C08 09/20/2010 Page 311

§8.2 General Design Considerations 311

Table 8.4 Group Interactions for Solvent Selection

Solvent

Group Solute 1 2 3 4 5 6 7 8 9

1 Acid, aromatic OH (phenol) 0     0 þ þ þ

2 Paraffinic OH (alcohol), water, imide or  0 þ þ þ þ þ þ þ
amide with active H
3 Ketone, aromatic nitrate, tertiary amine,  þ 0 þ þ  0 þ þ
pyridine, sulfone, trialkyl phosphate, or
phosphine oxide
4 Ester, aldehyde, carbonate, phosphate, nitrite or  þ þ 0 þ  þ þ þ
nitrate, amide without active H;
intramolecular bonding, e.g.,
o-nitrophenol
5 Ether, oxide, sulfide, sulfoxide, primary  þ þ þ 0  0 þ þ
and secondary amine or imine
6 Multihaloparaffin with active H 0 þ    0 0 þ 0
7 Aromatic, halogenated aromatic, olefin þ þ 0 þ 0 0 0 0 0
8 Paraffin þ þ þ þ þ þ 0 0 0
9 Monohaloparaffin or olefin þ þ þ þ þ 0 0 þ 0

(þ) Plus sign means that compounds in the column group tend to raise activity coefficients of compounds in the row group.
() Minus sign means a lowering of activity coefficients.
(0) Zero means no effect.
Choose a solvent that lowers the activity coefficient.
Source: R.W. Cusack, P. Fremeaux, and D. Glate, Chem Eng., 98(2), 66–76 (1991).

capacity for acetone. If the compound is environmentally
objectionable, though, it must be rejected.
A sophisticated solvent-selection method, based on the
UNIFAC group-contribution method for estimating activity
coefficients and utilizing a computer-aided constrained opti-
mization approach, has been developed by Naser and Four-
nier [23]. A preliminary solvent selection can also be made
using tables of partition coefficients and Godfrey Miscibility
Numbers in Perry’s Chemical Engineers’ Handbook [106].
§8.2.1 Representation of Ternary Data
Chapter 4 introduced ternary diagrams for representing
liquid–liquid equilibrium data at constant temperature. Such
diagrams are available for a large number of systems, as dis-
cussed by Humphrey et al. [6]. For ternary systems, the most
common diagram is Type I, shown in Figure 8.10a; much less
common are Type II systems, in Figure 8.10b, which include:
(1) n-heptane/aniline/methylcyclohexane, (2) styrene/ethyl-
benzene/diethylene glycol, and (3) chlorobenzene/water/
methyl ethyl ketone. For Type I, the solute and solvent are
miscible in all proportions, while in Type II they are not. For
Type I systems, the larger the two-phase region on line CS,
the greater the immiscibility of carrier and solvent. The
closer the top of the two-phase region is to apex A, the greater
the range of feed composition, along line AC, that can be sep-
arated with solvent S. In Figure 8.11, only feed solutions in
the composition range from C to F can be separated because,
regardless of how much solvent is added, two liquid phases
are not formed in the feed composition range of FA (i.e., FS
does not pass through the two-phase region). Figure 8.11a
has a wider range of feed composition than Figure 8.11b. For
Type II systems, a high degree of insolubility of S in C and C
in S will produce a desirable high relative selectivity, but at

Solute
A Solute
A

Plait Equilibrium
point curve
e
as

Extract
ph

phase
te

Raffinate Tie
na

phase lin
ffi

e
e
Ra

Tie lin
ph tract

Figure 8.10 Common classes of

e
as
Ex

Carrier Solvent Carrier Solvent ternary systems: (a) Type I, one

C S C S immiscible pair; (b) Type II, two
(a) (b) immiscible pairs.
C08 09/20/2010 Page 312
312 Chapter 8 Liquid–Liquid Extraction with Ternary Systems
A of interfacial tension by a significant solute presence in
A
F the interfacial film. If solvent is the dispersed phase, the
F
C S C S
(a) (b)
Figure 8.11 Effect of solubility on range of feed composition that
can be extracted.
the expense of solvent capacity. Solvents that result in Type I
systems are thus more desirable.
Whether a system is of Type I or Type II often depends on
temperature. The data of Darwent and Winkler [24] for the
ternary system n-hexane (H)/methylcyclopentane (M)/aniline
(A) for temperatures of 25, 34.5, and 45 C are shown in
Figure 8.12. At the lowest temperature, 25 C, the system is
Type II because both H and M are only partially miscible in
the aniline solvent. As temperature increases, solubility of M
in aniline increases more rapidly than the solubility of H in
aniline, until 34.5 C, the critical solution temperature for M
in aniline, where the system is at the border of Type II and
Type I. At 45 C, the system is clearly of Type I, with aniline
more selective for M than H. Type I systems have a plait
point (see Figure 8.10a); Type II systems do not.
Except in the near-critical region, pressure has little effect
on liquid–liquid equilibrium because it has little effect on
liquid-phase activity coefficients. The operating pressure
need only be greater than the bubble-point pressure. Most
extractors operate at near-ambient temperature, with the pro-
cess being nearly isothermal because of the small heat of
mixing. An exception is biochemical processes, where a
below-ambient temperature may be necessary to prevent
bioproduct degradation.
Laboratory or pilot-plant work, using actual plant feed and
solvent, is normally necessary to ascertain dispersion and
coalescence properties. Although rapid coalescence of drops
is desirable, this reduces interfacial area and leads to reduced
mass-transfer rates. Thus, compromises are necessary. Coa-
lescence is enhanced when the solvent phase is continuous
and mass transfer of solute is from the droplets. This
phenomenon, the Marangoni effect, is due to a lowering
M M
H A H A
T = 25°C T = 34.5°C
interfacial film is depleted of solute, causing an increase in
interfacial tension and inhibition of coalescence.
For a given (1) feed liquid, (2) degree of solute extraction,
(3) operating pressure and temperature, and (4) solvent for a
single-section cascade, there is a minimum solvent-to-feed
flow-rate ratio that corresponds to an infinite number of
countercurrent, equilibrium contacts. As with absorption and
stripping, a trade-off exists between the number of equili-
brium stages and the solvent-to-feed ratio. For a two-section
cascade, as for distillation, the trade-off involves the reflux
ratio and the number of stages. Algebraic methods—similar
to those for absorption and stripping—for computing the
trade-off are rapid but useful only for very dilute solutions,
where the solute activity coefficients are those near infinite
dilution. When the carrier and solvent are mutually insoluble,
the algebraic method of §5.3–5.4 applies. For more general
applications of ternary systems, the graphical methods
described in §8.3–8.4 are preferred. For higher-order, multi-
component systems, process simulator methods described in
Chapter 10 are necessary.
§8.3 HUNTER–NASH GRAPHICAL
EQUILIBRIUM-STAGE METHOD
Stagewise extraction calculations for Type I and Type II sys-
tems (Figure 8.10) are most conveniently carried out with
equilibrium diagrams [25]. In this section, procedures are
developed using mainly triangular diagrams. Other diagrams
are covered in §8.4.
Consider a countercurrent-flow, N-equilibrium-stage
extractor operating isothermally in steady-state, continuous
flow, as in Figure 8.13. Stages are numbered from the feed
end. Thus, the final extract is E1 and the final raffinate is RN.
Equilibrium is assumed at each stage, so for any stage n, the
components in extract En and raffinate Rn are in equilibrium.
Mass transfer of all species occurs at each stage. The feed, F,
contains the carrier, C, the solute, A, and the solvent, S, up to
the solubility limit. Entering solvent, S, can contain C and A,
but preferably contains little of either. Because most liquid–
liquid equilibrium data are listed and plotted in mass rather
than mole concentrations, let:
F ¼ mass flow rate of feed to the cascade;
S ¼ mass flow rate of solvent to the cascade;
M
Figure 8.12 Effect of temperature
on solubility for the system
H A n-hexane (H)/methylcyclopentane
T = 45°C (M)/aniline (A).
C08 09/20/2010 Page 313

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 313

Solvent S
Extract
E1 E2 E3 En En + 1 EN –1 EN

1 2 n N–1 N
Figure 8.13 Countercurrent-flow,
F
R1 R2 Rn–1 Rn RN–2 RN –1 Raffinate N-equilibrium-stage liquid–liquid
RN extraction cascade.
Feed

En ¼ mass flow rate of extract leaving stage n; direction. Sometimes the solutropy phenomenon disappears
Rn ¼ mass flow rate of raffinate leaving stage n; if mole-fraction coordinates, rather than mass-fraction coor-
dinates, are used.
(yi)n ¼ mass fraction of species i in extract leaving stage
n; and
(xi)n ¼ mass fraction of species i in raffinate leaving stage n. §8.3.1 Number of Equilibrium Stages

Although Figure 8.13 implies that the extract is the light From the degrees-of-freedom discussions in §4.1 and §5.8,
phase, either phase can be the light phase. Phase equilibrium the following sets of specifications for the ternary component
is represented, as in §4.5, on an equilateral-triangle diagram, cascade of Figure 8.13 can be made, where, in addition, all
as proposed by Hunter and Nash [26], or on a right-triangle sets include the specification of F, (xi)F, (yi)S, and T:
diagram, as proposed by Kinney [27]. Assume the ternary sys-
tem is A (solute), C (carrier), and S (solvent) at a temperature, Set 1. S and ðxi ÞRN Set 4. N and ðxi ÞRN
T, such that liquid–liquid equilibrium data are as shown in Set 2. S and ðyi ÞE1 Set 5. N and ðyi ÞE1
Figure 8.14, where the bold line is the equilibrium curve (also Set 3. ðxi ÞRN and ðyi ÞE1 Set 6. S and N
called the binodal curve because the plait point separates the
curve into an extract to the left and a raffinate to the right), where ðxi ÞRN and ðyi ÞE1 and all exiting phases lie on the equi-
and the dashed lines are tie lines connecting compositions of librium curve.
equilibrium phases. Because the tie lines slope upward from Calculations for sets 1 to 3 involve determination of N, and
the C side toward the S side, at equilibrium, A has a concentra- are made using triangular diagrams. Sets 4 to 6 involve a spec-
tion higher in S than in C. Thus, in this example, S is an effec- ified N, and require an iterative procedure. First consider set 1,
tive solvent for extracting A. Alternatively, because the tie with the procedures for sets 2 and 3 being minor modifica-
lines slope downward from the S side toward the C side, C is tions. The technique, sometimes called the Hunter–Nash
not an effective solvent for extracting A from S. method [26], involves three kinds of constructions on the
Some systems, such as isopropanol–water–benzene, triangular diagram, and is more difficult than the McCabe–
exhibit a phenomenon called solutropy, wherein moving Thiele staircase method for distillation. Although the proce-
from the plait point into the two-phase region of the diagram, dure is illustrated here only for a Type I system, parallel
the tie lines first slope in one direction; but then the slope principles apply to a Type II system. The constructions are
diminishes until an intermediate tie line becomes horizontal. shown in Figure 8.14, where A is the solute, C the carrier, and
Below that tie line, the remaining tie lines slope in the other S the solvent. On the equilibrium curve, extract compositions

(Solute)
A

90 10
Tie line
80 20 Operating line
Equilibrium curve
70 30

60 40
Plait
50 point 50

40 E1 60

30 70
F
20 M 80

Mmax 10 90
RN Figure 8.14 Construction 1: Location of
(Solvent) S C (Carrier)
90 80 70 60 50 40 30 20 10 product points.
C08 09/20/2010 Page 314

314 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

lie to the left of the plait point, while all raffinate compositions
lie to the right. Determination of N is as follows, given
feed and solvent flow rates and compositions, and raffinate
composition:
Construction 1 (Product Composition Points)
First, on Figure 8.14, locate mixing point M, which repre-
sents the combined composition of feed F plus entering
solvent S. Assume the following feed and solvent specifica-
tions, as plotted in Figure 8.14:
Feed Solvent
F ¼ 250 kg S ¼ 100 kg
(xA)F ¼ 0.24 (xA)S ¼ 0.00
(xC)F ¼ 0.76 (xC)S ¼ 0.00
(xS)F ¼ 0.00 (xS)S ¼ 1.00
By material balances, composition M of combined F and S
is:
M ¼ F þ S ¼ 250 þ 100 ¼ 350 kg
ðxA ÞM M ¼ ðxA ÞF F þ ðxA ÞS S
¼ 0:24ð250Þ þ 0ð100Þ ¼ 60 kg
ðxA ÞM ¼ 60=350 ¼ 0:171
ðxC ÞM M ¼ ðxC ÞF F þ ðxC ÞS S
¼ 0:76ð250Þ þ 0ð100Þ ¼ 190 kg
ðxC ÞM ¼ 190=350 ¼ 0:543
ðxS ÞM M ¼ ðxS ÞF F þ ðxS ÞS S
¼ 0ð250Þ þ 1ð100Þ ¼ 100 kg
ðxS ÞM ¼ 100=350 ¼ 0:286
From any two of the (xi)M values, point M is located, as
shown, in Figure 8.14. Based on properties of the triangular
diagram discussed in §4.5, point M must be located on the
straight line connecting F and S. Therefore, M can be located
knowing just one value of (xi)M, say, (xS)M. Also, the ratio S/F
is given by the inverse-lever-arm rule as
S=F ¼ MF=MS ¼ 100=250 ¼ 0:400
or
S=M ¼ MF=SF ¼ 100=350 ¼ 0:286
Thus, point M can be located by two points or by measure-
ment, employing either of these ratios.
With point M located, the composition of exiting extract
E1 is determined from overall material balances:
M ¼ RN þ E1 ¼ 350 kg
ðxA ÞM M ¼ 60 ¼ ðxA ÞRN RN þ ðxA ÞE1 E1
ðxC ÞM M ¼ 190 ¼ ðxC ÞRN RN þ ðxC ÞE1 E1
ðxS ÞM M ¼ 100 ¼ ðxS ÞRN RN þ ðxS ÞE1 E1
Specify ðxA ÞRN ¼ 0:025. Because it lies on the equilibrium
curve, RN can be located and the values of ðxC ÞRN and ðxS ÞRN
can be read from Figure 8.14. A straight line drawn from RN
through M locates E1 at the equilibrium-curve intersection,
from which the composition of E1 can be read. Values of the
flow rates RN and E1 can then be determined from the overall
material balances above, or from Figure 8.14 by the inverse-
lever-arm rule:
E1 =M ¼ MRN =E1 RN
RN =M ¼ ME 1 =E1 RN
with M ¼ 350 kg. By either method the results are:
Raffinate Product Extract Product
RN ¼ 198.6 kg E1 ¼ 151.4 kg
(xA) ¼ 0.025 (xA) ¼ 0.364
(xC) ¼ 0.90 (xC) ¼ 0.075
(xS) ¼ 0.075 (xS) ¼ 0.561
Also included in Figure 8.14 is point Mmax, which lies on
the equilibrium curve along the straight line connecting F
to S. Mmax corresponds to the maximum possible solvent
addition if two liquid phases are to exist.
Construction 2 (Operating Point and Operating Lines)
For vapor-liquid cascades, §6.3 and §7.2 describe an operat-
ing line that is the locus of passing streams in a cascade.
Referring to Figure 8.13, material balances around groups of
stages from the feed end are
F  E1 ¼    ¼ Rn1  En ¼    ¼ RN  S ¼ P ð8-5Þ
Because the passing streams are differenced, P defines a
difference point, not a mixing point, M. From the same geo-
metric considerations that apply to a mixing point, a differ-
ence point also lies on a line through the points involved.
However, whereas M lies inside the diagram and between the
two end points, P usually lies outside the triangular diagram
along an extrapolation of the line through two points such as
F and E1, RN and S, and so on.
To locate the difference point, two straight lines are drawn
through the passing-stream point pairs (E1, F) and (S, RN)
established by Construction 1 and shown in Figure 8.15.
These lines are extrapolated until they intersect at difference
point P. These lines and point P are shown in Figure 8.15.
From (8-5), straight lines drawn through points for any other
pair of passing streams, such as (En, Rn1), must also pass
through point P. Thus, the difference point becomes an oper-
ating point, and lines drawn through pairs of points for pass-
ing streams and extrapolated to point P are operating lines.
The difference point has properties similar to a mixing
point. If F  E1 ¼ P is rewritten as F ¼ E1 þ P; F can be
interpreted as the mixing point for P and E1. Therefore, by
the inverse-lever-arm rule, the length of line E1 P relative to
the length of line FP is
E1 P E1 þ P F
¼ ¼ ð8-6Þ
FP E1 E1
Thus, point P can be located with a ruler using either pair of
feed-product passing streams.
The operating point, P, lies on the feed or raffinate side of
the diagram in Figure 8.15. Depending on the relative
amounts of feed and solvent and the slope of the tie lines,
C08 09/20/2010 Page 315

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 315

Tie line
90 10
Operating line
80 20 Equilibrium curve
70 30

60 40
Plait
point
50 50

40 E1 60

30 70
F
20 80 Opera
ting li
ne Operating
10 90 point
P
RN
S C Operating line Figure 8.15 Construction 2: Location of
90 80 70 60 50 40 30 20 10
operating point.

point P may lie on the solvent or feed side of the diagram,
and inside or outside of the diagram.
Construction 3 (Tie Lines and Equilibrium Lines)
The third type of construction involves the tie lines that
define the equilibrium curve, which is divided into the two
sides (raffinate and extract) by the plait point. For Type I dia-
grams, the plait point is the composition of two equilibrium
phases that become one phase. A material balance around
stage n for any of the three components is
ðxi Þn1 Rn1 þ ðyi Þnþ1 Enþ1 ¼ ðxi Þn Rn þ ðyi Þn En ð8-7Þ
Because Rn and En are in equilibrium, their composition
points are at the two ends of a tie line. Typically, a diagram
will not contain all tie lines needed; however, they may be
added by centering them between existing tie lines, or by
using either of two interpolation procedures illustrated in
Figure 8.16. In Figure 8.16a, the conjugate line from the plait
C at the opposite end of a tie line (shown as a dashed line) con-
Plait
point
E
D with a tie line by Construction 3; from R2, E3 is located by
Plait
A H B
F G
point
D Procedures for specification sets 2 and 3 are similar to that
J
A J B
(a)
Figure 8.16 Use of conjugate curves to interpolate tie lines:
(a) method of International Critical Tables, Vol. III, McGraw Hill,
New York, p. 393 (1928); (b) method of T.K. Sherwood, Absorption
and Extraction, McGraw-Hill, New York, p. 242 (1937).
[From R.E. Treybal, Liquid Extraction, 2nd ed., McGraw-Hill, New York
(1963) with permission.]
point to J is determined from four tie lines and the plait point.
From tie line DE, lines DG and EF are drawn parallel to
triangle sides CB and AC, respectively. The intersection at
point H gives a second point on the conjugate curve. Subse-
quent intersections establish additional points from which the
conjugate curve is drawn. Then, using the curve, additional
tie lines are drawn by reversing the procedure. If it is desired
to keep the conjugate curve inside the two-liquid-phase
region of the triangular diagram, the procedure illustrated in
Figure 8.16b is used, where lines are drawn parallel to trian-
gle sides AB and AC.
Stepping Off Stages
Equilibrium stages are stepped off by alternate use of tie lines
and operating lines, as shown in Figure 8.17, where Con-
structions 1 and 2 have been employed to locate points F, E,
S, R1, and P. Starting at the feed end from point E1 and refer-
ring to Figure 8.13, it is seen that R1 is in equilibrium with
E1. Therefore, by Construction 3, R1 in Figure 8.17 must be
necting to E1. From Figure 8.13, R1 passes E2. Therefore, by
Construction 2, E2 lies at the intersection of the straight oper-
ating line drawn through points R1 and P, and back to the
extract side of the equilibrium curve. From E2, R2 is located
C Construction 2. Continuing in this fashion by alternating
between equilibrium tie lines and operating lines, the speci-
fied point RN is reached or passed. If the latter, a fraction of
the last stage is taken. In Figure 8.17, 2.8 equilibrium stages
E are required, where equilibrium stages are counted by the
number of equilibrium tie lines used.
H
for set 1. Sets 4 and 5 can be handled by iteration on assumed
values for S and following the procedure for set 1. Set 6 can
(b) also use the procedure of set 1 by iterating on E1.
From (8-6), it is seen that if the ratio F=E1 approaches a
value of 1, operating point P will be located at a large dis-
tance from the triangular diagram. In that case, by using an
arbitrary rectangular-coordinate system superimposed over
the triangular diagram, the coordinates of P can be calculated
from (8-6) using the equations for the two straight lines
C08 09/20/2010 Page 316

316 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

90 10
Tie line
80 20
Operating line
70 30 Equilibrium curve

60 40
Plait
point
50 50

40 E1 60

30 70
F
20 E2 80
R1 90
10 E3
R2 P Figure 8.17 Determination of the
S RN C number of equilibrium stages.
90 80 70 60 50 40 30 20 10

established in Construction 2. Operating lines for intermedi- line OL is drawn through the points S and RN and extended to
ate stages can then be located on the triangular diagram so as the left and right. This line is the locus of all possible mate-
to pass through P. Details of this procedure are given by rial balances determined by adding S to RN. Each tie line is
Treybal [25]. then assumed to be a pinch point by extending each tie line
until it intersects the line OL. In this manner, a sequence of
§8.3.2 Minimum and Maximum Solvent-to-Feed intersections P1, P2, etc., is found. If these points lie on the
Flow-Rate Ratios raffinate side of the diagram, as in Figure 8.18, the pinch
point corresponds to the point Pmin located at the greatest dis-
The procedure of §8.3.1 for determining the number of tance from RN. If the triangular diagram does not have a suf-
equilibrium stages needed to achieve a desired solute ficient number of tie lines to determine that point accurately,
extraction for a given solvent-to-feed ratio presupposes additional tie lines are introduced by a method illustrated in
that this ratio is greater than the minimum ratio, which Figure 8.16. If it is assumed in Figure 8.18 that no other tie
corresponds to an infinite number of stages, and less than line gives a point Pi that is farther away from RN than P1,
the maximum ratio that would prevent the formation of then P1 ¼ Pmin .
the required second liquid phase. In practice, one usually With Pmin known, an operating line can be drawn through
determines the minimum ratio before solving specification point F and extended to E1 at an intersection with the extract
sets 1 or 2. This is done by solving set 4 with N ¼ 1, side of the equilibrium curve. From the compositions of the
where, as in distillation, absorption, and stripping, the in- four points S, RN, F, and E1, the mixing point M can be found
finity of stages occurs at an equilibrium-curve and operat- and material balances can be used to solve for Smin=F:
ing-line pinch point. In ternary systems, the pinch point
occurs when a tie line coincides with an operating line. F þ Smin ¼ RN þ E1 ¼ M ð8-8Þ
The calculation is involved because the pinch point is not ðxA ÞF F þ ðxA ÞS Smin ¼ ðxA ÞM M ð8-9Þ
always at the feed end of the cascade.
from which
Consider the previous A–C–S system, shown in Figure
8.18. The composition points F, S, and RN are specified, but Smin ðxA ÞF  ðxA ÞM
¼ ð8-10Þ
E1 is not because the solvent rate is not known. The operating F ðxA ÞM  ðxA ÞS

90 10 Tie line
Operating line
80 20
Equilibrium curve
70 30

60 Plait 40
Extract point Raffinate
side 50 50 side
40 60

30 70
F
20 80

10 90 L
O RN Figure 8.18 Determining
S P2 P3 P1 = Pmin
90 80 70 60 50 40 30 20 10 C minimum solvent-to-feed ratio.
C08 09/20/2010 Page 317

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 317

A solvent flow rate greater than Smin is required for extraction
in a finite number of stages. In Figure 8.18, such a solvent
rate results in an operating point P to the right of Pmin, which
is a location farther away from RN. A reasonable value for S
might be 1.5 Smin. From Figure 8.18, (xA)M ¼ 0.185, from
which, by (8-10), Smin=F ¼ 0.30. In Figure 8.17, S=F ¼ 0.40,
giving S=Smin ¼ 1.33.
In Figure 8.18 the tie lines slope downward toward the
raffinate side. If the tie lines slope downward toward the
extract side of the diagram, the above procedure for find-
ing Smin=F is modified. The sequence of points P1, P2,
etc., is now on the other side of the diagram. However, the
pinch point now corresponds to the point, Pmin, that is
closest to point S, so an operating point, P, must be chosen
between points Pmin and S. For a system that exhibits solu-
tropy, intersections P1, P2, etc., are found on both sides of
the diagram. Those on the extract side determine the mini-
mum solvent-to-feed ratio.
In Figure 8.14, mixing point M lies in the two-phase
region. As this point moves along the line SF toward S, the
ratio S=F increases according to the inverse-lever-arm rule.
In the limit, a maximum S=F ratio is reached when M ¼
Mmax arrives at the equilibrium curve on the extract side.
Now all of the feed is dissolved in the solvent, no raffinate is
obtained, and only one stage is required. To avoid this
impractical condition, as well as the other extreme of infinite
stages, it is necessary to select a solvent ratio, S=F, such that
ðS=FÞmin < ðS=FÞ < ðS=FÞmax . In Figure 8.14, the mixing
point Mmax is located as shown, from which (S=F)max is
determined to be about 16.
EXAMPLE 8.1 Equilbrium Stages for Extraction.
Acetone is to be extracted from a 30 wt% acetone (A) and 70 wt%
ethyl acetate (C) feed at 30 C, using pure water (S) as the solvent,
by the cascade shown at the bottom of Figure 8.19, which estab-
lishes the nomenclature for this example. The final raffinate is
to contain 5 wt% acetone on a water-free basis. Determine the
minimum and maximum solvent-to-feed ratios and the equili-
brium stages required for two intermediate S=F ratios. The data
shown in Figure 8.19 are from Venkataranam and Rao [28] and
correspond to a Type I system, but with tie lines sloping down-
ward toward the extract side. Thus, although water is a convenient
solvent, it does not have a high capacity, relative to ethyl acetate,
for dissolving acetone. Also determine, for the feed, the maxi-
mum weight % acetone that can enter the extractor. This example,
as well as Example 8.2, are taken largely from an analysis by
Sawistowski and Smith [29].
Solution
Point B represents the solvent-free raffinate. By drawing a straight
line from B to S, the intersection with the equilibrium curve on the
raffinate side, B0 , is the raffinate composition leaving stage N.
Minimum S=F. Because the tie lines slope downward toward the
extract side, the extrapolated tie line that intersects the extrapolated
line SB closest to S is sought. This tie line, leading to Pmin, is shown
in Figure 8.20. The intersection is not shown because it occurs far to
the left of the diagram. Because this tie line is at the feed end of the
extractor, location of extract composition D0 min is found as shown in
Figure 8.20. The mixing point, Mmin, for (S=F)min is the intersection
of lines B0 D0 min and SF. By the inverse-lever-arm rule,
ðS=F Þmin ¼ FM min =SM min ¼ 0:60.

Acetone
A

10 90

20 80
Tie line
Operating line 30 70
W
ter

Equilibrium curve
t%
wa

40 60
ac
t%

eto

50 50
W

ne

Plait point
Raf
60 env finate 40
ct elo
pe F
tra pe
70 Ex elo 30
v
en
80 Tie line L1 20
M
V2 D′ L2
10
B
B′
P S 10 20 30 40 50 60 70 80 90 C
Water Wt % acetate Ethyl acetate

V2 VN
D′ S, solvent
Solvent
D 1 2 N
removal Solvent
F B Figure 8.19 Determination of stages for
L1 LN – 1 B′ removal
(30% A) (5% A) Example 8.1 with S=F ¼ 1.75.
C08 09/20/2010 Page 318

318 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Acetone
A

10 90
Tie line
Operating line 20 80
Equilibrium curve
30 70

W
ter
D

t%
wa
40 60

ac
t%

eto
50 50

ne
Extract Raffinate
side side
60 G 40
L1
70 D′min Tie line
To Pmin F

80 Mmin 20

90 10
Water S B
C
10 20 30 40 50 60 70 80 90 Acetate Figure 8.20 Minimum and
Extrapolated Mmax
line SB D′max
Wt % acetate
B′ maximum S=F for Example 8.1.
Pmax

§8.3.3 Use of Right-Triangle Diagrams

Maximum S=F. If M in Figure 8.20 is moved along line FS toward
S, the intersection for (S/F)max occurs at the point shown on the
extract side of the binodal curve. By the inverse-lever-arm rule,
using line FS, ðS=F Þmax ¼ FM max =SM max ¼ 12.
Equilibrium stages for other S=F ratios. First consider S=F ¼
1.75. In Figure 8.19, the composition of the saturated extract D0 is
obtained from an extractor material balance,
S þ F ¼ D0 þ B0 ¼ M
For S=F ¼ 1.75, point M is located such that FM=MS ¼ 1:75. A
straight line must pass through D0 , B0 , and M. Therefore, D0 is lo-
cated by extending B0 M to the extract envelope.
The difference point P is located to the left of the diagram.
Therefore, P ¼ S  B0 ¼ D0  F. It is located at the intersection of
extensions of lines FD0 and B0 S.
Step off stages, starting at D0 , by following a tie line to L1. Then
V2 is located by noting the intersection of the operating line L1 P
with the phase envelope. Additional stages are stepped off by alter-
nating between the tie lines and operating lines. Only the first stage
is shown; four are required.
For S=F ¼ 5ðS=FÞmin ¼ 3:0, M is determined and the stages are
stepped off in a similar manner to give two equilibrium stages. In
summary, for the countercurrent cascade,
S=F (solvent/feed ratio) 0.60 1.75
N (equilibrium stages) 1 4
xD (wt% acetone, solvent free) 64 62
If the wt% acetone in the feed mixture is increased from the base
value of 30%, a feed composition will be reached that cannot be
extracted because two liquid phases in equilibrium will not form
(no phase splitting). This feed composition is determined by extend-
ing a line from S, tangent to the equilibrium curve, until it intersects
AC at point D in Figure 8.20. The feed composition is 64 wt% ace-
tone. Feed mixtures with a higher acetone content cannot be
extracted with water.
Ternary, countercurrent extraction calculations can also be
made on a right-triangle diagram, discussed in §4.5 and
shown by Kinney [27]; no new principles are involved. The
disadvantage is that mass-percent compositions of only two
of the components are plotted; the third is determined, when
needed, by difference from 100%. The advantage of right-
triangle diagrams is that ordinary, rectangular-coordinate
graph paper can be used and either one of the coordinates can
be expanded, if necessary, to increase construction accuracy.
A right-triangle diagram can be developed from an equi-
lateral-triangle diagram, as shown in Figure 8.21a, where
coordinates are in mass or mole fractions. Point P on the
equilibrium curve and tie line RE in the equilateral triangle
become point P and tie line RE in the right-triangle diagram,
which uses rectangular coordinates xA and xC, where A is the
solute and C is the carrier.
Consider the right-triangle diagram in Figure 8.22 for the
A–C–S system of Figure 8.14. The compositions of S (the
solvent) and A (the solute) are plotted in mass fractions, xi.
For example, point M represents a liquid mixture of overall
composition ðxA ¼ 0:43; xS ¼ 0:28Þ. By contrast, xC, the
carrier, which is not shown, is 1  0.43  0.28 ¼ 0.29. Al-
3 12 though lines of constant xC are included in Figure 8.22, they
2 1 are usually omitted because they clutter the diagram. As with
50 30 the equilateral-triangle diagram, Figure 8.22 includes the
binodal curve with extract and raffinate sides, tie lines con-
necting equilibrium phases, and the plait point.
Because point M falls within the phase envelope, the mix-
ture separates into two liquid phases, given by points A0 and
A00 at the ends of the tie line passing through M. The extract
at A00 is richer in the solute (A) and the solvent (S) than the
raffinate at A0 .
Point M might be the result of mixing a feed, point F, con-
sisting of 26,100 kg/h of 60 wt% A in C (xA ¼ 0.6, xS ¼ 0),
with 10,000 kg/h of pure furfural, point S. The mixture splits
C08 09/20/2010 Page 319

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 319

A
1.0 A

yA
xA
P
P G
P D
E
E E 45° Line
R
R R
0
C D F S 0D F 1.0 C S xA
xC
(a) (b)

A
y Equilibrium
curve
y1
x, y 1 Figure 8.21 Development
of other coordinate systems
45° Line from the equilateral-triangle
Operating
ys+1
diagram: (a) to right-trian-
K curve
gle diagram; (b) to auxiliary
y1 distribution curve.
(c) Location of operating
xF F
point on auxiliary McCabe–
ys+1 J
xs Thiele diagram.
Operating I [From R.E. Treybal, Liquid
S ys
point
ys Extraction, 2nd ed., McGraw-
C xs xF x Hill, New York (1963) with
permission.]
(c)

into phases A0 and A00 . The location of M and the amounts of solute in the raffinate and extract, respectively, and L and V
extract and raffinate are given by the same mixing rule and are total amounts of raffinate and extract. The feed to stage N
inverse-lever-arm rule used for equilateral-triangle diagrams. is LNþ1 ¼ 180 kg of 35 wt% A in a saturated mixture with C
The mixture separates spontaneously into 11,600 kg/h of raf- and S ðxNþ1 ¼ 0:35Þ, and the solvent to stage 1 is VW ¼ 100
finate ðxS ¼ 0:08; xA ¼ 0:32Þ and 24,500 kg/h of extract kg of pure S (yW ¼ 0.0). Thus, the solvent-to-feed ratio is
ðxS ¼ 0:375; xA ¼ 0:48Þ. 100/180 ¼ 0.556. These points are shown in Figure 8.24.
Figure 8.23 represents the portion of an n-stage, counter- The mixing point for LN+1 and VW is M1, as determined by
current-flow cascade, where x and y are weight fractions of the inverse-lever-arm rule.
Suppose the final raffinate, LW, leaving stage 1 is to con-
S tain xW ¼ 0.05 glycol. By an overall balance,
1.0
M 1 ¼ V W þ LNþ1 ¼ V N þ LW ð8-11Þ
0.9 Phase envelope Because VW, LN+1, and M1 lie on a straight line, the mixing
0.8 rule requires that VN, LW, and M1 also lie on a straight line.
Furthermore, because VN leaves stage N at equilibrium and
0.7 0 LW leaves stage 1 at equilibrium, these streams lie on the
C
i on extract and raffinate sides, respectively, of the equilibrium
ct
1

0.6
Ex
Wt fraction S

ra
tf
Tie line

tra

1 W VN – 1 VN – 2
xC
ct x C

yN – 1 yN – 2 VW = 100
0.5
=

VN
0

VW, yW = 0
=

yN = ?
xC

0.
1

N–2
=

0.4 (Solvent)
xC

A′′ N N–1
0.
0

2
=
xC

0.

LN + 1 = 180
3
=

0.3 LW, xW = 0.05

xC

M
0.

xN + 1 = 0.35 LN LN – 1
4
=
xC

0.

(Feed) xN xN – 1
5
=

0.2
xC

0.

(a)
6
=
xC

0.

Plait
7
=

0.1 point VN2 VW = 100 VN3 VW = 100

0.
xC

Raffinate A′ yN2 yW = 0 yN3 yW = 0

=

F
0.

A LN + 1 = 55 LW LN + 1 = 600 LW
9

C 00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
xN + 1 = 0.35 xW = 0.05 xN + 1 = 0.35 xW = 0.05
0 1
Wt fraction A (b) (c)
Figure 8.22 Right-triangle diagram for system of Figure 8.14. Figure 8.23 Multistage, countercurrent contactors.
C08 09/20/2010 Page 320

320 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

P2 If stages are stepped off for case 3, starting from yN3, the
tie line and operating line coincide, giving a pinch point for a
solvent-to-feed ratio of 0.167. Thus, this ratio is the mini-
mum corresponding to 1 equilibrium stages.
For case 1, where the solvent-to-feed ratio is between that
S y ,V
of cases 2 and 3, the required number of stages lies between 1
1.0 W W and 1. The difference point and the steps for this case are
not shown in Figure 8.14, but the difference point is located
0.9
at a very large distance from the triangle because lines LW V W
0.8 yN2 VN2 and LNþ1 V N are nearly parallel. When the stages are stepped
off, using operating lines parallel to LW V W , between one and
Tie lines

0.7
Ex

yN, VN two stages are required.

tr

M2
ac
t
1

0.6
Wt fraction S

§8.3.4 Use of an Auxiliary Distribution Curve

0.5
As the number of stages on either one of the two types of
0.4 triangular diagrams increases, the diagram becomes clut-
0

M1 yN3, VN3 tered, and a triangular diagram in conjunction with the

0.3
McCabe–Thiele method, devised by Varteressian and Fenske
0.2 [30], becomes attractive. The y–x diagram discussed in §4.5
M3
and illustrated in Figure 8.21b is simply a plot of the tie-line
0.1
Raffinate
xN +1, LN +1 data in terms of solute mass fractions in the extract (yA) and
LW, xW equilibrium raffinate (xA). The curve begins at the origin and
C A
0 0.1 0.2 0.3 0.4 0.5 0.6
0 1
0.7 0.8 0.9 1.0
terminates at the plait point, where it intersects the 45 line.
Wt fraction A A tie line, such as RE in the triangular diagram, becomes a
point in the equilibrium curve of Figure 8.25.
If an operating line is added to the equilibrium curve in
Figure 8.21b, the McCabe–Thiele staircase construction
P3 determines the equilibrium stages required. However, unlike
Figure 8.24 Right-triangle diagram constructions for cascade in in distillation, where the operating line is straight, the operat-
Figure 8.23. ing line for liquid–liquid extraction is curved except in the
low-solute region. Fortunately, the curved operating line is
curve. The resulting points are shown in Figure 8.24, where it readily drawn using the technique of Varteressian and Fenske
is seen that the glycol in the extract is yN ¼ 0.34.
Figures 8.23b, 8.23c, and 8.24 include two additional
cases of solvent-to-feed ratio, each with the same composi- 0.6
tions for the solvent and feed, and the same value for xW:

Feed Solvent, Solvent-to Extract Mixing 0.5 e3 Plait

Case LN+1, kg VW, kg Feed Ratio Designation Point Stag Feed point
Mass fraction of solute, A, in extract, y

end
1 180 100 0.556 VN M1
e2
ag

2 55 100 1.818 VN2 M2 0.4

St

3 600 100 0.167 VN3 M3

curve m

ve
ibriu

ur

For case 2, a difference point, P2, is defined in terms of 0.3

gc

e
lin
Equil

tin

passing streams, as
°
45
era

P2 ¼ V N2  LNþ1 ¼ V W  LW ð8-12Þ
Op

0.2
e1

In Figure 8.24 P2 is at the top of the diagram, where LW V W

Stag

and LNþ1 V N2 intersect. For case 3, difference point P3 lies at

the bottom, where LW V W and LNþ1 V N3 intersect. 0.1

Equilibrium stages for Figure 8.24 are stepped off as

Solvent end
before by alternating the use of equilibrium tie lines and
operating lines that pass through the difference point. Consider- 0 0.1 0.2 0.3 0.4 0.5
ing case 2, with a solvent-to-feed ratio of 1.818, and stepping
off stages from stage N, a tie line from the point yN2 gives xN Mass fraction of solute, A, in raffinate, x
¼ 0.04. But this is less than the specified xW ¼ 0.05, so only a Figure 8.25 Stepping off stages by the McCabe–Thiele method for
fraction of a stage is required. the A–C–S system.