----14-----
EXTRACTION ANO LEACHING
Extraccion
xtraction is a process whereby a mixture of severa/
E substances in the liquid phase is at /east partially
separated upan addition of a liquid solvent in which
the original substances have different solubilities.
When sorne of the original substances are solids, the process
is called /eaching. In a sense, the role of solvent in extraction
is analogous to the role of enthalpy in distillation. The
solvent-rich phase is cal/ed the extract, and the solvent-poor
phase is called the raffinate. A high degree of separation may
be achieved with severa/ extraction stages in series,
particular/y in countercurrent flow.
Processes of separation by extraction, distillation,
crystallization, ar adsorption sometimes are equal/y possible.
Oifferences in solubility, and hence of separability by
extraction, are associated with differences in chemical
structure, whereas differences in vapor pressure are the basis
of separation by distillation. Extraction often is effective at
near-ambient temperatures, a valuable feature in the
separation of thermally unstable natural mixtures or
pharmaceutical substances such as penicillin.
The simplest separation by extraction involves two
substances and a solvent. Equilibria in such cases are
represented convenient/y on triangular diagrams, either
equilateral ar right-angled, as far example on Figures 14.1 and
14.2. Equivalent representations on rectangular coordinates
a/so are shown. Equilibria between any number of substances
are representable in terms of activity coefficient correlations
such as the UNIOUAC ar NRTL. In theory, these correlations
involve only parameters that are derivable from
measurements on binary mixtures, but in practice the
resulting accuracy may be poor and sorne multicomponent
equilibrium measurements a/so should be used to find the
parameters. Finding the parameters of these equations is a
comp/ex enough operation to require the use of a computer.
An extensive compilation of equilibrium diagrams and
UNIOUAC and NRTL parameters is that of Sorensen and Arlt
(1979-1980). Extensive bibliographies have been compiled by
Wisniak and Tamir (1980-1981).
The highest degree of separation with a minimum of
14.1. EQUILIBRIUM RELATIONS
On a ternary equilibriurn diagrarn like that of Figure 14.1, the lirnits
of mutual solubilities are rnarked by the binodal curve and the
cornpositions of phases in equilibriurn by tielines. The region within
the dome is two-phase and that outside is one-phase. The rnost
cornrnon systerns are those with one pair (Type 1, Fig. 14.1) and two
pairs (Type 11, Fig. 14.4) of partially miscible substances. For
instance, of the approxirnately 1000 sets of data collected and
analyzed by Sorensen and Arlt (1979), 75% are Type 1 and 20% are
Type 11. The rernaining srnall percentage of systerns exhibit a
considerable variety of behaviors, a few of which appear in Figure
14.4. As sorne of these exarnples show, the effect of ternperature on
phase behavior of liquids often is very pronounced.
Both equilateral and right triangular diagrarns have the
property that the cornpositions of mixtures of ali proportions of two
mixtures appear on the straight line connecting the original
459
solvent is attained with a series of countercurrent stages.
Such an assemb/y of mixing and separating equipment is
represented in Figure 14.3(a}, and more schematical/y in
Figure 14.3(b). In the /aboratory, the performance of a
continuous countercurrent extractor can be simulated with a
series of batch operations in separatory funnels, as in Figure
14.3(c). As the number of operations increases horizontal/y,
the terminal concentrations E1 and R3 approach asymptotical/y
those obtained in continuous equipment. Various kinds of
more sophisticated continuous equipment a/so are widely
used in /aboratories; some are described by Lo et al. ( 1983,
pp. 497-506). Laboratory work is of particular importance far
complex mixtures whose equilibrium refations are not known
and far which stage requirements cannot be calculated.
In mixer-separators the contact times can be made long
enough far any desired approach to equilibrium, but 80-90%
efficiencies are economical/y justifiable. lf five stages are
required to duplicate the performance of tour equilibrium
stages, the stage efficiency is 80%. Since mixer-separator
assemblies take much floor space, they usual/y are emp/oyed
in batteries of at most tour or five units. A large variety of
more compact equipment is being used. The simplest in
concept are various kinds of tower arrangements. The
relations between their dimensions, the operating conditions,
and the equivalent number of stages are the key infarmation.
Ca/culations of the relations between the input and
output amounts and compositions and the number of
extraction stages are based on material balances and
equi/ibrium relations. Knowledge of efficiencies and capacities
of the equipment then is applied to find its actual size and
configuration. Since extraction processes usual/y are
performed under adiabatic and isothermal conditions, in this
respect the design problem is simpler than far thermal
separations where enthalpy balances a/so are involved. On
the other hand, the design is complicated by the fact that
extraction is feasible only of nonideal liquid mixtures.
Consequently, the activity coefficient behaviors of two /iquid
phases must be taken into account or direct equilibrium data
must be availab/e.
mixtures. Moreover, the relative arnounts of the original mixtures
corresponding to an overall cornposition rnay be found frorn ratios
of line segrnents. Thus, on the figure of Exarnple 14.2, the arnounts
of extract and raffinate corresponding to an overall cornposition M
are in the ratio Eif RN = MRN / E 1M .
Experimental data on only 26 quaternary systerns were found
by Sorensen and Arlt (1979), and none of more cornplex systerns,
although a few scattered rneasurernents do appear in the literature.
Graphical representation of quaternary systerns is possible but
awkward, so that their behavior usually is analyzed with equations.
To a lirnited degree of accuracy, the phase behavior of cornplex
mixtures can be predicted frorn rneasurernents on binary mixtures,
and considerably better when sorne ternary rneasurernents also are
available. The data are correlated as activity coefficients by rneans
of the UNIQUAC or NRTL equations. The basic principie of
application is that at equilibriurn the activity of each cornponent is
the sarne in both phases. In terrns of activity coefficients this
460 EXTRACTION AND LEACHING

15

o
+
10
en

5
100 o
o Mol% B 100
Diluent A Solvent B
B-rich
(a) o
1.0

7o c ""''
< 1l
.!:::
o.
60 .!:::
u
·e:
0.5
ü ü
<!!- +
o
:2
ü
20

20 40 60 0.5
Mol % B C / (A + C), A-rich phase
(b) (e)
Figure 14.1. Equilibria in a ternary systern, type 1, with one pair of partially miscible liquids;
A = 1-hexene, B = tetrarnethylene sulfone, C = benzene, at 50ºC (R.M . De Fre, thesis, Gent, 1976).
(a) Equilateral triangular plot; point P is at 20% A, 10% B, and 70% C. (b) Right triangular plot with
tielines and tieline locus, the arnount of A can be read off along the perpendicular to the hypotenuse or
by difference. (e) Rectangular coordinate plot with tieline correlation below, also called Janecke and
solvent-free coordinates.

condition is for cornponent i, Cornpositions of petroleurn mixtures sornetirnes are repre-

Y;X; = rtxf , (14.1)
where • designates the second phase. This rnay be rearranged into a
relation of distributions of cornpositions between the phases,
(14.2)
where K; is the distribution coefficient. The activity coefficients are
functions of the cornposition of the mixture and the ternperature.
Applications to the calculation of stage requirernents for extraction
are described later.
Extraction behavior of highly cornplex mixtures usually can be
known only frorn experirnent. The sirnplest equiprnent for that
purpose is the separatory funnel, but cornplex operations can be
sirnulated with proper procedures, for instance, as in Figure 14.3(c).
Elaborate autornatic Iaboratory equiprnent is in use. One of thern
ernploys a 10,000-25,000 rprn mixer with a residence time of 0.3-
5.0 sec, followed by a highly efficient centrifuge and two
chrornatographs for analysis of the two phases (Lo et al., 1983, pp.
507).
sented adequately in terrns of sorne physical property. Three
exarnples appear in Figure 14.5. Straight line cornbining of mixtures
still is valid on such diagrarns.
Basically, cornpositions of phases in equilibriurn are indicated
with tielines. For convenience of interpolation and to reduce the
clutter, however, various kinds of tieline loci rnay be constructed,
usually as loci of intersections of projections frorn the two ends of
the tielines. In Figure 14.l the projections are parallel to the base
and to the hypotenuse, whereas in Figures 14.2 and 14.6 they are
horizontal and vertical.
Severa! tieline correlations in equation forrn have been
proposed, of which three rnay be presented. They are expressed in
weight fractions identified with these subscripts:
CA solute C in diluent phase A
es solute e in solvent phase s
SS solvent S in solvent phase S
AA diluent A in diluent phase A
AS diluent A in solvent phase S
SA solvent S in diluent phase A.
 Solute C
o 100

e
100

80

ü 60
o
o
40

20

100 o
o Mol % B 100 A 00 20 60 80
Diluent A Solvent B Mol % B

(a) (b)

10

ü
+

""''
<ll
.<=
o.
.<=
o
·¡::
cii
ü
+

C / (A + C), A-rich phase

{e)

Figure 14.2. Equilibria in a ternary system, type II, with two pairs of partially miscible liquids; A = hexane, B = aniline,
C = methylcyclopentane, at 34.5°C [Darwent and Winkler, J. Phys. Chem. 47, 442 ( 1943)]. (a) Equilateral triangular plot.
(b) Right triangular plot with tielines and tieline locus. (e) Rectangular coordinate plot with tieline correlation below, also called
Janecke and solvent-free coordinates.

461
 Stage Stage Stage
1 2 3
Final
extract
E, Ez E,

Settler

R, Final
R5 raffinate
Feed Salven!

(a)

s• Fresh solvenr R= Raf f inare

F • Feed to be exlracted E• Exr rocl

(b) (e)

Figure 14.3. Representation of countercurrent extraction batteries. (a) A battery of mixers and settlers (or separators).
(b) Schematic of a three-stage countercurrent battery. (c) Simulation of the performance of a three-stage continuous
countercurrent extraction battery with a series of batch extractions in separatory funnels which are designated by circles
on the sketch. The numbers in the circles are those of the stages. Constant amounts of feed F and solvent S are mixed at
the indicated points. As the number of operations is increased horizontally, the terminal compositions E1 and R 3
approach asymptotically the values obtained in continuous countercurrent extraction ( Treybal, 1963, p. 360).

(a) (b)

Figure 14.4. Less common examples of ternary equilibria and sorne temperature effects. (a) The system 2,2,4-tri-
methylpentane + nitroethane + perftuorobutylamine at 25ºC; the Roman numerals designate the number of phases in
that region [Vreeland and Dunlap, J. Phys. Chem. 61, 329 (1957)]. (b) Same as (a) but at 51.3ºC. (e) Gly-
col + dodecanol + nitroethane at 24ºC; 12 different regions exist at 14ºC [Francis, J. Phys. Chem. 60, 20 (1956)].
(d) Docosane + furfural + diphenylhexane at severa! temperatures [Varteressian and Fenske, Ind. Eng. Chem. 29, 270
(1937) ]. (e) Formic acid + benzene + tribromomethane at 70ºC; the pair formic acid/benzene is partially miscible with 15 and
90% of the former at equilibrium at 25ºC, 43 and 80% at 70ºC, but completely miscible at sorne higher temperature.
(f) Methylcyclohexane + water + -picoline at 20ºC, exhibiting positive and negative tieline slopes; the horizontal tieline is
called solutropic ( Landolt-Bornstein I/2b ).

462
 14.2. CALCULATION OF STAGE REOUIREMENTS 463

OipMnyfhexant1

o
OocosDM 20 30 40 50 60 70 60 Furfuraf
(e)
(d)

:t¡L-'----"'.....:..--'- Q S -4- _,,--;:n- z,:::==========as::::=:::::::::; ::;::: z1

CH2 02 x6 - C,H,. .r.- - H10
(e) (f)
Figure 14.4-(continued )

Ishida, Bull. Chem. Soc. Jpn. 33, 693 (1960):
(14.3)
Othrner and Tobias, Ind. Eng. Chem. 34, 693 (1942):
(14.4)
Hand, J. Phys. Chem. 34, 1961 (1930):
(14.5)
These equations should plot linearly on log-log coordinates; they
are tested in Exarnple 14.1.
A systern of plotting both binodal and tieline data in terrns of
certain ratios of concentrations was devised by Janecke and is
illustrated in Figure 14.l(c). lt is analogous to the enthalpy-
concentration or Merkel diagrarn that is useful in solving distillation
problerns. Straight line cornbining of mixture cornpositions is valid
in this rnode. Calculations for the transforrnation of data are rnade
rnost conveniently frorn tabulated tieline data. Those for Figure
14.1 are rnade in Exarnple 14.2. The x-y construction shown in
Figure 14.2 is the basis for a McCabe-Thiele construction for
finding the nurnber of extraction stages, as applied in Figure 14.7.
14.2. CALCULATION OF STAGE REQUIREMENTS
Although the rnost useful extraction process is with countercurrent
flow in a rnultistage battery, other rnodes have sorne application.
Calculations rnay be perforrned analytically or graphically. On
flowsketches like those of Exarnple 14.1 and elsewhere, a single box
represents an extraction stage that rnay be rnade up of an individual
mixer and separator. The performance of differential contactors
such as packed or spray towers is cornrnonly described as the height
equivalent to a theoretical stage (HETS) in ft or rn.
SINGLE STAGE EXTRACTION
The material balance is
feed + solvent = extract + raffinate,
(14.6)
F + S = E + R.
This nornenclature is shown with Exarnple 14.3. On the triangular
diagrarn, the proportions of feed and solvent locate the mix point
 464 EXTRACTION ANO LEACHING

•100._ .............
...
OISTILLllTE NITROSENZENE
i ,L_...t:::===========L_
(a) KEROSENE g5 •4 ETHYL ALCOHOL

(b)

PROPANE

0.90 1 .00 1 .10

OI L Speci fic gravi ty at 70F ASPHALT
(e)

Figure 14.S. Representation of solvent extraction behavior in terms of certain properties rather than direct compositions [Dunstan et al., Sci.
Pet., 1825-1855 (1938)]. (a) Behavior of a naphthenic distillate of VGC = 0.874 with nitrobenzene at lOºC. The viscosity-gravity constant is
low for paraffins and high for naphthenes. (b) Behavior of a kerosene with 95% ethanol at 17ºC. The aniline point is low for aromatics and
naphthenes and high for paraffins. (e) Behavior of a dewaxed crude oil with liquid propane at 70ºF, with composition expressed in terms of
specific gravity.

M. The extract E and raffinate R are located on opposite ends of IMMISCIBLE SOLVENTS
the tieline that goes through M.
The distribution of a solute between two mutually immiscible
CROSSCURRENT EXTRACTION solvents can be represented by the simple equation,

In this process the feed and subsequently the raffinate are treated in Y = K' X, (14. 7)
successive stages with fresh solvent. The sketch is with Example
14.3. With a fixed overall amount of solvent the most efficient where
process is with equal solvent ftow to each stage. The solution of
Example 14.3 shows that crosscurrent two stage operation is X = mass of solute/mass of diluent,
superior to one stage with the same total amount of solvent. Y = mass of solute/mass of solvent.
 14.2. CALCULATION OF STAGE REOUIREMENTS 465

A B xc i n raff i nate

Figure 14.6. Construction of points on the distribution and operating curves: Line ab is a tieline. The dashed line is the tieline locus. Point e is
on the equilibrium distribution curve, obtained as the intersection of paths be and ade. Line Pfg is a random line from the difference point P
and intersecting the binodal curve in f and g. Point j is on the operating curve, obtained as the intersection of paths gj and fhj.

When K' is not truly constant, sorne kind of mean value may be (14.8)
applicable, for instance, a geometric mean, or the performance of
the extraction battery may be calculated stage by stage with a In terms of the extraction ratio,
different value of K' for each. The material balance around the first
stage where the raffinate leaves and the feed enters and an
intermediate stage k (as in Fig. 14.8, for instance) is A = K(E /R ), (14.9)

EXAMPLE 14.1 The last correlation is inferior for this particular example as the
The Equations for Tieline Data plots show.
The tieline data of the system of Example 14.1 are plotted
according to the groups of variables in the equations of Ishida,
XAA XcA XsA XAs Xcs Xss
Hand, and Othmer and Tobías with these results:
98.945
92.197 o.o 1.055 5.615 o.o 94.385
Ishida: y = l.OOxº·61 [Eq. (14.3)], 6.471 1.332 5.811 3.875 90.313
83.572 14.612 1.816 6.354 9.758 83.889
Hand: y = 0.078x 1'11 [Eq. (14.5)], 75.356 22.277 2.367 7.131 15.365 77.504
Othmer and Tobías: y = 0.88xº·90 [Eq. (14.4)]. 68.283 28.376 3.341 8.376 20.686 70.939
60.771 34.345 4.884 9.545 26.248 64.207
54.034 39.239 6.727 11.375 31.230 57.394
47.748 42.849 9.403 13.505 35.020 51.475
39.225 45.594 15.181 18.134 39.073 42.793

"--''
UJ

<1: 1o'xASxSA _1
__ 1 _1
__1
u XcA/xAA Xcs/Xss
HAND XAA
"
Vl
XAAXSS XcAXss Xss
'::::>
o 6.34 o 0.0107 0.0595 o o
a: 9.30 0.0088 0.0846 0.1073 0.070 0.043
<1:
> 16.46 0.0129 0.1966 0.1928 0.178 0.116
; 28.90 0.0211 0.3270 0.2903 0.296 0.198
o'
" 58.22 0.0343 0.4645 0.4097 0.416 0.292
--' ... 119.47 0.0581 0.6455 0.5575 0.565 0.409
OTHMER & 339.77 0.0933 0.8507 0.7423 0.726 0.544
TOBIAS 516.67 0.1493 1.0943 0.9427 0.897 0.680
1640 0.3040 1.5494 1.3368 1.162 0.913

LOG X, VARI OUS SCALES

 466 EXTRACTION ANO LEACHING

EXAMPLE 14.2 Calculated ratios for the Jlinecke coordinate plot of Figure 14.1:
Tabulated Tieline and Distribution Data for the System
A = 1-Hexene, B = Tetramethylene Sulfone, C = Benzene, Left Phase Right Phase
Represented in Figure 14.1
Experimental tieline data in mol %: B e B e
A+C A+C A+C A+C
Left Phase
O.Q108 o 16.809 o Right Phase
0.0135 0.0656 9.932 D.4000
0.0185 0.1488 5.190 0.6041
A e B A B 0.0248 0.2329 3.445 0.6830 e
98.945 o.o 1.055 5.615 94.385 0.0346 0.2936 2.441 0.7118 o.o
92.197 6.471 1.332 5.811 90.313 0.0513 0.3625 1.794 0.7333
3.875
83.572 14.612 1.816 6.354 83.888 0.0721 0.4207 1.347 0.7330
9.758
75.356 22.277 2.367 7.131 77.504 0.1038 0.4730 1.061 0.7217
15.365
68.283 28.376 3.341 8.376 70.938 0.1790 0.5375 0.748 0.6830
20.686
60.771 34.345 4.884 9.545 64.207 26.248
54.034 39.239 6.727 11.375 57.394 The x-y plot like that of Figure 14.6 may be made with the tieline
31.230
47.748 42.849 9.403 13.505 51.475 data of columns 5 and 2 expressed as fractions or by projection
35.020from
39.225 45.59415.181 18.134 42.793 the triangular diagram as shown. 39.073

the material balance becomes solution for the number of stages is

(14.10)
When these balances are made stage-by-stage and intermediate
compositions are eliminated, assuming constant A throughout, the
result relates the terminal compositions and the number of stages.
The expression for the fraction extracted is
XF - Xn An+t _ A
(14.11)
c/J = X p - Ys/ K = An+t _ ¡.
This is of the same form as the Kremser-Brown equation for gas
abso'rption and stripping and the Turner equation for leaching. The
(14.12)
When A is the only unknown, it may be found by tria! solution of
these equations, or the Kremser-Brown stripping chart may be
used. Example 14.4 applies these results.
14.3. COUNTERCURRENT OPERATION
In countercurrent operation of severa! stages in series, feed enters
the first stage and final extract leaves it, and fresh solvent enters the
last stage and final raffinate leaves it. Severa! representations of

xCEJ - - -- -
....
u '
,tt_J
.... \ oper a t ipg
'
)(
(]) L ine 1
e
<....>
)(

p xcRN
A RN s xc i n ra f f i na te
(a)

Figure 14.7. Locations of operating points P and Q for feasible, total, and minimum extract reftux on triangular diagrams, and stage
requirements determined on rectangular distribution diagrams. (a) Stages required with feasible extract reftux. (b) Operation at total reftux
and minimum number of stages. (e) Operation at minimum reftux and infinite stages.
 14.3. COUNTERCURRENT OPERATION 467

.....
u
L
ro.
.....
X
Q)

e
u
X

XCRN XCF
A xc i n raff i nate
(b)

.....
u
ro
L .
.....
X
Q)

e
u
X

Xc RN Xc F
A xc i n raffi nate
lc)
Figure 14.7-(continued )

such processes are in Figure 14.3. A flowsketch of the process The operation of finding the number of stages consists of a
together with nomenclature is shown with Example 14.5. The number of steps:
overall material balance is
l. Either the solvent feed ratio or the compositions E 1 and RN
(14.13) serve to locate the mix point M.
2. The operating point P is located as the intersection of lines FE 1
or and RNS.
3. When starting with Ev the raffinate R 1 is located at the other
(14.14) end of the tieline.
4. The line PR 1 is drawn to intersect the binodal curve in E 2 •
The intersection of extended lines FE 1 and RNS locates the
operating point P. The material balance from stage 1 through k is The process is continued with the succeeding values R 2 , E 3 ,
R 3 , E4 , • • . until the final raffinate composition is reached.
(14.15) When number of stages and only one of the terminal
compositions are fixed, the other terminal composition is selected
or by tria! until the stepwise calculation finds the prescribed number of
stages. Example 14.6 applies this kind of calculation to find the
(14.16) stage requirements for systems with Types I and II equilibria.
Evaluation of the numbers of stages also can be made on
Accordingly, the raffinate from a particular stage and the extract rectangular distribution diagrams, with a McCabe-Thiele kind of
from a succeeding one are on a line through the operating point P. construction. Example 14.5 does this. The Janecke coordinate plots
Raffinate Rk and extract Ek streams from the same stage are located like those of Figures 14.1 and 14.2 also are convenient when many
at opposite ends of the same tieline. stages are needed, since then the triangular construction may
 468 EXTRACT\ON ANO LEACH\NG

EXAMPLE 14.3 balance are

Single Stage and Cross Current Extraction of Acetic Acid from
Methylisoutyl Ketone with Water E R
The original mixture contains 35% acetic acid and 65% MIBK. It is
charged at 100 kg/hr and extracted with water. Acetic acid 0.185 0.16
MIBK 0.035 0.751
Water 0.78 0.089
a. In a single stage extractor water is mixed in at 100 kg/hr. On the kg/hr 120 80
triangular diagram, mix point M is midway between F and S.
Extract and raffina\.e compositions are on the tieline through M. b. The flowsketch of the crosscurrent process is shown. Feed to the
Results read off the diagram and calculated with material first stage and water to both stages are at 100 kg/hr. The extract
and raffinate compositions are on the tielines passing through
mix points M1 and M2 • Point M is for one stage with the same
total amount of solvent. Two stage results are:
Acetic acid
E, R, E¿ R2

..
"O
'ü Acetic acid 0.185 0.160 0.077 0.058
· F MIBK 0.035 0.751 0.018 0.902
.".'
e:
u Water
kg/hr
0.780
120
0.089
80
0.905
113.4
0.040
66.6
o
·
u
!!

o
s F, XF
·1
Ketone
0.5
Mass fraction water
1.0
Water ...
E, y S, Ys

(Acetic acid)
0.5 ..-
---
---
--

..
? "?
"O
·¡ ;
u F
·u¡;
..
e:
R2
o
't 2
..
!!: ::

f
1
::;: R2

o
(S , S
1 2
)

E,
f
s,
0.5 1.0 E, s2
Ketone Mass fraction water Water

become crowded and difficult to execute accurately unless a very
large scale is adopted. The Janecke method was developed by
Maloney and Schubert [Trans. AIChE 36, 741 (1940)]. Severa)
detailed examples of this kind of calculation are worked by Treybal
(1963), Oliver ( Diffusional Separation Processes, Wiley, New York,
1966), and Laddha and Degaleesan (1978).
MIN\MUM SOLVENT/FEED RATIO
Both maximum and minimum limits exist of the solvent/feed ratio.
The maximum is the value that locates the mix point M on the
binodal curve near the solvent vertex, such as point Mmax on Figure
14.7(b). When an operating line coincides with a tieline, the
number of stages will be infinite and will correspond to the
mínimum solvent/feed ratio. The pinch point is deterrnined by the
intersection of sorne tieline with Jine RNS. Depending on whether
the slopes of the tielines are negative or positive, the intersection
that is closest or farthest from the solvent vertex locates the
operating point for minimum solvent. Figure 14.9 shows the two
cases. Frequently, the tieline through the feed point determines the
mínimum solvent quantity, but not for the two cases shown.
EXTRACT REFLUX
Normally, the concentration of solute in the final extract is limited
to the value in equilibrium with the feed, but a countercurrent
stream that is richer than the feed is available for enrichment of the
extract. This is essentially solvent-free extract as reflux. A
flowsketch and nomenclature of such a process are given with
Example 14.7. Now there are two operating points, one for above
the feed and one for below. These points are Jocated by the
following procedure:
l. The mix point is located by establishing the solvent/feed ratio.
2. Point Q is at the intersection of lines RNM and E 1 SE> where SE
refers to the solvent that is removed from the final extract, and
may or may not be of the same composition as the fresh solvent
S. Depending on the shape of the curve, point Q may be inside
 14.3. COUNTERCURRENT OPERATION 469
EXTRACT the binodal curve as in Example 14.7, or outside as in Figure
14.7.
3. Point P is at the intersection of lines RNM and E 1SE , where SE
refers to the solvent removed from the extract and may or may
not be the same composition as the fresh solvent S.

Determination of the stages uses Q as the operating point until

the raffinate composition Rk falls below line FQ. Then the
operation is continued with operating point P until RN is reached.

MINIMUM REFLUX
For a given extract composition E 1, a pinch point develops when an
operating line through either P or Q coincides with a tieline.
Frequently, the tieline that passes through the feed point F
determines the reflux ratio, but not on Figure 14.7(c). The tieline
that intersects line FSE nearest point s. locates the operating point
Qm for minimum reflux. In Figure 14.7(c), intersection with tieline
Fcde is further away from point SE than that with tieline abQm,
which is the one that locates the operating point for minimum reflux
in this case.

RAFFINATE
Figure 14.8. Model for liquid-liquid extraction. Subscript i refers to
a component: i = 1, 2, . . . , c. In the commonest case, F1 is the only
feed stream and FN is the solvent, or Fk may be a reflux stream.
Withdrawal streams Uk can be provided at any stage; they are not
incorporated in the material balances written here.
MINIMUM STAGES
As the solvent/feed ratio is increased, the mix point M approaches
the solvent point S, and poles P and Q likewise do so. At total
reflux ali of the points P, Q, S, SE> and M coincide; this is shown in
Figure 14.7(b).
Examples of triangular and McCabe-Thiele constructions for
feasible, total, and minimum reflux are shown in Figure 14.7.

EXAMPLE 14.4
Extraction with an Immiscible Solvent E, , Y,
A feed containing 30 wt % of propionic acid and 70 wt %
trichlorethylene is to be extracted with water. Equilibrium
distribution of the acid between water (Y) and TCE (X) is
represented by Y = K' X, with K' = 0.38. Section 14.3 is used.

a. The ratio E / R of water to TCE needed to recover 95% of the R 1, x 1

acid in four countercurrent stages will be found:

XF = 30/70,
Xn = 1.5/70,
Y5 = 0,
5
</> = (30 -1.5)/(30 - O) = 0.95 = (A5 - A/(A -1).

By tria!,

A = 1.734,
E / R =A/ K' = 1.734/0.38 = 4.563.
Em+ 2 1 Ym + 2

b. The number of stages needed to recover 95% of the acid with

E / R = 3.5 is found with Eq. 14.12.

A = K' E / R = 0.38(3.5) = 1.330, </> = 0.95,

n = _ 1+ ln[(A - <J>)/(1- </>)] -1
ln A
+ ln[(l.330 - 0.95)/(1- 0.95)]/ln(l.330) = 6.11
470 EXTRACTION AND LEACHING
EXAMPLE 14.5
Countercurrent Extraction Represented on Triangular aud
Rectangular Distribution Diagrams
The specified feed F and the desired extract E 1 and raffinate RN
compositions are shown. The solvent/feed ratio is in the ratio of the
line segments MS / MF, where the location of point M is shown as
the intersection of lines E1RN and FS.
Phase equilibrium is represented by the tieline locus. The
equilibrium distribution curve is constructed as the locus of
intersections of horizontal lines drawn from the right-hand end of a
---- F, xF
R1, x 1
p
Naturally, the latter constructions are analogous to those for
distillation since their forms of equilibrium and material balances
are the same. References to the literature where similar calculations
are performed with Janecke coordinates were given earlier in this
section.
Use of reflux is most effective with Type IIsystems since then
essentially pure products on a solvent-free basis can be made. In
contrast to distillation, however, extraction with reftux rarely is
beneficia!, and few if any practica! examples are known. A related
kind of process employs a second solvent to wash the extract
countercurrently. The requirements for this solvent are that it be
only slighly soluble in the extract and easíly removable from the
extract and raffinate. The sulfolane process is of this type; it is
described, for example, by Treybal (1980) and in more detail by Lo
et al. (1983, pp. 541-545).
tieline with horizontals from the left-hand end of the tielines and
reflected from the 45º line.
The operating curve is drawn similarly with horizontal
projections from pairs of random points of intersection of the
binodal curve by lines drawn through the difference point P.
Construction of these curves also is explained with Figure 14.6.
The rectangular construction shows that slightly less than eight
stages are needed and the triangular that slightly more than eight
are needed. A larger scale and greater care in construction could
bring these results closer together.
E,, y,
X0, in ralfinate
14.4. LEACHING OF SOLIOS
Leaching is the removal of solutes from admixture with a solid by
contracting it with a solvent. The solution phase sometimes is called
the overflow, but here it will be called extract. The term underflow
or raffinate is applied to the solid phase plus its entrained or
occluded solution.
Equilibrium relations in leaching usually are simpler than in
liquid-liquid equilibria, or perhaps only appear so because few
measurements have been published. The solution phase normally
contains no entrained solids so its composition appears on the
hypotenuse of a triangular diagram like that of Example 14.8. Data
for the raffinate phase may be measured as the holdup of solution
by the solid, K lb solution/lb dry (oíl-free) solid, as a function of the
concentration of the solution, y lb oíl/lb solution. The correspond-
 14.4. LEACHING OF SOLIOS 471

EXAMPLE 14.6 The material balance on heptane is

Stage Requirements Cor the Separation of a Type 1and a Type
11 System
a. The system with A = heptane, B = tetramethylene sulfone, and 40 = 0.6E + 0.05(100 - E),
C = toluene at 50ºC [Triparthi, Ram, and Bhimeshwara, J.
Chem. Eng. Data 20, 261 (1975)]: The feed contains 40% C, the whence E = 63.6 lb/100 lb feed, and the TMS/feed ratio is
extract 70% C on a TMS-free basis or 60% overall, and raffinate
5% C. The construction shows that slightly more than two
equilibrium stages are needed for this separation. The 0.13(63.6) + 0.93(36.4) = 42 lb/100 lb feed.
compositions of the streams are read off the diagram:
b. The type 11 system with A = octane, B = nitroethane, and
Feed Extract Raffinate C = 2,2,4-trimethylpentane at 25ºC [Hwa, Techo, and Ziegler, J.
Heptane 60 27 2 Chem. Eng. Data 8, 409 (1963)]: The feed contains 40% TMP,
TMS o 13 93 the extract 60% TMP, and the raffinate 5% TMP. Again, slightly
Toluene 40 60 5 more than two stages are adequate.

C E'
1

e E,
70
60
(.)
Tieline
Locus
o
::E

20
RN 10 E2
R3 o E3 ----
o 20 40 60 100
A Mol% B B A Mol % B B

ing weight fraction of oil in the raffinate or underflow is
x = Ky /( K + 1). (14.17)
Since the raffinate is a mixture of the solution and dry solid, the
equilibrium value in the raffinate is on the line connecting the origin
with the corresponding solution composition y, at the value of x
given by Eq. (14.17). Such a raffinate line is constructed in Example
14.8.
Material balance in countercurrent leaching still is represented
by Eqs. (14.14) and (14.16). Compositions Rk and Ek+l are on a
line through the operating point P, which is at the intersection of

p
m
B

(a) (b)

Figure 14.9. Minimum solvent amount and maximum extract concentration. Determined by location of the intersection of extended tielines
with extended line RNS. (a) When the tielines slope down to the left, the furthest intersection is the correct one. (b) When the tielines slope
down to the right, the nearest intersection is the correct one. At maximum solvent amount, the mix point Mm is on the binodal curve.
472 EXTRACTION AND LEACHING

EXAMPLE 14.7 made with operating point P, which is the intersection of lines FQ
Countercurrent Extraction Employing Extract Reftux and SRN.
The feed F, extract E 1, and raffinate RN are located on the In this example, only one stage is needed above the feed F and
triangular diagram. The ratio of solvent/feed is specified by the five to six stages below the feed. The ratio of solvent to feed is
location of the point M on line SF.
Other nomenclature is identified on the flowsketch. The
S / F = FM / MS = 0.196,
solvent-free reflux point R 0 is located on the extension of line SE 1 .
Operating point Q is located at the intersection of lines SR 0 and
RNM. Lines through Q intersect the binodal curve in compositions and the extemal reflux ratio is
of raffinate and reflux related by material balance: for instance, Rn
and En+t· When the line QF is crossed, further constructions are

Solvent
Removal

SE
e Solvent
Reflux -- --- E1P
Ro E1 Product

F\-1 f F
leed
Rt l:i.1

s
Flowsketch and triangular diagram construction with extract reflux.

EXAMPLE 14.8 and sorne calculated values are

Leaching of an Oil·Bearing Solid in a Countercurrent Battery
Oil is to be leached from granulated halibut livers with pure ether as y o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
solvent. Content of oil in the feed is 0.32 lb/lb dry (oil-free) solids X 0 0.0174 0.0397 0.0694 0.1080 0.1565 0.2147 0.2821 0.3578
and 95% is to be recovered. The economic upper limit to extract
concentration is 70% oil. Ravenscroft [Ind. Eng. Chem. 28, 851 Points on the raffinate line of the triangular diagram are located on
(1934)] measured the relation between the concentration of oil in lines connecting values of y on the hypotenuse (solids-free) with the
the solution, y, and the entrainment or occlusion of solution by the origin, at the values of x and corresponding y from the preceding
solid phase, K lb solution/lb dry solid, which is represented by the tabulation.
equation
Feed composition is xF = 0.32/1.32 = 0.2424.
2 Oil content of extract is y1= O.7.
K = 0.19 + 0.126y + 0.810y .
Oil content of solvent is Ys = O.
Amount of oil in the raffinate is 0.32(0.05) = 0.016 lb/lb dry,
The oil content in the entrained solution then is given by and the corresponding entrainment ratio is

x = K/(K + I)y, wt fraction, KN = 0.016/yN = 0.19 + 0.126yN + 0.81y1.

 14.5. NUMERICAL CALCULATION OF MULTICOMPONENT EXTRACTION 473

EXAMPLE 14.8---( continued ) are located as intersections of random lines through P with these
results:
Solving by trial,
YN = 0.0781, y o 0.1 0.2 0.3 0.4 0.5 0.6 0.7
X 0.013 0.043 0.079 0.120 0.171 0.229 0.295 0.368
KN =0.2049,
xN = 0.0133 (final raffinate composition).
The McCabe-Thiele construction also shows that six stages are
The operating point P is at the intersection of lines FE 1 and SRN. needed.
The triangular diagram construction shows that six stages are Point P is at the intersection of lines E 1F and SRN. Equilibrium
needed. compositions are related on lines through the origin, point A.
The equilibrium line of the rectangular diagram is constructed Material balance compositions are related on lines through the
with the preceding tabulation. Points on the material balance line operating point P.

o.a

...J
.......
GI
111
.....
o
ra
o z:.
%
...a.u
1- ra
u
<
o::
.""..
..... )(

>-

x ( raff i nate )

lines FE 1 and SRN. Similarly, equilibrium compositions Rk and Ek
are on a line through the origin. Example 14.8 evaluates stage
requirements with both triangular diagram and McCabe-Thiele
constructions. The mode of construction of the McCabe-Thiele
diagram is described there.
These calculations are of equilibrium stages. The assumption is
made that the oil retained by the solids appears only as entrained
solution of the same composition as the bulk of the liquid phase. In
sorne cases the salute may be adsorbed or retained within the
interstices of the salid as solution of different concentrations. Such
deviations from the kind of equilibrium assumed will result in stage
efficiencies less than 100% and must be found experimentally.
14.5. NUMERICAL CALCULATION OF
MULTICOMPONENT EXTRACTION
Extraction calculations involving more than three components
cannot be done graphically but must be done by numerical solution
of equations representing the phase equilibria and material balances
over ali the stages. Since extraction processes usually are adiabatic
and nearly isothermal, enthalpy balances need not be made. The
solution of the resulting set of equations and of the prior
determination of the parameters of activity coefficient correlations
requires computer implementation. Once such programs have been
developed, they also may be advantageous far ternary extractions,
474 EXTRACTION ANO LEACHJNG

particularly when the number of stages is large or several cases must
be worked out. Ternary graphical calculations also could be done
on a computer screen with a little effort and sorne available
software.
The notation to be used in making material balances is shown
on Figure 14.8. For generality, a feed stream Fk is shown at every
stage, and a withdrawal stream Uk also could be shown but is not
incorporated in the balances written here. The first of the double
subscripts identifies the component i and the second the stage
number k ; a single subscript refers to a stage.
For each component, the condition of equilibrium is that its
activity is the same in every phase in contact. In terms of activity
coefficients and concentrations, this condition on stage k is written:
YiJch = Yftx;k (14.18)
or
Y;k = K;kXik' (14.19)
where
K;k = y fl f y:Jc (14.20)
is the distribution ratio. The activity coefficients are functions of the
temperature and the composition of their respective phases:
y:Jc = f ( Tk , Y1k' Yzk, · ·· , Yck), (14.21)
Y =f ( Tk, X1k' X zk , · · · ' Xck ). (14.22)
The most useful relations of this type are the NRTL and
UNIQUAC which are shown in Table 14.1.
Around the kth stage, the material balance is
(14.23)
When combined with Eq. (14.19), the material balance becomes
Rk-1X;.k-1 - (Rk + Ek K;k )x;k + Ek+1K;,k+1X;,k+1 = -Fk z;k·
(14.24)
In the top stage, k = 1 and R0 = O so that
When ali of the coefficients are known, this can be solved for the
concentrations of component i in every stage. A straightforward
method for solving a tridiagonal matrix is known as the Thomas
algorithm to which references are made in Sec. 13.10, "Basis for
Computer Evaluation of Multicomponent Separations: Specifi-
cations."
INITIAL ESTIMATES
Solution of the equations is a process in which the coefficients of
Eq. (14.28) are iteratively improved. To start, estimates must be
made of the flow rates of ali components in every stage. One
procedure is to assume complete removal of a "light" key into the
extract and of the "heavy" key into the raffinate, and to keep the
solvent in the extract phase throughout the system. The distribution
of the keys in the intermediate stages is assumed to vary linearly,
and they must be made consistent with the overall balance, Eq.
(14.27), for each component. With these estimated flowrates, the
values of X;k and Y;k are evaluated and may be used to find the
activity coefficients and distribution ratios, K;k· This procedure is
used in Example 14.9.
PROCEDURE
The iterative calculation procedure is outlined in Figure 14.10. The
method is an adaptation to extraction by Tsuboka and Katayama
(1976) of the distillation calculation procedure of Wang and Henke
[Hydrocarb. Proc. 45(8), 155-163 (1967)). It is also presented by
Henley and Seader (1981, pp. 586-594).
l. The initial values of the flowrates and compositions x;k and Y;k
are estimated as explained earlier.
2. The values of activity coefficients and distribution ratios are
evaluated.
3. The coefficients in the tridiagonal matrix are evaluated from
Eqs. (14.24)-(14.26). The matrix is solved once for each
component.
4. The computed values of iteration (r + 1) are compared with
those of the preceding iteration as
C N
-r1= L 2: jxfk+l) - xi,;:>¡ :s E 1 = O.OlNC. (14.29)
i=1 k=l

(14.25)
The magnitude, O.OlNC, of the convergence criterion is
In the bottom stage, k = N and EN+l = O so that arbitrary.
S. For succeeding evaluations of activity coefficients, the values of
the mol fractions are normalized as
(14.26)

The overall balance from stage 1 through stage k is (x;k)nonnalized = X;k / .f X;k,
i=l
(14.30)
k
Rk = Ek+1 - El+¿ Fk , (14.27) (y;k)normalized = Yik / :f Y;k•
1 i=l

which is used to find raffinate flows when values of the extract fiows 6. When the values of x,k have converged, a new set of y,k is
have been estimated. calculated with
For ali stages for a component i, Eqs. (14.24)-(14.26)
constitute a tridiagonal matrix which is written (14.19)

C1 7. A new set of extract flow rates is calculated from

¡··
Az B2 C2
Ai Bi
AN-1
(14.28) Ek - k L.,,, ik'
e
(s+1) _ E(s ) .._, y
i=l
(14.31)

where s is the outer loop index number.

 14.5. NUMERICAL CALCULATION OF MULTICOMPONENT EXTRACTION 475

TABLE 14.1. NRTL and UNIQUAC Correlations for Activity Coefficients of Three-
Component Mixturesª

NRTL

•¡¡= o
G¡¡ = 1

UNIQUAC
e/>¡ 9¡ e/>¡
In'Y;= 1n-+ Sq¡ 1n- + I ¡ - -(x 111 + x 212 + x3/ 3 ) + q¡( l - ln(9¡<¡¡ + 92•u + 93<3;))
X¡ e/>¡ X¡

t"¡¡ = 1
r¡x¡

I¡ = S( r¡ - q¡) - r ; +1

ª NRTL equation: There is a pair of parameters 9;k and 9k; for each pair of substances in the
mixture; for three substances, there are three pairs. The other terms of the equations are related
to the basic ones by

r k = 9;k / RT,
G¡k = exp(-<r kr k).

For liquid-liquid systems usually, a-¡k = 0.4.

UNIOUAC equation: There is a pair of parameters u¡k and uki for each pair of substances in
the mixture:

The terms with single subscripts are properties of the pure materials which are usually known or
can be estimated.
The equations are extended readily to more components.
(See, for example, Walas, Phase Equilibria in Chemica/ Engineering, Butterworths, 1985).

8. The criterion for convergence is 9. If convergence has not been attained, new values of Rk are
N calculated from Eq. (14.27).
<2 = 2: (1- Ei:l / Ei:+ >)2
1
st:2 = O.OlN. (14.32) 10. Distribution ratios K;k are based on normalized values of x;k
k-1 and Y;k·
The magnitude, O.OlN, of the convergence criterion is 11. The iteration process continues through the inner and outer
arbitrary. loops.
476 EXTRACTION AND LEACHING

EXAMPLE 14.9 TABLE 2. Estimated Mol Fractions

Trial Estimates and Converged Flow Rates and Compositions
in Ali Stages of an Extraction Battery for a Four-Component
Y1, X,1
Mixture Stage
Benzene is to be recovered from a mixture with hexane using j H B D w H B D w
aqueous dimethylformamide as solvent in a five-stage extraction
battery. Tria! estimates of flow rates for starting a numerical
1 o.o 0.0909 0.6818 0.2273 0.7895 0.2105 o.o o.o
2 o.o 0.0741 0.6944 0.2315 0.8333 0.1667 o.o o.o
solution are made by first assuming that ali of the benzene and ali of 3 o.o 0.0566 0.7076 0.2359 0.8824 0.1176 o.o o.o
the solvent ultimately appear in the extract and ali of the hexane 4 o.o 0.0385 0.7211 0.2404 0.9375 0.0625 o.o o.o
appears in the raffinate. Then flow rates throughout the battery are 5 o.o 0.0196 0.7353 0.2451 1.0000 o.o o.o o.o
assumed to vary Jinearly with stage number. Table 1 shows these
estimated flowrates and Table 2 shows the corresponding mol
fractions. Tables 3 and 4 shows the converged solution made by TABLE 3. Converged Mol Fractions
Henley and Seader (1981, p. 592); they do not give any details of
Yi1 X,1
the solution but the algorithm of Figure 14.10 was followed. Stge
J H B D w H B D w
1 0.0263 0.0866 0.6626 0.2245 0.7586 0.1628 0.0777 0.0009
2 0.0238 0.0545 0.6952 0.2265 0.8326 0.1035 0.0633 0.0006
3 0.0213 0.0309 0.7131 0.2347 0.8858 0.0606 0.0532 0.0004
4 0.0198 0.0157 0.7246 0.2399 0.9211 0.0315 0.0471 0.0003
TABLE 1. Estimated mol/hr 5 0.0190 0.0062 0.7316 0.2432 0.9438 0.0125 0.0434 0.0003

Extraet Raffinate
TABLE 4. Converged mol/hr
Stage Total H B D w Total H B D w
Extraet Raffinate
o
1 1100 o 100 750 250 400 300 100 o o Hexane 29.3 270.7
2 1080 o 80 750 250 380 300 80 o o Benzene 96.4 3.6
3 1060 o 60 750 250 360 300 60 o o DMF 737.5 12.5
4 1040 o 40 750 250 340 300 40 o o Water 249.0 0.1
5 1020 o 20 750 250 320 300 20 o o
N+1 1000 o o 750 250 300 300 o o o Total 1113.1 286.9

Solutions of four cases of three- and four-component systems
are presented by Tsuboka and Katayama (1976); the number of
outer loop iterations ranged from 7 to 41. The four component case
worked out by Henley and Seader (1981) is summarized in Example
14.9; they solved two cases with different water contents of the
solvent, dimethylformamide.
14.6. EQUIPMENT FOR EXTRACTION
Equipment for extraction and leaching must be capable of providing
intimate contact between two phases so as to effect transfer of
solute between them and also of ultimately effecting a complete
separation of the phases. For so general an operation, naturally a
substantial variety of equipment has been devised. A very general
classification of equipment, their main characteristics and industrial
applications is in Table 14.2. A detailed table of comparisons and
ratings of 20 kinds of equipment on 14 characteristics has been
prepared by Pratt and Hanson (in Lo et al., 1983, p. 476). Sorne
comparisons of required sizes and costs are in Table 14.3.
Selected examples of the main categories of extractors are
represented in Figures 14.11-14.15. Their capacities and perfor-
mance will be described in general terms insofar as possible, but
sizing of Jiquid-liquid extraction equipment always requires sorne
pilot plant data or acquaintance with analogous cases. Little
detailed information about such analogous situations appears in the
open literature. Engineers familiar with particular kinds of
equipment, such as their manufacturers, usually can predict
performance with a minimum amount of pilot plant data.
Literature data is almost entirely for small equipment whose
capacity and efficiency cannot be scaled up to commercial sizes,
although it is of qualitative value. Extraction processes are sensitive
because they operate with small density differences that are
sensitive to temperature and the amount of solute transfer. They
also are affected by interfacial tensions, the large changes in phase
flow rates that commonly occur, and even by the direction of mass
transfer. For comparison, none of these factors is of major
significance in vapor-liquid contacting.
CHOICE OF DISPERSE PHASE
Customarily the phase with the highest volumetric rate is dispersed
since a Jarger inteñacial area results in this way with a given droplet
size. In equipment that is subject to backmixing, such as spray and
packed towers but not sieve tray towers, the disperse phase is made
the one with the smaller volumetric rate. When a substantial
difference in resistances of extract and raffinate films to mass
transfer exists, the high phase resistance should be compensated for
with increased suñace by dispersion. From this point of view,
Laddha and Degaleesan (1978, pp. 194) point out that water should
be the dispersed phase in the system water + diethylamine +
toluene. The dispersed phase should be the one that wets the
material of construction less well. Since the holdup of continuous
phase usually is greater, the phase that is Jess hazardous or less
expensive should be continuous. It is best usually to disperse a
highly viscous phase.