Select (and justify the selection) of appropriate solvents to effect a desired liquid-liquid
extraction

6.2 The Ideal Gas and the Ideal Solution

It is convenient to introduce the concept of an ideal gas, i.e. a hypothetical fluid that obeys the
following equation of state:
PV = nRT (6-1)

An ideal gas consists of molecules that occupy zero volume and that do not interact. Each
chemical species in an ideal gas mixture therefore contributes its own properties independent of

1
the presence of the other species. When several different gasses occupy the same container each
gas contributes a partial pressure equal to the pressure it would exert were it present on its own:
pi = ni R T / V (6-2)
According to Avogadro’s hypothesis, equal volumes of gases at the same pressure and
temperature contain equal numbers of molecules. Dalton’s law states that the total pressure of a
mixture of ideal gases is given by the sum of the partial pressures of the various component gases
present:
ni RT RT nRT
P = ∑ pi = ∑ V
=
V
∑n i =
V
(6-3)

or
P = n RT V (6-4)
Combining equations (6-2) and (6-4) gives Dalton’s law of partial pressures:
pi = y i P (6-5)
An ideal solution, in the context of VLE, is one that obeys Raoult's Law: the partial
pressure of a pure component in a vapour mixture is equal to its mole fraction in the liquid times
its vapour pressure at the system temperature.
pi = xi Pi * (6-6)
To correct for non-ideal solution behaviour, the activity coefficient correction factor is
introduced:
pi = γ i xi Pi * (6-7)
Intuitively, phase equilibrium implies that the partial pressures exerted by each component in the
gas and solution must be the same, i.e.
yi P = γ i xi Pi * (6-8)
To account for real gas behaviour, this equation needs to be modified even further. At low
pressures it is sufficient to introduce the fugacity coefficient to correct for gas phase deviations
from ideal behaviour:
Φ i yi P = γ i xi Pi * (6-9)
The distribution coefficient or equilibrium constant is, for medium to low pressures, is then
approximately given by:
yi γ i Pi *
Ki = = (6-10)
xi Φi P
When a component is supercritical (i.e. liquid does not exist at the prevailing conditions),
we can use Henry’s Law instead:

2
 yi γ *i k i
Ki = = (6-11)
xi Φi P
At low pressures the vapour phase is often ideal but liquid solutions are rarely ideal. When
both the vapour and liquid phases are ideal, equation (6-7) simplifies to:
yi Pi *
Ki = = (6-12)
xi P

When two liquid phases, α and β, are in equilibrium, then

xiα γ iβ
Ki = = (6-13)
xiβ γ iα
Equilibrium separation processes take advantage of the fact that, in general, Ki ≠1.

1
α =8
α =4
0.8 α =2

0.6
y1
0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
x1

Figure 6-1: Isothermal binary VLE diagram for ideal systems with constant relative volatility α.

6.3 Binary distillation and binary flash (Ideal mixtures)

Vapour pressure data are often correlated using the empirical Antoine correlation:
B
n P * = A − (6-14)
C +T
Here P* is the vapour pressure of a condensed phase, i.e. either liquid or solid, at temperature T
(usually expressed in °C!).
The relative volatility is a useful measure of the separability of two components in a
distillation column. It is defined as:
y i xi γ i Pi *
α ij ≡ ≈ (6-15)
yj xj γ j Pj*

3
It is a useful measure of separability of two components in a distillation column. If α12 is equal to
unity, an azeotrope exists and no separation of components 1 and 2 can be effected. For an ideal
binary system α12 = P1* P2* . The Pi* depend on temperature only. Thus, under isothermal
conditions, the relative volatility takes on a constant value, i.e. it does not change with
composition! Neither azeotropes nor multiple liquid phases can exist in ideal mixtures. The
assumption of constant relative volatility significantly simplifies distillation design calculations.
The shape of the VLE curve can be predicted knowing only the vapour pressures together with the
equilibrium condition, equation (6-12) and the equation for the total system pressure:
P = γ 1 x1P1* + γ 1 x2 P2* (6-16)
Consider 1 mole of a binary mixture with constant α12 = α . The shape of the vapour-
liquid equilibrium (VLE) curve is shown in Figure 6-1 and it is described by:
α x1
y1 = (6-17)
1 + (α −1) x1

1.0

0.8 L/V → ∞

L/V = 1
0.6
y
L/V = 0
0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0

Figure 6-2: The basic principle of binary separation by either flashing a saturated liquid or partially
condensing a vapour.
Assume that the mixture has an overall composition of z1 and that it splits into V moles of vapour
and L moles of liquid. The composition of the two phases can be calculated from an equilibrium
relationship, e.g. equation (6-13) and the following mass balances:
4
 V + L = 1 and y1V + x1L = z1 (6-18)
1 L
This leads to: y1 =   z1 −   x1 (6-19)
V  V 
The last equation describes a straight line with slope –L/V that intersects the y1 = x1
diagonal at the point z1. The equilibrium composition corresponds to the intersection of this straight
line with the equilibrium curve as illustrated in Figure 6-2. Figure 6-3 illustrates typical near-ideal
systems.

6.4 Origin of non-ideal solution behaviour (General Low Pressure Systems).

Even if the molecules behave ideally in the liquid phase, increasing pressure creates non-ideal
behaviour in the gas phase. In practice this is taken into account using fugacity coefficients that
are calculated from equations of state, e.g. the Redlich-Kwong equation. However, in the systems
usually encountered, the non-ideal behaviour of the liquid phase grossly outweighs that of the
vapour phase.
Non-ideal behaviour in the condensed state is caused by differences in molecular size and
shape as well as differences in the strength of interactions between the different functional groups
on the interacting molecules. As a rule of thumb, intermolecular force differences are usually most
important, size differences are next, and shape differences are usually least important. The "activity
coefficient" is used as a multiplier on the vapour pressure to correct for these effects. The activity
coefficient is not a "fudge factor", but a true thermodynamic quantity related to the excess Gibbs
free energy GE:
G E = RT Σ x i n γ i (6-20)
A common standard state assumed here for calculating activity coefficients is the pure
liquid compound at the system temperature and pressure.
The vapour pressure and the boiling point of a solution are measures of the attractive forces
between the molecules present in the liquid. If a new component is added to the solution, these
forces will change. Occasionally, the forces attracting unlike pairs of molecules are stronger than
forces between like molecules. In most cases however, positive deviations are the norm.
Non-ideal behaviour of the liquid phase is usually taken into account using an appropriate
activity coefficient correlation. The simplest of these is the Hildebrandt-Van Laar or Porter regular
solution model. Below it is used to illustrate the construction of an isothermal binary phase
diagram.

5
 Isobaric Isothermal

100 450
Methanol (1) - 1-propanol (2) acetone (1) - ethanol (2)
400
90 1 atm liquid
350
2φ vapour

P, mmHg
T, oC

300
80 2φ
250

70 liquid 200 vapour

150 T = 40oC
60
0.0 x1 0.5 y1 1.0 100
0 0.2 0.4 0.6 0.8 1
x1
x1, y1

Figure 6-3: Typical near-ideal, binary VLE phase diagram

1
cyclohexane (1)

0.8

0.6
Vapour mole
fraction, y1
0.4
Component (2):
0.2 (a) heptane
(b) 2-propanol

0
0 0.2 0.4 0.6 0.8 1
Liquid mole fraction, x1

Figure 6-4: Effect of positive deviations from ideal behaviour on a binary VLE diagram. Curve
(a) indicates Raoult's law for a mixture with a constant relative volatility. Curve (b) indicates fairly
large activity coefficients, resulting in the formation of an azeotrope. Curve (c) has activity
coefficients so large that two liquid phases have formed.

Here is a procedure for calculating the equilibrium curve in the y-x diagram: Choose an
appropriate activity coefficient correlation. The Porter model is the simplest and it reads:

n γ 1 = A (1− x1 ) 2 and n γ 2 = A x1
2
(i)

Write the total pressure equation: P = x1γ 1 P1* + x2γ 2 P2* (ii)

6
Equilibrium relationship: y1 = x1γ 1 P1* / P (iii)
Procedure:
1. From the given temperature calculate P1* , P2*
2. Choose a value for x1 and calculate γ 1 and γ 2 from equation (i)
3. Determine the system pressure from equation (ii) and calculate y1 from equation (iii)

6.5 Azeotropes
Consider the isothermal behaviour of a hypothetical binary system composed of components (1)
and (2) with vapour pressures P1* and P2* . Figure 6-4 graphically examines the behaviour of such
a system as the activity coefficients are systematically increased.
It is conventional for the more volatile component to be on the right hand side. Thus the
right hand side of the diagram represents the situation in the rectification or top part of a distillation
column. Similarly, the left-hand side represents the stripping or bottom part of the column. Figure
6-4 shows that, as the activity coefficient is increased, stripping becomes easier but rectification
becomes more difficult or even impossible.
As the deviations are increased a maximum appears in the total pressure curve. This is the
condition that applies at an azeotrope. Thus, the existence of an azeotrope (maximum) depends on
the difference in volatility (vapour pressures) and the magnitude of deviations from ideal solution
behaviour. When the volatility difference is small, only small deviations are needed to produce a
maximum and thus form an azeotrope, while large volatility differences require large deviations.
Similar conclusions are obtained if negative, rather than positive, deviations are considered. In
that case one would be looking for a minimum in the total pressure curve instead of a maximum.
A graphical representation of these phenomena for a binary system is shown in Figure 6-5.
Examination of this Figure leads to the following conclusions:
• A system that exhibits a minimum on a pressure composition diagram shows a maximum on a
temperature composition diagram under similar conditions
• A system that exhibits a maximum on a pressure composition diagram shows a minimum on a
temperature composition diagram under similar conditions.
• The azeotrope point corresponds to the extreme value in either plot where the composition of
the vapour and liquid phases are identical, i.e. the liquid and vapour curves touch at the
azeotropic point.

7
 (a) (b)

2000 190
methanol (1) - CCl4 (2) acetone (1) - chloroform (2)
1800 180

170 20oC
1600

P, mmHg
P, mmHg

160
1400
150
1200 80oC 140

1000 130

800
120
0 0.2 0.4 0.6 0.8 1 0 0.5 1

x1, y1 x1, y1

22
methanol (1) - CCl4 (2) acetone (1) - chloroform (2)
105
20
P = 1800 mmHg

18
T, oC

95
T, oC

16
85
14 P = 130 mmHg

75
12
0 0.2 0.4 0.6 0.8 1 0 0.5 1

x 1, y 1 x1, y1

1.00
1
methanol (1) - CCl4 (2) acetone (1)-chloroform (2)
0.80
0.8

0.60
0.6
y1 y1
0.40
0.4
80oC T = 20oC
0.20 0.2

0.00 0
0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
x1
x1

Figure 6-5: Equilibrium – composition diagrams for azeotropic mixtures caused by (a) positive
and (b) negative deviations from ideal behaviour.
8
The significant feature of azeotropic mixtures is that their composition will not change upon
further boiling since the vapour leaving has the same composition as the liquid. Notice that for the
azeotropic composition α12 = 1. This azeotropic point separates the composition range into two
parts, one for which α12 < 1 and another for which α12 > 1. This means that the vapour on one side
of the azeotropic composition will be richer than the liquid in the more volatile component, and
that the opposite will be true on the other side.
Virtually all azeotropes are minimum-boiling. That is, the boiling point of the mixture is
lower than either pure component. This is a consequence of almost all systems exhibiting positive
deviations from ideal behaviour. As mentioned, such minimum-boiling azeotropes are usually
encountered at the top of distillation columns.
Negative deviations require attracting forces between unlike pairs of molecules that are
stronger than forces between like molecules. In these, less common cases, the excess Gibbs free
energy is negative and the activity coefficients will be less than unity. This negative deviation from
ideal behaviour is usually observed when the two species can form chemical complexes, e.g. as in
the water-ammonia and the formic acid-water systems.
Most azeotropes encountered in practice are therefore a natural consequence of positive deviations
from ideal behaviour (increasing activity coefficients). The greater the aggregation among like
molecules, the more positive the value of the excess Gibbs free energy and the higher the activity
coefficient. The higher the activity coefficient for a component, the greater the partial pressure it
will exhibit. However, the partial pressure cannot exceed the vapour pressure under any
circumstances.
Azeotropic data are simply a single vapour-liquid equilibrium data point. Since yi = xi at
the azeotrope, and since the temperature or pressure is given, by computation of the vapour
pressures the activity coefficients are calculated approximately by:
γ 1 = P P1* (6-21)
This provides the activity coefficients for both components at one composition and these can be
used to determine two parameters in an activity coefficient equation such as Van Laar, Margules,
Wilson, etc. Thus a single azeotropic composition at a specified temperature or pressure is the very
minimum amount of data required to model a system.
In systems that do not form azeotropes, it is possible by distillation to separate any mixture
into its pure components. However, when an azeotrope exists, a mixture can only be separated into
a pure component and the azeotrope. The presence of an azeotrope therefore causes difficulty when
a separation of components by distillation is desired. When faced with the unwelcome prospect
of an azeotrope, in a system that we need to fractionate, it is of interest to know whether changing
9
the operating pressure can shift the azeotrope point. With the availability of sufficient data, simply
calculating the T-x-y diagram at various pressures easily checks this. Unfortunately such data is
rarely available.
At the known azeotropic point: P = γ 1P1* = γ 2 P2* (6-22)

P1* γ
Eliminating the system pressure P: = 2 (6-23)
P2*
γ1
The left-hand side is a function of temperature only via the relevant Antoine equations. The right-
hand side depends on composition and temperature. If we assume the validity of the Porter regular
solution model:

Ω Ω
n γ 1 = (1− x1 ) 2 and n γ 2 = x1
2

RT RT
(6-24)
And, furthermore, that in the neighbourhood of the azeotropic point, the vapour pressures can be
approximated by:
n P1* =
A1 − B1 T and n P2* =
A2 − B2 T (6-25)
It can be shown that the shift in the azeotrope point with temperature (and by implication pressure),
is given by:
=
x1,azeo aTazeo + b (6-26)

where a and b are constants.

There are certain situations in which the occurrence of an azeotrope can be used to
advantage in a separation process. The ethanol-water system forms an azeotrope at 78.2°C and 1
atm. Oxygenates, mainly MBTE but also major amounts of anhydrous ethanol, are widely used to
improve the clean burning behaviour of gasoline. The challenge is the removal of the ≈5% of water
in the azeotrope. In practice this is achieved by adding pentane, which forms an even lower
temperature azeotrope with water (34.6°C @ 1 atm.). Being lower-boiling than the ethanol - water
azeotrope, it effectively removes water from the mixture. Because the liquid pentane is immiscible
with liquid water, it can simply be decanted and recycled to the column. Thus, one azeotrope is
used to break another! Another example is purification of acetic acid contaminated with by-product
formic acid and water. This ternary features an azeotrope at 107°C. This can azeotrope can be
broken when benzene is used as entrainer. It carries the formic acid and water overhead at about
71°C and 1 atm.

10
6.6 Significance of infinite dilution activity coefficients.
At infinite dilution solute molecules are completely surrounded by dissimilar solvent molecules.
It is in this state that they tend to exhibit their highest degree of non-ideal behaviour. The volatility
or escaping tendency of a molecule is proportional to the product of vapour pressure and the
activity coefficient. The infinite dilution condition represents a limiting physical situation at the
top and bottom parts of distillation columns. For reliable design of such columns it is therefore
imperative that γ i∞ be known accurately. Extrapolation of activity coefficients, measured at finite
concentrations, is unreliable, especially if the Van Laar correlation is used. Fortunately it is
possible to measure γ i∞ values accurately using such techniques as ebulliometry or Gas-Liquid-
Chromatography (GLC) without the need to make actual composition measurements.
Screening for potential azeotropes. Binary and ternary azeotropes are widely tabulated but
no compilation can ever be complete. Assuming an ideal vapour phase the following criteria, can
be used to test for the existence of an azeotrope:
• For a minimum boiling azeotrope: γ 1∞ > P2o P1o > 1 γ 2∞ (6-27)
(positive deviations from Raoult)
• For a maximum boiling azeotrope: γ 1∞ < P2o P1o < 1 γ 2∞ (6-28)
(negative deviations from Raoult)
Note that the above relations assume that only one azeotrope exists. The benzene - hexafluoro-
benzene system is unique in that it features two azeotropes.

6.7 VLE with immiscible liquids

VLE calculations for immiscible systems are particularly simple and illustrate the concept of
“steam distillation”. Consider Figure 6.6:
Region I:
y2 P = x 2 γ 2 P2* but since x 2 = 1 , therefore γ 2 = 1.

P2* P2*
⇒ y2 = ; y1 = 1 − (6-29)
P P
Region II:
y1 P = x1 γ 1P1* but since x 1 = 1 , therefore γ 1 = 1.

P1* P1*
⇒ y1 = ; y2 = 1− (6-30)
P P
Azeotrope Point:

I II P2* P1*
y 1 = y 1 i.e. 1 − =
P P

11
 ⇒ P = P1* + P2* (6-31)
i.e. guess temperature until this relationship holds and then
P1*
y1azeo = (6-32)
P1* + P2*

120
water (1) - toluene (2) at 1 atm
Tb, toluene
110
Vapour

100 Tb,water
I
T, oC

vapour
90 vapour + liquid 2
II +
liquid 1
Tazeo

80
liquid 1 + liquid 2

70 y1,azeo
0 0.5 1

x1, y1

Figure 6.6: Isobaric phase diagram for the water (1) – toluene (2) immiscible liquid-liquid
system.

6.8 K-values and multi-component flash calculations.

The distribution coefficient or equilibrium constant is, for medium to low pressures, approximately
given by:
yi γ i Pi *
Ki = = (6-33)
xi Φi P
where the reference state for the liquid is the saturated liquid at the system pressure. Consider 1
mole of mixture with composition zi partitioning into L moles of liquid and V moles vapour:
V + L = 1
yi V + xi L = zi
⇒ z i = x i (1 −V ) + K i x i V = x i [1 + V (K i − 1)]

12
 zi
⇒ xi = and since Σ x i = 1 :
1 + V (K i − 1)

zi
∑1 + V (K i − 1)
= 1 (6-34)

Similarly
K i zi
yi = and since Σ y i = 1 :
1 + V (K i − 1)

K i zi
∑ 1 + V (K i −1)
= 1 (6-35)

zi
(a) Dew point calculation: L = 0; V = 1 ⇒ ∑K i
= 1 (6-36)

(b) Bubble point calculation: V = 0; L = 1 ⇒ ∑K i zi = 1 (6-37)

(c) Flash calculations:

Subtraction of Equation (6-36) from (6-37) provides a very suitable form for the application of
Newton’s method for flash calculations:
(K i − 1) z i
F ≡ ∑ 1 + V (K i − 1)
= 0 (6-38)

dF z i (K i − 1) 2
F ' (V ) =
dV
= − ∑ [1 + V (K i − 1)] 2
< 0 (6-39)

Since F’(V) < 0; F is a strictly decreasing function of V and convergence to the root, using
Newton’s root finding method is rapid and assured:
Fn
Vn + 1 = Vn − (6-40)
Fn'
A good starting point is to first evaluate the dew point and bubble point pressures. A good starting
value for V is then given by:
Pbubble − P
V0 = (6-41)
Pbubble − Pdew

3.9 Binary Liquid-liquid phase behaviour

Figure 6-7 shows the effect of temperature in binary systems that exhibit partial miscibility.
Typically the mutual solubility declines as the temperature decreases. The butanol (1) – water (2)
system shows the fairly common upper critical solution behaviour. Systems showing lower critical
solution temperature e.g. cyclohexyl-3-pyrrolidone – water, are less common. Some systems, e.g.
n-butanol (1) – water (2) exhibit both lower and upper critical solution behaviour.

13
 140 140
120 130
100 120
80 110
THF (1) - H2O (2)

T, oC
T, oC

60 100
40 90

20 80
n-butanol (1) - H2O (2)
0 70

-20 60
0 20 40 60 0 20 40 60
x1, mol % x1, mol %

Figure 6-7: Binary liquid-liquid phase separation.

1.05
n-butanol (1) - cyclohexane (2)

1.00
x2γ2

0.95

0.90
0.50 0.60 0.70 0.80 0.90
x1γ1

Figure 6-8: The Null method for determining the mutual solubility’s of two components forming
two different phases.

When the two phases are in equilibrium, the following set of equations is satisfied:
x1α γ 1α = x1β γ 1β and x2α γ 2α = x2β γ 2β (6-42)
Null proposed a useful graphical procedure for finding the compositions of the two liquid phases
in equilibrium. If x2γ 2 is plotted against x1 γ 1 with x1 as parameter, the plot loops back on itself

as illustrated in Figure 6-8 when phase separation does exist. The compositions x1α and x1β can
be found from the intersection point.

14
6.10 Liquid-liquid extraction
Figure 6-9 compares flashing with solvent extraction. Solvent extraction or liquid-liquid extraction
is a separation process based on the different distribution of components between two liquid
phases. At least three components are involved in an extraction. The feed to a solvent extraction
unit is the solution that contains the components to be separated: the solute (A) that is to be
removed and the key component (P) that will remain in the raffinate stream leaving the extractor.
The extraction solvent (S) must be selective towards the solute (A). It leaves the extractor enriched
with component (A) as the extract. A typical example involves the removal of the BTX aromatics
(benzene, toluene and the xylenes) from non-aromatic hydrocarbons in petroleum feed stocks. The
recovery by ordinary distillation is often impossible because the boiling points of many non-
aromatics are too close to those of the BTX components. The desired fractionation can be achieved
by extractive distillation, solvent extraction or azeotropic distillation. If the aromatic content is
high (>90%), special azeotrope formers such as acetone may be added that allow the non-aromatics
to be distil off azeotropically with the aromatics remaining as residues. In the case of extractive
distillation the added selective solvent reduces the volatility of the aromatic cut. Both the solvent
and the aromatics are recovered in the bottoms product.

Flashing (a) Solvent (b)

extraction
→ Vapour → Raffinate
Feed →
P
A+P

Feed →

→ Liquid Solvent → → Extract

S S+A

Figure 6-9: Comparing the principles of flashing and liquid-liquid extraction

The selective solvents rely on interactions of polar groups in their structure with the electron rich
and easily polarisable π-electron clouds of the aromatics. Paraffins, isoparaffins and cycloparaffins
are not able to interact in this manner and are dissolved to a lesser extent. Typical selective solvents
feature dipolar aprotic structures e.g. amide groups or dipolar protic structures e.g. alkanol amines.
The most important characteristics of a selective solvent are high selectivity towards- and high

15
capacity for the aromatics. The simplest extractive separation model requires a ternary system and
benzene and hexane are useful model compounds. The process pre-supposes the existence of a
two-phase region. Figure 6-10 shows a typical open ternary solubility diagram. The phase
equilibrium curve, the bimodal, characterizes the solubility boundaries with tie lines indicating the
coexisting phases. In Figure 6-10 the aliphatic-aromatic binary is fully miscible. It is conventional
to assume that they form near-ideal solutions. The selective solvent (S) shows miscibility gaps
with both components.
The liquid-liquid equilibrium or distribution coefficient KA defines the “capacity” or
solvent power of the solvent phase:
x AE γ AR 1
KA = R = E ≈ E (6-43)
xA γA γA
The separation factor or selectivity, SA,P, is a parameter analogous to the relative volatility in
distillation. It is defined by:
x AE x AR γ ARγ PE γ PE
S A, P = E R = E R ≈ E (6-44)
xP xP γ Aγ P γA
In order to use these parameters for quick and cost-effective characterization of candidate selective
solvents, it has become common practise to use the limiting values for infinite dilution:
1
Limiting capacity: K ∞A = (6-45)
γ A∞

γ P∞
Limiting selectivity: S ∞A ,P = (6-46)
γ A∞

An increase in selectivity implies that fewer number of separation stages will be required.
Capacity determines the ratio between extractant and feedstock. A rise in capacity diminishes the
volume of the required solvent stream and thus the size of the extraction units. Thus both selectivity
and capacity influence the investment cost of the extraction plant. However, capacity also has a
major effect on the running costs of the extraction plant. In most cases solvents with high capacity
show only low selectivity and vice versa. Since capacity and selectivity manifest themselves in
opposition, it has been suggested that the parameter ωo defined as the product of the limiting
capacity and the limiting selectivity provides a technically more realistic evaluation of a selective
solvent:

ω o = S A∞, P K A∞ = γ P∞ / γ A∞
2
(6-47)

16
 Aromatic
100.00

75.00
Raffinate or
R-phase

50.00

tie lines Extract or

25.00 E-phase

0.00
Parafinic0.00 50.00 Solvent
100.00

Figure 6-10: Schematic representation of an open ternary system suitable for solvent extraction.

Table 6-1 compares the capacity, selectivity and ωo values of some selective solvents from the
point of view of their separation properties with respect to the benzene hexane model system. Note
however, that in commercial practice an entire group of aromatics needs to be recovered.

Table 6-1: Capacity, selectivity and ωo values of different solvents related to the hexane/benzene
model system at 60°C.
Selective solvent Limiting capacity Limiting selectivity ωo value
K ∞A = 1 / γ A∞ S ∞A ,P = γ P∞ γ A∞ ω = S A∞, P K A∞
o

Ethylene glycol 0,044 13,7 0,6

Diethylene glycol 0,16 10,2 1,6
Triethylene glycol 0,23 9,5 2,2
DMSO 0,31 12,2 3,8
Sulpholane 0,42 20,0 8,4
DMF 0,58 9,2 5,3

17