APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID

EXTRACTION PROCESSES: PURIFICATION OF

BIOLOGICALS, BOTANICALS, AND STRATEGIC METALS
1. Introduction

Liquid-liquid extraction (LLE) or solvent extraction (SX) is the selective separation

of substances from a liquid mixture using combinations of non-miscible or almost
non-miscible liquids. One of the major advantages is the high throughput while
energy consumption is kept to a minimum. Therefore, it represents a viable
alternative to distillation and is used for separation of temperature-sensitive
substances, which are mostly of biological origin or metal ions. A Further advan-
tage, apart from the lower process temperature, is the direct control of selectivity
by selecting the extraction agent. In this way separation of small contents of high-
boiling impurities, mainly in aqueous solutions, is possible, whereas in the usual
thermal separation technique (like distillation), the complete water content has to
be withdrawn by a highly energy-intensive evaporation process. The extractant
typically must be reprocessed in an additional process step so that it can be reused,
which often results in more steps for regeneration and scrubbing than for the
actual extraction process itself as can be seen in Fig. 1 (1, 2).
Although the energy consumption of the actual extraction process is usually
negligibly low, the steps of solvent regeneration and potential re-extraction must
be taken into account for the complete assessment of resource and energy
efficiency.
Hence, the basics of liquid-liquid extraction are discussed in this chapter.
This is followed by an introduction and explanation of the process variants and
operating methods as well as a description of the individual extraction appara-
tuses. Subsequently, the criteria of solvent selection and the phenomena of mass
transport will be discussed. Finally, the experiments to determine the parameters
necessary for modeling and description of liquid-liquid extraction as well as
application examples are presented.

2. Fundamentals

2.1. Phase Diagrams. The basic requirement for liquid-liquid extraction

is the presence of a two-phase area between raffinate and extract phase. Ideally,
this mixing gap is as large as possible (2, 3).
Two liquid phases always occur when the activity coefficients reach high
values, which is given by a strong positive deviation from Raoult's law. Due to the
different progression of the activity coefficients as a function of composition and
temperature, the solubility behavior also differs with temperature (3).
This is shown for the binary case and constant pressure in the Fig. 2 in form of
three different temperature-composition diagrams, based on the so-called binodal
curves. In most cases, the solubility increases with increasing temperature until
the homogeneous liquid phase above the upper critical demixing temperature is
reached (Fig. 2, mid). Seldom the other two conditions might occur, which are also
displayed in Fig. 2. On the left side, the mutual solubility increases with

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright  2018 John Wiley & Sons, Inc. All rights reserved.
DOI: 10.1002/0471238961.koe00041
 2 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Feed Scrub Product

(TC, SCs, IPs) Solution (TC)

Raffinate Extraction Scrubbing Stripping Strip

Scrub
(SCs, IPs) solution
Raffinate
SCs, IPs SCs, IPs TC

TC, SCs, IPs TC, IPs IPs

Solvent Extract Extract Solvent
(Regenerated) (scrubbed) (Stripped)
Regeneration

IPs

Aqueous phase TC: Target component

Organic phase SCs: Side components
IPs: Impurities

Fig. 1. General flow sheet for liquid–liquid extraction processes.

decreasing temperature until the two-phase area completely disappears

below the lower critical demixing temperature. Sometimes even both critical
demixing temperatures (Fig. 2, right) emerge. Additionally, a fourth case that
the mixing gap extends over the entire temperature range might occur (not
shown in Fig. 2). Those cases can be regarded as unusual cases of the ones shown
in Fig. 2, mid.
For the ternary case, the two most frequent curve progressions are discussed
in the form of ternary diagrams. Fig. 3 shows that there is a gap between the
carrier A and the solvent B. This gap becomes smaller with increasing amount of

Two-liquid phases One-liquid

M
One-liquid M
phase phase
Temperature

Temperature

Temperature

S Two-liquid phases
D G

M One-liquid
One-liquid phase K Two-liquid phases L Mʹ
K phase
100%A 100%B 100%A 100%B 100%A 100%B
Composition Composition Composition
0 1.0

Fig. 2. Temperature dependence of binary liquid–liquid equilibria. (Reproduced with permission

from Ref. (4). Copyright 1963, John Wiley & Sons.).
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 3

C
W
Auxiliary line

Raffinate Extract
C.P
.

A B
WB Tie lines

Fig. 3. General overview of ternary phase diagrams. (Reproduced with permission from
Ref. (2). Copyright 2005, Springer.)

the target component C until it disappears. The single-phase area is separated

from the two-phase area by the binodal curve. Mixtures of substances within the
mixing gap separate along the tie line into two phases. The intersection points of
the tie line with the binodal curve represent the compositions of the extract
and raffinate phase, which are in equilibrium. As the distance to the critical point
(C. P.) decreases, both the density difference and the interfacial tension decrease
until they finally become zero at the critical point. Due to low interfacial tension
close to this point, the extraction cannot be carried out here. This is based on the
decreasing size of droplets, which leads to a condition where the phases cannot
longer be separated (2, 5).
In one case (closed system), observed in about 75% of systems, only one binary
pair has a mixing gap. In these systems, a critical point occurs where the two liquid
phases have the same concentration (3). The other case occurs when in a system
two binary pairs are only partially miscible (open system). This behavior occurs in
about 20% of all cases (3). The temperature dependence of the liquid–liquid
equilibria of ternary systems can be very different from those of binary systems.
The most common behavior for a closed system is shown in Figure 4. In this case,
the solubility increases with increasing temperature, ie, the concentration range in
which two liquid phases coexist becomes smaller and smaller until the heteroge-
neous area disappears from the upper critical separation temperature of the
binary system AB (3).
2.2. Operating Modes. In practice, there are several variants of how a
liquid–liquid extraction process is used. On the one hand it is applied as simple
single-stage extraction, which is commonly used on a laboratory scale. On the other
hand it may be used as a multi-stage extraction, where the phases are conducted
either in crossflow or countercurrent configuration. These three variants are
briefly introduced and explained in the following sections with an emphasis on
the equilibrium representation in ternary diagrams.
4 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

A B

P5
C
P4
t4
Temperature

P3
t3
t1
P2 t2
t2 t3 P1
P
P3 2
C t4
P4
P1
A B
t1 A P5 B

Fig. 4. Qualitative progression of temperature dependence of ternary liquid–liquid equi-

libria. (Reproduced with permission from Ref. (4). Copyright 1963, John Wiley & Sons.)

Single Stage LLE. In single-stage extraction, the feed F_ ˆ R_ 0 mixture

consisting of a carrier A and an ingredient/transition component C is first of all
intensively mixed with the solvent E_ 0 in the mixing zone (Fig. 5).
The aim is to create a phase interface as large as possible, which is a
prerequisite for an elevated level of mass transfer, and thus enables the equili-
brium of distribution to be reached or be achieved rapidly. The composition is set at
the mixing point M _ . Since this point is in the two-phase area, the phases are
separated into two liquid phases after mixing has been completed. These are

∙ ∙ C
E0 E1

∙ ∙
F R1
C
W

C.P
∙
F
∙ ∙
M E1
∙
R1

∙
A E0 B
WB

Fig. 5. Construction of single-stage extraction in ternary diagram. (Reproduced with

permission from Ref. (2). Copyright 2005, Springer.)
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 5

located on the tie line crossing the mixing point at the intersection with the
binodal. The extract phase E_ 1 consists of the transition component C and the
solvent B, whereas the resulting refining phase R1 consists of the remaining
transition component and the carrier A. (2,5).
The mass balance of the single-stage extraction is as follows:

F_ ‡ E_ 0 ˆ M
_ 1ˆR
_ 1 ‡ E_ 1 (1)

The mass balance equation of the transition component can be used to determine
the position of the mixing point.

F_
xF ‡ E_ 0
y0 ˆ M
_ 1
z1 ˆ R
_ 1
x1 ‡ E_ 1
y
1 (2)

The composition of the mixing point can be determined by rearrangement to z1 .

Alternatively, the quantities of the resulting raffinate and extract phase can also
be determined by the lever rule (2,5).
Crosscurrent LLE. In crosscurrent extraction, the raffinate phase R_ (in the
first stage F_ ) is mixed with fresh or regenerated solvent in each stage, while the
combined extract phase is recycled (Fig. 6). It is assumed that the equilibrium is
reached at each stage.
The first mixing point M _ 1 is obtained by the ratio of F_ =E_ 0 . As described
already, this mixture separates into the extract phase and raffinate phase. The
resulting raffinate phase is then mixed again with fresh solvent and the mixing
point M 2 is formed. This is repeated until the desired depletion of the raffinate
phase has been achieved. The number of theoretical separation stages required for
the crosscurrent cascade then corresponds to the number of state points in the
raffinate or extract phase. The extract, on the other hand, is less and less
concentrated with increasing number of steps (2,5).

∙ C
F
∙ ∙
E 01 1 E1
∙
R1
∙ ∙
E 02 2 E2
C
W

∙ C.P
R2
∙ ∙
∙ ∙ F E1 ∙
E 03 3 E3 ∙
∙ M E2
∙ ∙ 1 ∙
R3 ∙ R1 M E3
∙ R2 ∙2 ∙
∙ ∙ R M3 E4
E 04 4 E4 ∙ 3 ∙
R4 M4 ∙
∙ A E0 B
R4
WB

Fig. 6. Construction of cross-current extraction in ternary diagram. (Reproduced with

permission from Ref. (2). Copyright 2005, Springer.)
6 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

∙ ∙ C
F E1
1
∙ ∙
R1 E2

C
W
∙ ∙ C.P
R2 E3 ∙
∙ E1
F ∙ ∙
3 E2
R ∙
∙ ∙ ∙ 1 M ∙
E4 R2 E3
π R3 ∙
R3 ∙
E4
4 ∙
∙ ∙ A E0 B
R4 E0 WB

Fig. 7. Construction of countercurrent extraction in ternary diagram.(Reproduced with

permission from Ref. (2). Copyright 2005, Springer.)

Countercurrent LLE. The principle of countercurrent four-stage solvent

extraction is shown in Figure 7.
The feed mixture F is depleted from the initial concentration xF to the final
_ n (here: R
concentration xn (here: x4 ) of the raffinate stream R _ 4 ). The concentration
_
of the solvent stream E 0 increases from the concentration y0 to y1 , whereas for the
pure solvent E_ 0 is exactly on the triangular tip B. Analogous to equation 1, the
miscibility point of the entire process can be determined via the mass balance.

F_ ‡E_ 0 ˆ M
_ ˆR
_ n ‡ E_ 1 (3)

The location of the mixing point is determined again by the mass balance:

F_
xF ‡ E_ 0
y0 ˆ M
_
z ˆ R
_ n
xn ‡ E_ 1
y
1 (4)

The position of the mixing point is thus determined from the ratio of the feed
stream to the solvent stream. If the refining stream R _ n is now connected to the
mixing point via a straight line, the position of the extract phase E_ 1 can be
determined and thus the concentration with which the solvent leaves the process
(2,5).
The pole point π is formed from the intersection of the lines E_ 0 R_ n and E_ 1 F_ .
The stage construction can be started by knowing the position of the pole point. At
the inlet stage, the mixture separates into two phases according to phase equi-
librium along the tie line. The tie lines are shown in Figure 7 as dashed lines. From
the extract phase E_ 1 , the raffinate stream R_ 1 of the first stage can be determined.
One of these raffinate streams and the pole point intersects the binodal on the
extract side in E_ 2 , which returns R _ 2 via the corresponding node and repeats this
procedure as often as required until the desired final concentration of R_ n is
reached or fallen below. The number of n steps required corresponds to the
required number of theoretical separation stages.
2.3. Mass Transfer. In any system that is not in equilibrium, balancing
processes take place to ensure that a state of equilibrium is achieved in the long
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 7

Fig. 8. Mass transfer by diffusion.

term. The mass transport is caused by concentration gradients and takes place in
technical processes by diffusion and convection (1,5).
Diffusion is responsible for mass transport in a stationary layer such as phase
boundary layers. Figure 8 shows how a component is transferred through a resting
delta x thickness boundary layer in the direction of the lower concentration. A
concentration gradient occurs only in the boundary layer. Outside the boundary
layer, it is assumed that mixing is optimal.
The diffusive mass transfer is described by Fick’s first law.

dci
n_ i ˆ Dij
A
(5)
dx

The material flow ni of a component i is therefore proportional to the interface A

and the concentration gradient dci =dx as well as the diffusion coefficient Dij . In this
case, mass transfer works against the direction of diffusion. As the pressure
increases, the diffusion coefficient decreases, while the temperature increases
the diffusion cofficient in liquids. The diffusion coefficient in liquids can be
approximated by the Stokes Einstein equation (6,7).
kB
T
Dij ˆ (6)
6
π
η
Ro

The diffusion coefficient is determined by the Boltzmann constant kB and the

temperature T in relation to the dynamic viscosity of the solvent, the hydro-
dynamic radius of the diffusing particles, and the sixfold number π (6,7).
In moving fluids there is convection in addition to diffusion. In convection, the
mass transfer is no longer based on the mass transfer of two resting fluid phases,
but also on macroscopic movements of the liquid, which can be either free or forced.
As a result of the movement, the contact between the liquids is more intense and
the convective mass transfer is considerably faster than the diffusive (2,8).
8 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Fig. 9. Two-film theory of mass transfer.

The convective mass transport can be described by the following equation.


n_ i ˆ βi AΔci ˆ βi A ci;BP ci;I (7)

The mass transport n_ i depends proportionally on the mass transfer coefficient βi ,

the interface A as well as on the concentration difference between the boundary
layer and the bulk phase. The mass transfer coefficient depends, on one hand, on
the properties of the material and on the other hand, on the type of flow, whether it
is laminar or turbulent (5,8).
Both mass transport phenomena occur simultaneously. The two-film theory
developed by Whitman and Lewis in 1923 considers the transport of matter from
the core of the first phase, via the phase boundary and both phase interfaces, to the
core of the second phase (Fig. 9) (9,10).
In the bulk phases, the concentrations are constant as a result of convection,
since the mass transfer is greatly accelerated due to turbulence. Within the
boundary layers it is to be assumed that due to the adhesion conditions it is an
almost stationary fluid or a pure laminar flow. Due to the predominant transport

Fig. 10. Equilibrium concentrations.

APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 9

mechanism of diffusion, a linear concentration profile is formed. The concentrations

at the phase boundary are not identical (9). The correlation between bulk, inter-
facial, and equilibrium concentrations is illustrated in Figure 10.
Phase II concentrations are plotted on the x-axis and phase I concentrations
are plotted on the y-axis. The two interfacial concentrations (cI;I and cI;II ) and the
equilibrium concentrations c∗I and c∗II are in equilibrium with the bulk concentra-
tions of the other phase. These form points on the equilibrium line. Using the
distribution and equilibrium coefficient K i , these correlations can be represented
mathematically (1,8).
cI;II ˆ K i
cI;II (8)
cBP;I ˆ K i
c∗II (9)
cBP;II
K i ˆ
c∗I (10)

Equation (12) is reformed for both phases according to the respective concentration
gradient, since the concentration at the phase boundary cannot be measured.

n_ i
cBP;I cI;I ˆ (11)
βI
A
n_ i
cI;II cBP;II ˆ (12)
βII
A

By introducing the equations (13) and (15) in equation (17), the correlation
between the concentrations c∗I and cI;I are obtained.

n_ i
K i
cI;I c∗I ˆ (13)
βII
A

The unknown interfacial concentration can be removed by further applying the

equation (11).
 
n_ i 1 Ki
cBP c∗I ˆ
‡ (14)
A βI βII

The term in brackets is the reciprocal of the mass transfer coefficient k.

1 1 Ki
ˆ ‡ (15)
kI βI βII

This consists of the individual resistances (transfer coefficients) and represents the
total resistance of the mass transfer process. If the mass transfer coefficient is used
in equation (14), the result is obtained

n_ i ˆ kI
A
cBP;I c∗I † (16)

10 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

The following equation is obtained for the transition mass flow of the second
phases.
n_ i ˆ kII
A
cBP;II c∗II † (17)

The associated phase II-related mass transfer coefficient is calculated as follows

(11).
1 1 1
ˆ ‡ (18)
kII βII K i
βI

2.4. Extraction Mechanisms. Extractions are distinguished according

to their characteristics in physical extractions, dissociative extractions, and
reactive extractions.
Physical Extraction. In physical extraction, the distribution of a compo-
nent can be described in sufficient approximation by the Nernst distribution. This
represents the linear dependence of the concentrations in the extract and raffinate
phase. The distribution of physical extraction is only based on the different
molecular structures and thus the different physical properties, such as polarity.
Physical extraction systems are therefore relatively free of disturbing circum-
stances, such as those arising from interactions in another phase. As expected,
such methods based on purely physical effects are rarely very selective and the
achievable separation factors are low. For the application of solvent extraction,
only a few substances can be found whose phase transition is unhydrated,
undissociated, and not associative. After Kertes and Grauer, only mercury and
some of its compounds, as well as halogens and the halides of the metals of the IVB
group (Ti, Zr, Hf,) are mentioned in this context, which fulfill this definition of
physical extraction in its strict form (12) (Fig. 11).
The idealized case of a purely physical distribution of a component between
two phases occurs when the two phases have no or negligible solubility in each
other. The equilibrium relations can thus be represented in sufficient approxima-
tion in linear dependence of the concentrations of aqueous and organic phase and
thus in analogy to Nernst’s distribution theorem. The distribution of component A
between the two phases (eg toluene and water) is then described by means of the
physical partition coefficient K d .

corganic
Kd ˆ (19)
caqueous

Fig. 11. Extraction mechanism based on physical partitioning (13).

APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 11

Fig. 12. Extraction mechanism based on dissociative partitioning (13).

Dissociative Extraction. Organic compounds with acidic or basic groups

are subject to a dissociation reaction in the aqueous phase. In the case of these
molecules, the actual dissociation reaction follows after extraction from the organic
phase into the aqueous phase. According to Wise and Willams, this is referred to as
dissociation extraction (14). The components extracted in this way are organic
molecules with acidic or basic groups (14). Examples are carboxylic acids such as
citric acid, salicylic acid, or phenols or compounds with amine groups such as
aniline, penicillin-G, DL-phenylalanine. The following illustrations show the prin-
ciple mechanism of dissociation extraction as an example (15) (Fig. 12).

‰ROHŠorg
K phys ˆ (20)
‰ROHŠw}ssrig
 
‰RO Š
H‡
K diss ˆ (21)
‰ROHŠ

The dissociation constant depends on the pH of the solution. The pH value can shift
the dissociation reaction in the direction of the dissociated or nondissociated
molecules. Due to the physical partition coefficient, further nondissociated com-
ponents are transferred from the organic to the aqueous phase. The yield depends
on the partition coefficient and thus on the pH value.
Reactive Extraction. Reactive extraction is a coupling between the chemi-
cal conversion of the transfer component and physical mass transport. Extraction
by chemical reaction is suitable for the separation of biochemical compounds,
organic bases, and acids as well as metals ions. The selectivity of the separation
process can be positively influenced by the selection of suitable reactive partners
(Fig. 13).

Fig. 13. Extraction mechanism based on reactive partitioning (13).

12 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Rigid Circulating Oscillating Deformed

Drops Drops Drops Drops

Drop velocity

Drop Diameter

Fig. 14. Droplet rise velocity as a function of diameter and shape.

2.5. Hydrodynamics. Figure 14 shows the rising velocity as a function of

the drop diameter.
In the case of small drops, there is a linear relationship between the diameter
and the drop rise velocity. In this area, the drops are spherical and slightly
deformed. With increasing size, the drops begin to form internal circulation flows.
These lead to an increased rise rate (16).
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4
j ρd ρc †j
g
dd
uT ˆ (22)
3
ρc
c w

The gradient then decreases and the speed reaches its maximum. The larger drop
volume means that the drops lose their stability and begin to vibrate. The drops
become flatter and eventually rise more slowly than rigid spheres of comparable
size (16).
As the droplets increase in size, they become flatter and more deformed. This
further increases the flow resistance and reduces sedimentation velocity. As of a
certain size, the drops become unstable and disintegrate (16). The calculation of
the stationary rate of rising velocity ud of a single drop is done by the equilibrium of
weight, buoyancy, and resistance (13,17). The drop velocity depends on the density
difference of the drop and the continuous phase, the earth acceleration g, the drop
diameter dd and the coefficient of resistance cw . As shown in Figure 14, the mobility
and thus the coefficient of resistance depends on the size of the drop and the type of
flow conditions inside the drop (17).
The mass transfer in extraction devices depends largely on the phase
interfacial area between the continuous and disperse phase. The size of the
interfacial surface consists of the mean specific surface and its hold up.
The holdup ϕ is the volume of the disperse phase V d in relation to the void
volume of the extraction apparatus V.

Vd
ϕˆ (23)
V
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 13

The holdup can also be considered in smaller volumes which, for example,
represent theoretical separation stages. These hold ups are not constant over
the column height and change due to internals, drop size, and mass transfer (18).
The holdup can also be calculated using the drop rise velocity. However, since the
velocity of a single drop differs fundamentally from that of a drop in a droplet swarm,
the hold up in columns must be determined by the droplet swarm velocity (19).
ud uc
us ˆ ‡ (24)
ϕ 1 ϕ

This equation uses the layer model and establishes a relationship between the
averaged relative velocity of the droplet swarm us to the two void velocity of the
continuous uc and disperse phase ud as well as the hold up. If the swarm velocity is
known, the holdup can be determined from this equation (13,19).
The liquid load L of the column is determined from the ratio of the two
entering volume flows of the disperse V_ d;in and the continuous phase V_ c;in to the
column cross-sectional area AC according to the following equation (1,2).

V_ c;in ‡ V_ d;in
Lˆ (25)
AC

The maximum liquid load is called flooding point. This parameter limits the size of
the volume flows. If too much continuous phase is introduced into the column, the
disperse phase is entrained. If, on the other hand, the volume flow rate of the
disperse phase is set too high, the continuous phase can no longer sink down (1).
Figure 15 shows the relationship between the liquid loads of the disperse and
continuous phase, the flooding limit, and the holdup as an example for a sieve plate
column.
∙ ∙
Vd / Vc: 1:3 1:2 1:1.5 1:1
3 6
Volume flow continuous phase (m3/m2h)

60 9

12 1,5.1
6
Flooding limit
40 15 Loading
8 2:1 Phase ratio
9 10 18 Number of theoretical stages
Holdup (%)
11 21 3:1
10

20
0%

12
Fl

27
oo
di
gn
80

lim
40

60

it
%

0
0 20 40 60
Volumeflow dispersed phase (m3/m2h)

Fig. 15. Operating diagram for a pulsed sieve plate column with water (c) + butyl acetate (d)
+ acetone (a f = 11.7 mm/s). (Reproduced with permission from Ref. (1). Copyright 2006,
John Wiley & Sons.)
14 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

The holdup increases with increasing liquid load of both phases. The depen-
dency of the holdup on the disperse phase is higher. By increasing the flow of
disperse phases and a constant continuous phase, the hold up to the flooding point
increases slightly exponentially. When the liquid load approaches the flooding
point, the number of theoretical separation stages (column efficiency) increases.
Therefore, the highest possible flow rates are desirable. Since further operation of
the column is not possible when the flooding loading is exceeded, columns are
operated at a maximum of 80% of the flooding loading. Furthermore, it can be seen
that a higher proportion of disperse phase has a positive influence on the number of
separation stages (1,2).
2.6. Interfacial Phenomena. Since the interfacial tension depends on
the concentration of the transfer component, an interfacial tension distribution
occurs during mass transfer between two phases at the phase interface. This
causes the phase boundary fluid to move from areas of low interfacial tension to
areas of higher interfacial tension. Therefore, liquids with low interfacial tension
spread on liquid interfaces with higher interfacial tension. If the stress gradient of
the interface is very large, eruptions can occur at the interface. Phenomena of this
kind are called interfacial convection or Marangoni effect. Sternling and Scriven
carried out a mathematical analysis of such interfacial instabilities and estab-
lished criteria for the occurrence of the Marangoni effect (20). According to this,
Marangoni effects occur at

• mass transfer from the phase with the higher viscosity to the phase with the
lower viscosity,
• the mass transfer from the phase with the smaller to the phase with the
larger diffusion coefficients,
• large differences between the kinematic viscosities and the diffusion coef-
ficients in both phases,
• steep concentration gradients at the phase boundary,
• strong dependence of interfacial tension on concentration,
• low viscosities and diffusion coefficients in both phases,
• absence of surface active substances, and
• rapid expansion of the interface.

With these criteria the occurrence of Marangoni effects can be qualitatively

estimated. Modigell (21), after investigating interfacial instabilities, gave a simple
stability criterion from the results found experimentally to complement the
criteria of Sterling and Scriven: If the dependence of the interfacial tension on
the concentration of the transition component is known, it can be deduced from the
maximum inducible local interfacial tension difference whether the material
system is stable or not. If this is less than a critical threshold of 10^3 n/m, no
instabilities occur (21).
The Marangoni effect has a double effect on the mass transfer. Under the
influence of the Marangoni effect, the coalescence mechanism and the mass
transfer coefficient are strongly dependent on the direction of mass transfer.
For a material system with a negative interfacial tension gradient, there is a
greater probability of droplet coalescence in the direction of mass exchange from
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 15

the disperse to the continuous phase (d!c). The film between two drops or between
one drop and the interface has a lower interfacial tension due to the high
concentration of the transition component. As the film strives to expand, the
liquid flows out of the film and coalescence occurs. If the mass exchange takes place
in the direction from the continuous to the disperse phase(c!d), liquid flows out of
the environment through the higher interfacial tension in the film and thus leads
to a stabilization of the film. As a result, the coalescence time in a mass transfer
direction d!c increases to the range of a few seconds instead of a few milliseconds
if no mass transfer takes place (22).

3. Equipment

There is a large number of extraction equipment that can be classified, as shown

Figure 16.
In all liquid–liquid extraction apparatus, the phase contact occurs when one
phase is dispersed in the other. This phase contact enables the mass transfer,
which can be carried out in one or more stages. Single mixer–settlers are operated
in single-stage mode, while several mixer settlers connected in series can also be
operated in so-called batteries in countercurrent or cross-current mode. The most
common extraction apparatuses that work in countercurrent mode are columns. A
distinction is made here between columns without energy supply, such as spray
columns or packed columns, and columns with energy supply, which include
pulsed or stirred columns (11,23).

Liquid–Liquid Extraction

SINGLESTAGE MULTISTAGE

Direct Current Counter Current

Mixer Settler Centrifugal Extraction

Extractor Columns

without with
MSB Separators
Energy Input Energy Input

By By
Agitators Pulsation

Single Apparatus Plate Separator Podbielniak Spray Column Scheibel Dumped Packing
series connected Column
Chamber Dumped Packing RZE
Tubular
Separator Column Kühni Sieve Tray Column
Box Design Centrifuge
a) Pulsed Liquid: PSE
Quadronic Sieve Tray Column SHE b) moved Trays: Karr
Tower Design

Ordered Packing RDC Ordered Packing

Column ARD Column

Fig. 16. Extraction equipment overview.

16 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

3.1. Mixer–Settler. Mixer–settler consist of two spatially separated

devices, a stirring unit (agitator, static mixer, or similar), and a directly adjacent
settler. This straightforward design is one of the oldest extractors in history. The
mixing time must be long enough to achieve equilibrium between the two phases.
The drop size is adjusted by the stirring speed. It should be sufficiently small
to provide the mass transfer with a large specific phase interface, but at the
same time large enough that the separation time does not become too long.
The minimum of the total mixing and separation time is then used to determine
the agitator speed. In the inlet area of the separator, coalescence aids in the
form of fillers or sheet metal packings can be applied to accelerate the settling
rate. In practice, mixer-settler are operated with stepwise phase contact. They
are particularly attractive for large throughputs and few separation stages
(5,24).
3.2. Extraction Columns. An extraction column can be divided into
three areas. In the lower part of the column is the column bottom. In the column
bottom, the heavy phase is removed from the column and the light phase is added.
When inserting the light phase, care must be taken to ensure that the drops are
optimally sized and distributed over the entire column cross section. The size of the
drops has a significant influence on the separation efficiency of a column. Large
droplets have a poor surface to volume ratio and thus may lead to a insufficient
mass transfer. If the drops become too small, there is a risk that a stable emulsion
will form, which cannot be separated during the process.
The contact zone is located in the middle area, where the mass is transported
from the heavy phase to the light phase. In the upper area, also called column head,
the drops coalesce and can be removed from the column. In the column head, the
heavy phase is also fed into the column.
The general column structure can be supplemented by fixtures. These can be
pulsation pumps or moving internals, among other things, in order to enable an
additional energy input into the column. The installations may also be nonmoving
internal or packings. These can produce additional turbulence in the column. The
internals serve to improve the mass transfer in the column (25).
In extraction columns, the countercurrent is caused by the difference in
density between the two phases. If the light phase is dispersed, it rises up in
through a vertical cylindrical body and coalesces at the head of the column. If the
heavy phase is dispersed, then the phase separation area is correspondingly
located at the bottom of the column. This applies to both static columns (without
external energy supply) and to columns with external energy supply, which include
stirred and pulsed columns.
The simplest construction of a static column is the spray column. It has no
built-in components and is therefore only suitable for mixtures with a high
difference in density, since the differences are too small, large volume vortices
can be formed that lead to backmixing. This effect of backmixing is counteracted by
installations in other designs. These installations can be either sieve bottoms,
fillers, or structured packings. In the case of sieve plate columns the droplets are
formed at each stage. Due to the missing energy input, the size distribution of the
droplets is not optimal and the separation efficiency of the column is comparatively
low. Therefore, these columns can only be used for problems that require a small
number of separation stages (26,27).
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 17

On the other hand, there are columns with external energy supply. These
include extractors with rotating internals such as stirred and pulsed columns.
With stirred columns, agitators are mounted on a rotor shaft that runs centrally
through the column. The rotation of the stirrers ensures that droplets are
continuously splitted to improve the mass transfer. The speed of the stirrer
must be above a critical speed in order to achieve an effect. However, it should
be noted that a reduction in the size of the droplets reduces the rise velocity of these
drops. In return, the speed of the continuous phase must therefore be reduced and
the throughput of the phases deteriorates accordingly. As a result, the cross section
of the column must be enlarged (28).
In pulsed columns, the pulsation is generated either by a piston pump or
pulsating membrane in the bottom of the column or by moving internals.
Periodic pulsation causes rapid up and down movement of the entire
column contents through the packing. Usually a pulsation intensity of af 1–
2.5 cm/s is aimed for. This results in turbulence, which leads to a higher
dispersion of the drops and thus to an improvement of the mass transfer. The
separation efficiency of pulsed columns compared to nonpulsed columns
improves by about twice as much. But also, the smaller drops may lead to larger
column diameters (2,5,24).
3.3. Centrifugal Extractors. In the case of liquid phases with a very
small difference in density, the Earth’s gravity field alone is not sufficient to
separate the phases in a reasonable time. Centrifuges accelerate the liquid/liquid
separation by the action of centrifugal force due to rotation of the centrifuge drum.
They can thereby generate a much stronger gravitational field and thus enable
even the separation of mixtures with a low difference in density (Δρ smaller than
30 g/m3) in acceptable time (24,29).
The separation principle of a centrifugal extractor is explained in the
following. Two immiscible liquids are fed into the annular mixing zone, which
forms between the rotor and the stationary housing, through separate inputs. The
liquid–liquid dispersion generated by the rotation of the rotor in the annular
mixing zone flows through gravity to the lower inlet. The actual separation zone is
located inside the rotor and the dispersion is quickly separated into the respective
heavy and light phase under the high centrifugal force. The separate phases leave
the apparatus via separate outputs.

4. Process Design

4.1. Criteria of Solvent Selection. The solvent selection is the central

part of planning the extraction process. A compromise can be found between many
factors and properties. The most important requirement of the solvent is a high
selectivity and capacity in relation to the target component.
The capacity determines how well a component is distributed between
solvent and raffinate. It is therefore a decisive criterion for the quantity of solvent
required. The larger it is for the components to be extracted, the less solvent is
needed for the extraction and the fewer extraction steps are required. This means
that smaller extraction columns with fewer stages can be used and a smaller
amount of liquid must be prepared after extraction. The distribution coefficient,
18 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

the quotient of the concentrations of a component in both liquid phases, is a

measure of capacity. Hence, the lower the distribution coefficient of the target
component is, the larger is the required size of extraction equipment.
The selectivity, on the other hand, specifies how much better the target
component is extracted compared to the other components in the raffinate.
This means that the greater the selectivity, the less number of theoretical
separation stages are required. The consequence of low selectivity is that many
unwanted components are present in the solvent and the subsequent concentra-
tion becomes more complicated and expensive (30). Selectivity is therefore another
crucial characteristic. If possible, a solvent should only extract the desired compo-
nents. The more efficiently this happens, the better the separation performance of
the extraction. Two definitions of selectivity are given in the literature. On the one
hand, the activity coefficients of the components to be separated are used with
infinite dilution in the extractant. The quotient of these values represents the
selectivity Sij between these two components. On the other hand, the ratio of the
distribution coefficients of the different components is used. The quotient of the
coefficients gives the separation factor Sij, which also represents selectivity.
Furthermore, mutual solubility should be low. For liquid–liquid extraction,
the occurrence of two liquid phases is essential. Without a mixing gap between the
extractant and the component in the raffinate stream, extraction is not possible.
The mixing gap should therefore be as large as possible, as this results in increased
separation efficiency and less solvent loss due to the raffinate stream. Considering
the ternary LLE with the solvent and two components to be separated, a larger
mixing gap means that a wider range of feed compositions can be processed.
A further requirement is that the difference in density between extract and
raffinate must be large enough, since the densities can change during the process
and there is a risk of emulsion formation if the densities are too similar. In
addition, a large difference in density is advantageous when separating the
phases. If the difference in density is too small, and all other properties listed
here are very satisfactory, centrifugal extractors can be used to accelerate the
separation of the liquid phases. If the density of the phases is identical for an
extraction step, the separation cannot be performed in the extractor.
Another important factor is the interfacial tension. If it is too low, there is a
risk of emulsion formation as the drops are too easily dispersed. If it is too large, on
the other hand, too much energy is needed to disperse the drops and thus improve
the mass transfer.
A low viscosity of the substances is also desirable, since a high viscosity has
negative effects on the mass transfer behavior and fluid dynamics (5,30). Viscosity
of both phases needs to be considered as well since high values may prove to be
negative for extraction. This makes it more difficult to pump the component
around. In addition, the heat transfer in possible heat exchangers of the plant
slows down. At the same time, the diffusion of the components is also required
according to the Stokes Einstein equation, which triggers problems in mass
transfer that reduce the extraction speed.
Ecological and safety requirements must also be observed. Ideally, the
materials used should be nontoxic, noncorrosive, flammable, or environmentally
hazardous. However, in the case of conventional liquid–liquid extraction with
organic solvents, this cannot be fulfilled.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 19

Also, a low boiling point and vapor pressure are advantageous for subsequent
treatment, as this reduces the heat required during distillation. This also enables
temperature-sensitive substances to be protected (2,5,30). Vapor pressures should
vary as much as possible in order to simplify postextraction treatment by distilla-
tion and to reduce the likelihood of an azeotrope occurring. In addition, at least one
of the vapor pressures should not be too low in order to avoid high temperatures
during distillation. This could adversely affect the stability of the molecules and
lead to higher heating costs.
This is why the separation factors in distillation must be considered as well.
In order to be able to measure the cost input for the subsequent separation of the
components from the solvent after extraction by means of distillation, the separa-
tion factor between the components to be separated and the solvent must be
considered during distillation. Separation factors close to 1 indicate a difficult
separation, which is only possible with the use of a high number of stages in a
distillation column. This is economically unfavorable. Therefore, solvents with
separation factors that are very different from 1 should be preferred.
In this context it is important that azeotropic behavior between the compo-
nents to be separated and the solvent should be avoided, as under these circum-
stances the complete separation of the solvent is not possible by distillation.
Hampe, Wauquier, Rydberg et al., Perry et al., and Cockrem et al. gave
an overview of the most important properties of a solvent for liquid–liquid
extraction.
4.2. Model-Based Process Design and Parameter Determination.
Liquid–liquid extraction still requires time-consuming and device-specific tests on
a laboratory and mini plant scale to obtain all the necessary parameters required
for selecting and dimensioning extraction apparatuses (31).
This knowledge makes it possible to predict the costs of the extraction process.
The number of influencing factors of liquid–liquid extraction is limited, but these are
linked in complex ways, especially in columns. The influencing variables and
possibilities for gaining knowledge about these are shown in Figure 17 (31,32).
The process development of liquid–liquid extraction can be based on the
concept illustrated in Figure 18.
In laboratory experiments, parameters are determined for the modeling. These
provide the necessary operating points for the extraction apparatus. Furthermore,

Batch‐settling exp Shaking exp

Sedimentation Equilibrium
Drop size and drop
size distribution

Coalescence and
Mass transfer
breakage
Holdup
Swarm behavior

Flooding Drop exp

Column exp

Fig. 17. Influence variables (black) and their possible extraction studies (colored).
20 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Separation Material Process

Equipment
task data concept

System
volumes
Shake Flaska ~ 6 mL

Mixer–settler battery ~ 1.5 L

Experiment

Parameters
Droplet
~ 0.5 L

Validation
measurements

Param

Parameters
Columns, centrifuge ~4L

Valid.
Param
Equilibrium stages + kaff (ldf)
Modeling
Distributed plug flow

Economic evaluation

Time + ressources

Fig. 18. Process development of liquid–liquid extraction. (Reproduced with permission

from Ref. (33). Copyright 2012, Teknoscienze Srl.)

the models can be validated by experiments. The experiments to determine the

parameters are presented and explained in the following sections (31,34).
Shaking Experiments. Shaking tests are one of the fastest, simplest, and
least expensive extraction tests. In shaking tests, dispersion between two or more
phases is created by shaking or stirring. The phases are to be mixed until both
phases are in equilibrium with each other and the mass transfer is thus completed.
In addition to the concentration of target and side components in light and heavy
phases, the distribution coefficient can be determined from their ratio.

cLP
K ˆ (26)
cHP

Furthermore, the separation factor α can be determined from the ratio of the
distribution coefficient of the target component to that of a side component.

K TC cLP;TC
cHP;SC
ሠˆ (27)
K SC cHP;TC
cLP;SC

The separation factor is a measure of how much better the target component is
extracted than the side components. In addition, initial findings on yield and
purity of the single-stage process can be obtained in shaking tests (2).
Batch-Settling Experiments. The intention of batch-settling experiments
is to determine the settling time and behavior of two-phase systems. During this
process, for example, the mixing time, stirrer speed, stirrer type, or phase
composition is varied and the settling time is measured. Furthermore, estimates
of typical drop sizes can be determined. During the mixing process, one phase
(disperse phase) is dispersed in another phase (continuous phase) in the form of
drops. The adjustable parameters mainly influence the drop size distribution.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 21

Oil dispersed
Initial
in water

Heights of coalescing and sedimenting interfaces hs and hs

H0 coalescence time τ0

Clear oil Initial

coalescence
rate ψ0 Maximum slope ψi
Coalescing interface
hp
hc Dense-packed Height of coalescing interface hc
Zone
Dense-packed zone

hp
Sedimentation zone
Sedimentation zone Height of sedimenting interface hs

hs Sedimenting intercace
Maximum slope v0
Clear water Initial drop diameter ϕ0
Ti Tf
Time, t
(a) (b)

Fig. 19. Batch-settling experiment (a) and corresponding sedimentation curve (b).
(Reprinted with permission from Ref. (36). Copyright 1998, American Chemical Society.)

Whether the light or heavy phase is dispersed depends mainly on the phase ratio.
Figure 19 shows an example of a sedimentation test in which the heavy aqueous
phase is dispersed in the light organic phase (5,35,36).
After mixing, the heavy phase begins to sediment in the direction of the
dispersed phase. If sedimentation proceeds faster than coalescence, an interphase
is formed in which the droplets accumulate and grow due to coalescence of
the droplets with each other (see Fig. 19). Finally, these drops coalesce at the
phase boundary and the interphase dissolves (35,37).
Droplet Measurement Cell. Single drops in extraction columns are influ-
enced by four basic phenomena: sedimentation, mass transfer, droplet splitting,
and coalescence. Drop measuring cells were developed with the aim of under-
standing and describing these effects independently in a controllable environment
(16,32,34,38,39).
The setup of a typical drop measuring cell is shown in Figure 20.
Defined drop volumes are injected via a syringe pump. The droplets then
rise through a cylindrical chamber, which can be filled with internals such as
stirrers or a structured packings, if required. The sedimentation rate can be
determined by the length of the rising part and the measurement of the time
required for the drop to pass through it. The residence time in the system can
be controlled by switching a centrifugal pump on and off. This pump is pumping
the continuous phase in countercurrent and the droplet is held in a conical part of
the measuring cell for the desired time in suspension. The droplets are then
collected in a funnel, removed with a syringe pump, and then analyzed. This
enables, for example, a measurement of the mass transfer rate depending on the
residence time and the drop size. It is also possible to investigate droplet
breakage and coalescence as well as swarm behavior. When the measuring
22 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Dispersed phase
outlet

Continuous
Drop outlet
phase inlet

Levitation area

Drop generation

Continuous
phase outlet

Dispersed phase
inlet

Fig. 20. Schematic design of a drop measuring cell. (Reproduced with permission from Ref.
(39). Copyright 2015, John Wiley & Sons.)

cell is filled with internals, its influence on the residence time and mass transfer
can also be determined (16,32,34,39).
4.3. Mixer–Settler Design. Extraction processes are mainly determined
by at least the following sequences of events (17):

1. Dispersion of the two liquid phases

2. Mass transfer between the carrier and the solvent
3. Separation of the phases.

These phenomena cannot be decoupled from each other and occur simulta-
neously, especially in the case of mass transfer. Steps 1 and 3 are also antagonistic
to each other, that is, the speed of phase separation decreases with increasing
dispersion work. Both phenomena need to be balanced in terms of mass transfer
rate. Consequently, the design of the mixer is essentially determined by the mass
transfer rate. This in turn depends on the mean droplet size, which in turn is
influenced by the physicochemical properties of the mixture, the mass transfer
direction, as well as the mixer geometry and the power input (Fig. 21) (17).
Early Prediction of Settler Dimensions. To get a rough estimation of the
dimensions of the apparatus, volume flow (V_ ) is calculated, which is dependent on
the entire volume to be processed (V total ) of the ATPS in relation to the prede-
termined process duration (ttotal ):

V total
V_ ˆ (28)
ttotal
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 23

Light phase Light phase

Inlet outlet 1 outlet
2
2 1 3
3
4 Inlet 4 Ld
5 6

Ld 5
6 Heavy phase
Heavy phase outlet
outlet

1 Continuous light phase 4 Sedimentation zone

2 High holdup packed zone 5 Densely packed sedimentation zone; low holdup
3 Inlet dispersion cone 6 Continuous heavy phase

Fig. 21. Development of the dispersion layer in horizontal and vertical settler devices.
(Reproduced with permission from Ref. (17). Copyright 1981, John Wiley & Sons.)

The length LS can then be calculated using volume flow and the predetermined
residence time (tRDT ), as well as the cross-sectional area of the device (AS ):

V_  tRDT
LS ˆ (29)
AS

Further influencing variables, such as the drop size, sedimentation rates, or other
parameters influencing coalescence are not taken into account. This quick esti-
mation is often referred to as minimum-apparatus-volume V min (MAV):

V min ˆ V_ tRDT (30)

Batch-Settling Data-Based Prediction of Settler Dimensions. A more

sophisticated procedure to calculate the dimensions of horizontal settler units can
be based on batch-settling data (40). In the model, the power input, sedimentation
rate, and coalescence behavior of the single droplets are taken into account. To
examine these influencing factors more precisely, the length of the separator (LS )
is divided into two regions: the inlet area of the dispersion (Lin ) and the coalescence
region (LC ) (Fig. 22).
In the inlet area, occurring turbulences and the formation of a droplet layer
are considered. Within the apparatus, drop coalescence and the formation of the
densely packed layer (hp ) are of greater interest. The change in the densely packed
height can be calculated as follows:

dhp V_ Dis  11:3  s  ηDis ‡ 126  ηc ‡ ηd ††

ˆ
 (31)
dl hp  D3S  εp  1 εp  Δρ  g

It is dependent on the dispersion volume flow (V_ Dis ), the dispersion viscosity (ηDis ),
a slippage parameter (s), the diameter of the settler (DS ), and the dispersed volume
fraction in the densely packed layer (εp ). The settling time (tS ) is normalized to
24 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

LS
Lin Lc

LPout
ATPSin
Densely packed layer

HPout

Fig. 22. Structure of the settler (41). (http://www.mdpi.com/2073-4468/6/4/21. Licensed

under CC BY 4.0.)

level out the influence of long coalescence times at the boundary surface. The
compensation parameter is (C∗ ). With the aid of these two parameters, the
coalescence length of the separator can be determined as follows:
tS  V_ d;0
Lc ˆ vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!0:5 (32)
u
u ϕ
DS  tH p  H 0:3  C∗
1:3 32;0;settle
d;0 =
ϕ32;0;settler

It is dependent on the final height of dispersed phase in a batch-settling experi-

ment (H d;0 ), the volume flow of dispersed phase (V_ Dis ), and the starting droplet
diameter in a batch-settling experiment (ϕ32;0;settle ) and inside the settler
(ϕ32;0;settler ). The height of the densely packed layer inside the settler (H p ) should
be predefined. The inlet length must also be determined. It is dependent on the
velocity inside the settler (vS ), the volume of the inlet (V in ), the diameter of the inlet
(Din ), earth gravity (g), and the dispersion band height inside the settler (H Dis ). In
this case, the mean density (ρ) and the hold-up (ε0 ), which corresponds to the
volume fraction of the dispersed phase in the total volume, are determined:

0:4 0:5 0:5

43:7 ρ0:3  v0:5
S  DS  V in  Din
Lin ˆ
0:4 (33)
Δρ0:3 ϕ32;0 ‡ H Dis  1 ε0 †0:4 g0:5

ρ ˆ ε 0  ρd ‡ 1 ε0 †  ρ c (34)
H d;0
ε0 ˆ (35)
H c;0 ‡ H d;0

The total length of the separator (LS ) is composed of the inlet length (Lin ) and the
coalescence length (Lc ):

LS ˆ Lin ‡ Lc (36)

The geometry of the dispersion layer is not further considered in the simplified
model. For further explanations, refer Henschke et al. (40). Additional approaches
for detailed calculation of settler dimensions can be found in Ref. (18).
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 25

4.4. Column Design. Although extraction can be used for a wide range
of tasks, it is usually only used when conventional methods such as distillation or
steam stripping are not applicable. One of the main reasons for this is the design
of extraction columns. Especially, the complex prediction of phase equilibria of
the substance systems as well as the interrelationships between fluid dynamics
and mass transport mechanisms make it difficult to design extraction columns.
Therefore, even today extraction columns are designed based on pilot plant
experiments. For dimensioning, knowledge of equilibrium, hydrodynamics, and
mass transport kinetics are necessary. Usually, these must be determined
experimentally for scale-up, at least in scales of DN50-150. This experiment-
based methodology requires large amounts of feedstock, especially in the early
phase of the project. Research so far is based on equations for droplet rise, break-
up, and coalescence, taking into account backmixing and mass transfer. These
models have mainly been tested on EFCE standard test systems. An overview of
factors that play a role in the separation performance and their influence are
listed in Table 1.
The uncertainties arising from these interrelationships can usually only be
resolved by extensive experiments on laboratory and pilot plant scale (13). In order
to reduce the costs of the experiments, numerical simulations are increasingly
being developed. These are based on mathematical modeling. Model parameters
must be determined for modeling, which are used to describe effects in the
extraction. To determine these model parameters, drop or droplet swarm experi-
ments with the original solution can be used.
The accuracy and confidence range of the simulation depends on the com-
plexity of the selected model. The relationship between the complexity of a model
and the resulting accuracy is also referred to as the modeling depth. The available
models for liquid–liquid extraction and the corresponding modeling depth are

Table 1. Influences on the Separation Efficiency of Extraction Columns

Influence Description
flow parameters: small droplets and narrow drop size distributions
enhance the transport of material
• drop formation
• drop size
• drop size distribution
• drop swarm movement
• drop coalescence and breakage
interfacial effects: surface turbulence caused by interfacial tension
gradients improves mass transport to the phase
• interfacial tension gradient interface compared to mass transfer by pure diffusion
• viscosity gradient
• Marangoni effect
mixing effects: driving force for mass transport is reduced by mixing
effects compared to the ideal case of plug flow
• longitudinal mixing
• back mixing
• cross-mixing
Source: Reproduced with permission from Ref. [5]. Copyright 2010, John Wiley & Sons.
26 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

x Y
Black box Ċj+1 Ḋ j
Y = f(x)
Ċ
Equilibrium stages
Short cut ṅ
dx
Xα Yα Ċj Ḋ j-1
In (ε-1)- + +1
Xω Yω
Nth = –1 dτ = konst.
In ε
Distributed plug flow
Equilibrium stages w. LDF Ḋ Leistner,
franke (*)
w. mass transfer
Ċj+1 Ḋ j Ċ
HTU/NTU Population
balances
ṅ
Distributed plug flow ṅ
w. inner diffusion Groups:
dτ = f(x,y) Pfennig.
Ċj Ḋ j-1 kr -
Cc bart,
Ḋ stichlmair,
CD,x kraume,
CFD etc.
X

(*) Leistner, J. 2005: Shaker Verlag Aachen;

Molecular modeling
Franke, M. 2006: Fortschnitt-Berichte VDI Dosseldort

Fig. 23. Available models and respective modeling depth.

shown in Figure 23. The complexity and thus the modeling depth of the models
increases from left to right.
The following sections present the most important models and their basic
principles.
Black Box. The left edge of Figure 23 shows the black box model, which
offers the lowest model depth in this overview. In a black box, only the
incoming and outgoing streams are considered. The material flows are linked
by a simple relationship. This relationship does not include a substantiated
physical or chemical model and is based solely on empirical data or mass
balance.
Equilibrium Stage Model. The equilibrium stage model has a higher
modeling depth than the black box model. Within the scope of the equilibrium
stage model, the observed balance area is mentally divided into individual
theoretical separation stages (1). Within the separation stages, a complete
mass transfer takes place between the phases. Accordingly, phase equilibrium
is achieved at each theoretical separation stage.
However, real apparatuses and systems differ from the model in terms of
phase equilibrium. In a real separation stage, the phase equilibrium is usually not
achieved. Factors such as mass transfer resistance, the contact time of the phases,
fluid properties, and flow conditions play an important role here. The deviation of
the real separation stage from a theoretical separation stage can be quantified by
the stage efficiency.
If the model is transferred to real apparatus, the mixer–settler cascade or
column with a spatial separation comes closest to the model of the theoretical
separation stages. At a stage efficiency of 1, a mixer–settler or a section in the
column thus forms exactly one separation stage.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 27

Packing columns are not clearly separated. In a known equilibrium curve, the
Kremser equation can be used to determine the theoretical stages (2).

 
Xα Yα
ln ε 1†
‡ ‡1
Xw Xw
Nt ˆ 1 (37)
ln ε†

The extraction factor in the Kremser equation can be determined as follows.

ε ˆ K∗
L T T ˆ K∗
ν (38)

The equilibrium stage model has a higher model depth than the black box;
however, in this model only the distribution coefficients and the flow conditions
are considered. Therefore, many of the influencing factors of extraction are not
taken into account in the equilibrium stage model, such as drop size, backmixing,
or coalescence.
Method of Transfer Units. The method of transfer units is of great impor-
tance in practice. In contrast to the equilibrium stage model, the influences of axial
backmixing can be taken into account. The so-called HTU value (height of a
transfer unit) is determined from the mass balance for a differential column
element of height dh, without taking the axial backmixing into account:

dx keff ρc aAcolumn
x x∗ y† † ˆ 0 (39)
dh m_ c
dy keff ρd aAcolumn ∗
y x† y† ˆ 0 (40)
dh m_ d

Through integration you get

htop xtop
m_ c dx
hcolumn ˆ dh ˆ ˆ HTUc NTU c (41)
∫ keff ρc aAcolumn ∫ x x∗ y†
0 0xbottom

htop ytop
_ d
m dy
hcolumn ˆ dh ˆ ˆ HTUd NTU d (42)
∫ keff ρd a Acolumn ∫ y∗ x† y
0 ybottom

with the definition

m_ c
HTUc ˆ (43)
keff ρc aAcolumn
m_ d
HTU d ˆ (44)
keff ρd aAcolumn
28 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

The HTU value denotes the height of a transfer unit. While the NTU value denotes
the number of a transfer unit:

xtop
dx
NTUc ˆ (45)
∫ x x∗ y†
xbottom

ytop
dy
NTUd ˆ (46)
∫ y∗ x† y
ybottom

In technical columns, the HTU value increases through the occurrence of mixing
effects. Bauer has given the following equation (42):
 
dcolumn;tech 0:33
HTUtech ˆ HTUtest (47)
dcolumn;test

HTUtech and dcolumn;tech are the HTU value and the diameter of the technical
column and HTUtest and dcolumn;test are the corresponding values of the test column.
The scale-up factor of 0.33 takes into account that the separation effect decreases
with an increase in the column diameter.
According to Stemerding and Zuiderweg, the HTU value can be divided into
two parts (43):

HTU ˆ HTU ∗ ‡ HDU (48)

HTU∗ is the term used for a nonmixing transfer unit. They assume that the HTU∗
value is independent of the column height and the HDU value only depends on the
mixing. They link the HDU value to the dispersion coefficient and specify the
following equation:
0   1 1
vd
B 1 0:8ln m
C
HDU ˆ B @ Dax;d ‡ vc C
A (49)
vd vd
ϕ ‡ 1 ϕ†Dax;c m m 1 hcolumn
ud v2c vc

Also, the theoretical separating stage can be obtained in a similar way:

hcolumn ˆ HETP NTS (50)

Here, HETP is the height equivalent of a theoretical stage or plate and NTS is the
number of stages. According to Qi, the following conversion applies between HETP
and HTU (22):
    
vd vd
HETP ˆ ln m = m 1 HTU (51)
vc vc
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 29

Distributed Plug Flow Model. The distributed plug flow model is based on
the work of Danckwert (44), who pursued the idea of considering mixing as a stationary
process with a plug flow superimposed on it. The control volume is divided into discrete
spaces. With an increase of the number of discrete spaces against infinite, the height of
the control volumes becomes differential small. Discretions do not correspond to the
steps used in the equilibrium stage model. The discrete spaces serve as simplification for
the numerical solution of the mass transfer equation (44). The following illustration
shows the schematic structure of a distributed plug model for an extraction column
operated in countercurrent (13) (Fig. 24).
The model can be subdivided into the continuous extract phase (index c) and
the dispersed feed phase (index d). On the basis of the dispersion model, a
differential equation for the continuous and disperse phase follows, which
describes the axial concentration curve along the column axis:
@ccon @c2 @cc 6 ϕ

ˆ Dax
c2 uc
‡ keff

cd cd;Eq (52)
@t @z @z ddrop 1 ϕ

@cd @c2 @cd 6


ˆ Dax
d2 ‡ udis
keff

cdis cd;Eq (53)
@t @z @z ddrop

The Danckwerts boundary conditions at the column top are

@cdi
Ddax;i
ˆ vd
cd;ein cdi jz ˆ L (54)
@z zˆL
i

@cci
ˆ0 (55)
@z zˆL

Feed Extract

n
n n-1
n-1 n-2
n-2

j+1 j z +dz
j j-1
n∙ transa z

3
3 2
2 1
1

Raffinate Solvent
Fig. 24. Control volume in the DPF model (13).
30 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

At the column bottom the boundary conditions are

@cdi
ˆ0 (56)
@z zˆ0

@cci
Dcax;i
ˆ vc
cc;ein cci (57)
@z zˆ0
i
zˆ0

Population Balance-Based Modeling. Both the hydrodynamics and the

mass transfer kinetics in liquid–liquid extraction in a column are strongly influ-
enced by the properties of the dispersed phase and its change as it passes through
the apparatus. The breaking of a mother drop into a certain number of daughter
drops and the coalescence of a number of drops into a large drop represents the
elementary processes that enter into complex interactions, both in terms of mass
transport, particularly, and in terms of the forward and backward mixing of the
dispersed phase and the formation of the column flooding.
First works dealing with this problem using a population balance model were
already developed in the 1960s, but did not yet fully consider the fundamental
processes of drop breakage and drop coalescence (45–48). The work of Casamatta,
Al Khani, and Gourdon (Toulouse) as well as Haverland and Vogelpohl (Clausthal)
in the 1980s resulted in a process model that among the first described drop
breakage, drop coalescence, and mass transport and experimentally validated the
model-based prediction of the flooding point (26,49–51).
Here, the fundamental equation in terms of holdup, drop size distribution,
as well as the terms for drop coalescence and drop breakage are briefly presented
as a starting point for the reader to delve into the topic of population balance-
based modeling. The fundamental element of a PBE model is the drop size
distribution function P(h,d), which specifies the volumetric distribution of the
drops with a drop diameter d at each height h of the column. The holdup at any
position h of the column corresponds to the volume fraction of the dispersed
phase in the column section and can therefore be accessed via the integral of the
droplet size distribution:

dmax

ϕ h† ˆ P h; d†δd (58)
∫
0

The drop size distribution function P(h,d) related to the volume of the column
describes the volumetric drop distribution as a function of the height h in the
column and the drop diameter d. Integration over the entire drop spectrum
provides the holdup as a function of the height h in the column. For a dispersed
phase fraction with the droplet diameter d at the height h of the column, the
differential equation follows:

δ δ δ δ
P h; d† ˆ ‰P h; d†ud h; d†Š ‡ Dax;d h; d† P h; d† ‡ R h; d† (59)
δt δh δh δh
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 31

where the swelling term R(h,d) is composed of drop breakage Rb(h,d) and coales-
cence Rc(h,dTr):

R h; d† ˆ R‡B h; d† RB h; d† ‡ R‡C h; d† RC h; d† (60)

The individual parts describe the volume gain or loss resulting from the coales-
cence or breakage of drops from the drop fraction under consideration.

5. Process Integration and Applications

5.1. Biologicals. Monoclonal antibodies (mAb) are produced on an indus-

trial scale of up to 1 ton per year. Examples are Rituximab and Etanercept (52). In
general, mAbs are produced in paragraphs, with unit operations (UO), sequen-
tially followed by a storage tank (53,54) between each UO. The manufacturing
process is divided into upstream (USP) and downstream processing (DSP) (55).
Figure 25 shows an exemplary process diagram for the production of mAbs.
In the USP, the target molecule is produced by Chinese hamster ovary (CHO)
cells (57). This cell type is capable of complexly glycosylating mAbs. The animal
cells are increased in the inoculum until the required cell density is reached to
inoculate the fermenter. The production of the antibodies takes place in

TRAIN A TRAIN B TRAIN C

Inoculum

Media prep.

Buffer prep.

Virus Protein a chrom. Ultrafiltration storage tank Microfiltration Centrifuge

Inactivation

Product

Ion-exchange Storage tank Virusfiltration Hydrophobic Storage tank Diafiltration Sterile filtration
chrom. interaction chrom.

Fig. 25. Standard platform process for the manufacturing of mAbs. (Reproduced with
permission from Ref. (56). Copyright 2005, Elsevier.)
32 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

bioreactors with a size of 15–25 m3 (52,58). The cultivation period is between 14

and 21 days (25 days). mAbs are extracellular after fermentation. Therefore, the
cells do not have to be digested before purification, but can be separated directly
from the fermentation broth. Cell separation is usually carried out by centrifuga-
tion or filtration (52). Alternative retention methods are sedimentation, acoustic
filtration, or aqueous two-phase extraction (59,60). A gentle separation of the cells
minimizes the contamination of the product stream with host cell proteins (HCP)
by lysed cells.
The DSP can be divided into capture, intermediate purification, and pol-
ishing. This subdivision is called the three-step cleaning strategy. In the capture
step, the mAb is separated from the fermentation broth and purified. Protein A
chromatography (61) is most commonly used for this purpose. In the subsequent
intermediate purification, aggregates, DNA, and host cell proteins are further
separated using ion exchange chromatography (IEX), for example. The final step
ensures the purity required for pharmaceutical applications. This can be achieved
by hydrophobic interaction chromatography (HIC). After purification, a diafiltra-
tion can be used for the formulation for the final buffer change.
In contrast to intermittent operation, a steady state occurs in continuous
operation. Continuous production can also be achieved by cyclic processes. Con-
tinuous production is not yet widespread in the manufacture of biopharmaceut-
icals. One reason for this is the low production volume of some recombinant
proteins in the range of several kilograms per year. In addition, the product is
defined by the process according to previous regulations (53). A change in produc-
tion would lead to a new and expensive registration procedure. However, the new
definition for the term batch opens up possibilities for the approval of continuous
processes. A batch is defined by the U. S. Food and Drug Agency (FDA) through a
defined quantity of product and not through the manufacturing process (62).
Advantages of continuous production are as follows:

• Reduction of investment costs and plant size

• Higher space-time yield
• Greater flexibility
• Improved product quality (54).

The use of continuous processes is particularly suitable for applications with

poor product stability, because the shorter residence times of the protein dena-
turation counteract. An example of an established continuous process is the
production of Factor VIII, as the product is very unstable (63). In the production
of biosimilars, biotechnologically produced drugs that are approved after the
expiry of the patent period of the original product, a new registration (64) is
necessary. The introduction of a continuous process can lead to cost reductions and
thus to a competitive advantage. Schmidt describes an alternative process, which
enables continuous operation by implementing ATPE as clarification and capture
operation (41) (Fig. 26).
Unlike in the conventional production process of mAb, centrifugation and
microfiltration as harvest and clearance operations are replaced with the outlined
ATPE process. After the filtration trains, the product stream can, depending on the
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 33

Cultivation Precipitation

Bioreactor
Precipitation agent

Polymer tank
Buffer tank

Filtration trains
ATPE

Horizontal settler LP + product

Static mixer iCCC

HP + oils
Fig. 26. Alternative production process based on ATPE as cell harvest operation (41).
(http://www.mdpi.com/2073-4468/6/4/21. Licensed under CC BY 4.0.)

future process strategy, be further purified by precipitation or by integrated

countercurrent chromatography (iCCC) for continuous capture, replacing tradi-
tional protein A chromatography, as well as ion exchange and hydrophobic
interaction chromatography for further purification and polishing.
To evaluate the validity of prediction of settler dimensions, which were done
as presented in Section 4.3, experimental studies were performed at lab and pilot
scale. Figure 27 shows the results of eight continuous settling experiments that
were conducted utilizing a settler with a nominal diameter (DN) of 50 mm (750 mL
holdup). Product yield was, across the experiments, higher than 80%. Cell reduc-
tion was at least 1 log or higher, except for experiments S6 and S7. The reason for
this was a change in flow rate. S6 and S7 were part of a phase inversion study. By
increasing the volume flow, effects on yield, phase separation, and cell clearance
should be determined. The relatively high yields of 90–100% indicate that mass
transfer, as well as phase separation, is not negatively affected by the increase in
flow rate.

2.00
100% 1.80
1.60
Log cell reduction (-)

80% 1.40
1.20
Yield (%)

60%
1.00

40% 0.80
0.60
20% 0.40
0.20
0% 0.00
S1 S2 S3 S4 S5 S6 S7 S8 S1 S2 S3 S4 S5 S6 S7 S8

Fig. 27. Yield and cell clearance for eight continuous settling experiments (41). (http://www.
mdpi.com/2073-4468/6/4/21. Licensed under CC BY 4.0.)
34 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Fig. 28. Side view of the pilot-scale settler (DN150, 12.6 L hold-up) during continuous
settling operation (900 s residence time).

Finally, a pilot-scale operation of the continuous settling process is discussed.

Figure 15 shows the experimental setup. Like in the lab-scale studies, phase-
forming components, after passing through a static mixer pipe, continuously enter
the apparatus. To ensure total sedimentation of the light phase, a residence time of
900 s was adjusted. Due to a small measurement inaccuracy of the flow meters, the
PEG content in the system was lower and the percentage of phosphate rose. This
results in phase inversion, which is also visible in Figure 28. In order to obtain a
predominantly light phase for the filtration train, large parts of the light phase
with the cells were withdrawn through the immersion tube during operation,
which led to a product yield of 80.4%. For better comparability, the same process
was carried out in batch-settling mode simultaneously. The comparisons in yield,
cell reduction, and subsequent filterability are listed in Table 2.
Both operation modes clarified approximately 160 L total volume within the
aimed-at time-frame of less than 3 h. No filter blockage occurred; however, due to
the discussed shift in system composition, the product phase from the continuous
settling process showed a higher yield loss. Also, the necessary filtration pressure
was significantly higher.
Compared to the settler dimensions, modeled only on the basis of batch
settling, it is very close to the calculated 10.9 L settler volume (H540), and in
accordance to the thick dispersion layer significantly smaller than 40 L (H1340),
which would be necessary for a narrow dispersion band (Fig. 29).

Table 2. Comparison of Batch and Continuous ATPE of 160 L System Volume.

Mode Yield IgG (%) Log cell reduction ( ) Pressure increase (bar)
batch 95 1.29 0.1
continuous 80.4 0.44 0.9
The pressure increase refers to 117 L/m2 area-specific filtration volumes.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 35

10,000 10,000
MAV300 H300
MAV540 H540
1000 MAV1340 1000 H1340
Settler volume (L)

Settler volume (L)

100 100

10 10

1 1

0.1 0.1

0.01 0.01
2 4 10 160 4000 24,000 2 4 10 160 4000 24,000
(a) ATPS volume (L) (b) ATPS volume (L)

Fig. 29. Settler volume as a function of residence time for different system scales obtained
for the MAV (a) and simplified Henschke method (b). The dashed lines indicate the
experimentally tested scenarios (from top to bottom): 12.6, 0.75, and 0.175 L settler volumes
for 160, 10, and 4 L ATPS volumes (41). (http://www.mdpi.com/2073-4468/6/4/21. Licensed
under CC BY 4.0.)

In this case study, the development and implementation strategies for

aqueous two-phase extraction as a cell harvest operation were shown. After
defining the operation space for the specific separation task in terms of cell
reduction, yield, and process time by a limited number of shaking flask experi-
ments to obtain equilibrium composition and material data by lab-scale batch and
continuous settling tests, early predictions of necessary settler dimensions are
possible (Fig. 30).
The batch-settling experiments, in particular, enable the identification of the
separation times that are necessary to obtain the desired sedimentation profile
inside the settler. Knowledge about the sedimentation rates can then be used for
more sophisticated calculation of settler dimensions. For 3 h, continuous

Separation Material Process Integration

task data concept scale-Up

Experiment Shaking flask Phase control concept

Batch-settling exp Comparison

Cont.-settling exp Measurement and control

Parameter

Comparison

Construction

Model Mav
Mistry
Henschke
Settler design Final apparatus

Fig. 30. Process development strategy (41). (http://www.mdpi.com/2073-4468/6/4/21.

Licensed under CC BY 4.0.)
36 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

processing of up to 12,000 L cultivation broth, a settler with a length of 3.6 m, and a

volume of approximately 1.4 m3 is proposed.
The investigated operating point is characterized by a yield of between 80 and
close to 100%, as well as up to 2 log step reductions in cell number. For small system
volumes (2–10 L), the calculated settler dimensions are consistent with laboratory
tests as well as with pilot-scale volumes of up to 160 L cultivation broth. In order to
achieve the above-described yield, the correct system composition must be observed.
In the case of deviations, this may lead to a reversed direction of phase inversion. Since
the heavy phase must be dispersed for a reduction of cells, this is of utmost importance.
5.2. Botanicals. For the next few years, an increase in the demand for
drugs of plant origin is predicted. Today, a total of around 12,000 plants are known
worldwide which form pharmaceutically active metabolites and are thus of inter-
est to the pharmaceutical industry. These account for a total of about 30% of the
active ingredients currently used. Examples of phytopharmaceuticals are the
analgesic morphine, which was originally isolated from opium, or the cardiac
glycoside digoxin, which is extracted from the fungi thimble. In addition, plant
ingredients are used in food supplements and cosmetic products (65–68).
Another very important phytopharmacon is the antimalarial drug artemi-
sinin, which is obtained from the annual wormwood (Artemisia annua L.). It has
proven to be the most effective and tolerable drug against the multiresistant
pathogen of malaria tropica. Malaria is one of the most frequent and serious
infectious diseases in the tropics and affects 216 million people a year with around
450,000 deaths per year. Not all ill people can be treated today because not enough
artemisinin is produced at low prices (65,66,69).
The production of artemisinin begins with the cultivation of the plant
Artemisia annua L. This is followed by solid–liquid extraction (SLE) from the
dried leaves and young bloom buds of the plant material (70). This is followed by
purification with different separation processes (71–74):

• Liquid–liquid extraction
• Chromatography
• Adsorption by activated carbon
• Crystallization

Following the patented manufacturing process, as can be seen in Figure 31,

an extractive purification path in which hot nonpolar solvent is used to extract
directly from the plant material (71,75). Common solvents used are n-hexane,

Waste Solvent Waste Waste/

biomass recycling precipitants recycling Ethanol
Hexan

Dryed plant
Fresh solvent Raw extract Crystals Kristalle
Artemisinin
Steam Artemisinin Artemisinin

Hexan/
Crystallization Crystallization
ethylacetat Evaporation Waschschritt Chromatography
#1 #2
extraction

Fig. 31. Patented process for the manufacturing of artemisinin.

APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 37

Upstream Clarification Capture Purification Polishing

Solvent Solvent
Percolation Precipitation LLE change Chromatography change Crystallization

Fig. 32. Alternative process for the manufacturing of artemisinin.

petroleum ether, toluene, and chloroform (76–79). Some of these are modified with
ethyl acetate to increase the solubility of artemisinin. The dried, cut plant material
is available as a packed bed in a column and the solvent flows through it. The SLE
runs in batches (75). Subsequently, the raw extract is evaporated in an evaporator
up to 90% by volume. The raw artemisinin is then crystallized and washed with a
warm solvent. The target component is to be separated from plant components
such as waxes, essential oils, or chlorophyll. The washed raw artemisinin is now
recrystallized several times with ethanol and water in the presence of active
carbon adsorbent. This is followed by a chromatographic fine purification of the
crude extract (72,75). Process A is now implemented in China and Vietnam (65).
The alternative process in the present study however achieves high levels of
purity by integration of extraction and scrub columns. Therefore, the process
development strategy as presented in Figure 32 was applied. The process adds
several changes to the standard process, but for the sake of conciseness, only the
results of LLE are discussed.
Extensive solvent screening resulted in toluene as the best solvent, as it has a
large mixing gap with water/acetone at a high boiling point. In addition, screening
with butyl acetate and MTBE as alternatives to toluene was performed. In the
shaking tests, toluene has the highest partition coefficients and separation factors,
followed by butyl acetate. Here, MTBE was excluded as solvent for LLE because it
has a low boiling point and the smallest mixing gap with water/acetone. Therefore,
only a low concentration of artemisinin extract would be possible with MTBE.
Sedimentation tests showed no significant difference between toluene and butyl
acetate, therefore, additional settling tests were carried out. In these, the solvent
and the SLE extract are dispersed together and the settling time is then absorbed.
Butyl acetate has a significantly shorter settling time, which reduces the risk of
emulsion formation in the column tests (Fig. 33).
For this reason butyl acetate was selected for the process. In the next step, the
mass transfer value for the artemisinin is determined by experiments on the drop
measuring cell and tests are carried out to determine the solvent ratio in the salt
wash. At the operating point of the Kühni column, a phase ratio of 1:6 is set and the
concentration of artemisinin in the extract phase increases with a lower risk of
flooding of the Kühni column (Fig. 34).
The Na2CO3 wash then removes a large number of polar secondary compo-
nents, as well as the previously skipped acetone. Here, the purity can be signifi-
cantly increased by increasing the volume flow of the salt phase.
38 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Type kühni ≈ 1/5.3

Phase ratio
LP: n-butylacetat (d) ≈ 1/2.65

HP: SLE-extrakt (c) Loading B = 10.2 m³/m²/h

(d) (c) Stirrer rate n = 100 min-1
LPout
HPin 4.00
(SLE-E.) 3.50

3.00 Aceton
Height (m)

2.50

2.00

1.50 Feed 1:5,3

Extract
1.00 Raffinate

0.50
LPin
(BAC) 0.00
Water n-butyl acetat
HPout
0.0 0.5 1.0 1.5 2.0 2.5
Concentration (g/L)
Continuous (HP) Dispersed (LP)

Fig. 33. Concentration profile of the extraction column.

TypeMontz Pak B750 ≈ 1/2.65

PhaseRatio
LP: n-butyl acetate (d) ≈ 1/7.14

HP: 5 wt-% Na2CO3 (aq) (c) Loading B = 23,4 m³/m²/h

(d) (c) Pulsation = 0 mm/s
LPout Aceton
4.0
HPin
(Na2CO3) 3.5

3.0
Feed
Height (m)

2.5

2.0

1.5
Extract
Raffinate
1.0
Water n-butyl acetate
LPin 0.5
(BAC/Ace.)
0.0
Dispersed (LP)
HPout
0.0 0.2 0.4 0.6 0.8 1.0
Purity (-)
Fig. 34. Column profile for the purity of artemisinin during the scub process.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 39

Finally, over both columns a purity of 90% of initial 40% and a concentration
of over 6 g/L of initial 0.9 g/L can be achieved.
5.3. Metals. There are many different process strategies for separating
rare earths. After mining and grinding, the ore is refined by flotation, magnetic, or
gravity separation processes and then processed as a rare earth concentrate in
hydrometallurgical processes to obtain the pure rare earth metals or rare earth
metal compounds. The process strategy is usually standardized. First of all, the ore
is broken down and leached. Subsequently, simple secondary components, such as
iron or cobalt, are removed from the system by precipitation. After a clearing step,
the individual REE compounds are then separated by multistage solvent extrac-
tion or, in some cases, by ion exchange columns (80,81). An example process is
shown in Figure 35. Nowadays, solvent extraction is the most commonly used

Monazite

NaOH Decomposition

Washing
Na3 PO4 Radioactive
(U, Th, RE) (OH)x waste treatment

HCl Dissolution

RECl3

Solvent Sludge
D2 EHPA extraction dissolution
by HNO3
Solvent TBP
La, Ce, Pr, Nd Sm, Fu, Gd Tb · · · Lu extraction

REO Solvent Solvent Solvent

extraction extraction extraction Th(NO3)4

REF3 Eu2 O3 Y2 O3 ThO2

Polishing La2 O3
powder Phosphor U3 O8
Pr6 O11 Solvent
RE2 (CO3)3 Solvent extraction (NH4)2 U2 O
Nd2 O3 extraction

Gd2 O3 Tb4 O7
CeO2
Sm2 O3
RECl3 Dy2 O3
CeCl3
R&D
Molten
salt Molten Lu2 O3
electrolysis salt
Mischmetal electro- SmCo powder
Lighter lysis
flints
Ce metal

Fig. 35. Typical separation process for rare earth metals. (Reproduced with permission
from Ref. (82). Copyright 1982, Elsevier.)
40 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

method of separating rare earths. In the simplest case, the basic function of liquid–
liquid extraction is based on the different physical distribution of the target
component between extract and refining phase. Extractions are distinguished
according to their characteristics in physical extractions, dissociation extractions,
and reactive extractions. The properties of the rare earths as a group, and thus also
their distribution behavior, differ significantly from those of other metals. There-
fore, they can usually be isolated as a group in a few first steps. However, the
different REEs are very similar to each other. This means that even with an
optimal choice of solvents, there are usually only minor separation factors. A multi-
stage purification countercurrent is therefore essential to keep the individual REE
components separate. Batteries with several hundred mixer separators are therefore
frequently used industrially (80,81). The choice of solvent directly influences the
distribution properties of the different components. Attached to this are the achieva-
ble separation factors and, finally, also the minimum number of separation stages,
which correlates directly with the final plant size. The choice of the optimal extraction
agent is therefore of enormous importance for the resulting investment and operating
costs. Complexing agents and phosphorus-based ion exchangers are the most impor-
tant extractants. While complexing agents are frequently investigated in university
research due to their high selectivity, only phosphorus-based extractants (80) are used
industrially due to their lower operating costs.
Ion-exchange chromatography has been used since the 1950s for separating
rare earths, but has been largely replaced by solvent extraction at the latest in the
1960s (80,83). Since then, however, the technical maturity of these basic opera-
tions has been significantly improved in analytics, biotechnological processes, and
the food industry (84). The potential of state-of-the-art ion exchange chromatog-
raphy is therefore particularly high for separating the REE.
Helling described a case study based on experimental and literature data
(85). Different processing concepts are generated and optimized by process model-
ing. Finally, a cost calculation and sensitivity analysis point out the potential for a
combined processing concept of chromatographic and reactive liquid–liquid extrac-
tion processes and main cost drivers are indicated (85). In the study, a stage model
is used with distribution coefficients from literature (86). Rare earth oxides are
solved in 1M HClO4 or NaClO4, respectively, and are therefore in the water phase.
0.2M HTTA (2-thenoyltrifluoroacetone) dissolved in CCl4 is applied as organic
counter phase. The liquid–liquid extraction of metal oxides is implicitly carried out
by chelate formation. Therefore, the pH value has a critical influence on the
extraction performance. In acidic milieu, a logarithmic dependency between the
distribution coefficient and H+ ion concentration could be determined. For high
distribution coefficients (pH value between 3.5 and 5), a deviation from this
linearity was detected. This is caused by the complex formation of metal ions
in aqueous phase, by trace impurities or mutual solubility of the phases (86). For a
constant pH value, an extraction efficiency, which is defined here as the difference
of the distribution coefficients, in dependence of ion radius is detected. Thereby,
the distribution coefficient increases with higher number of atoms (87). For the
liquid–liquid extraction process of metal ions, reactive extraction phenomena
should be taken into account (13).
This can be done by the use of process models, which consider the recycling of
internal phase and wash (or scrub) to increase the product yield and purity (33). In
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 41

Wash solvent Backextraction

solvent

Solvent
recycling W RE

Backextract
Feed M

E Solvent
recycling

Raffinate

Fig. 36. Extraction process. (Reproduced with permission from Ref. (85). Copyright 2013,
John Wiley & Sons.)

Figure 36, the total extraction process with corresponding recycling streams is
shown schematically. In this study, the conceptual process design and the eco-
nomic evaluation start with constant distribution coefficients for the determina-
tion of necessary stage number and column diameter. In order to keep optimal
distribution and separation factors locally constant, technically an addition of acid
and base at different separation stages is necessary. The model-based process
design is realized for a multistep extraction process with wash/scrub for purity
increase and re-extraction for extraction media recycling. The innovation potential
of this work is a sufficient separation of the rare earth elements by a minimized
number of separation units at extremely high purity demands. Most patents and
publications use a different feed mixture that is easier to separate than in this
example. Furthermore, the target concentrations of the rare earth elements differ
according to their applications. In this study, the separation of Eu2O3 and Tb4O7 is
the most difficult caused by their high similarity: close retention factor in chro-
matography as well as small differences in their distribution coefficients for
liquid–liquid extraction. In general, different processes, such as different leaching
and crystallization/precipitation types or thermal treatments, are used (88), but in
most cases no information is provided on the final concentrations. Liquid–liquid
extraction is mostly used as a mixer–setter assembly, whereas chromatography
systems are mainly used for analytical purposes only. Examples for the chromato-
graphic separation of rare earth elements on an industrial scale are published and
patented, but usually only for simpler separation tasks. In addition to the scal-
ability of the chromatographic processes, the feasibility of liquid–liquid extraction
in a countercurrent column is investigated. This is currently not the state of the art
for a complete separation of rare earth elements, due to the low distribution
42 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Fig. 37. Processing alternatives for recycling of rare earth compounds. (Reproduced with
permission from Ref. (85). Copyright 2013, John Wiley & Sons.)

coefficients. The advantage of an extraction column in comparison to mixer/

separator batteries is the high number of theoretical stages that can be achieved.
A process concept for the processing of a rare earth recycled feed is proposed by
Helling et al., which is based on the distribution coefficients from the literature and
the selectivity of ion exchange chromatography determined experimentally. This
method can lead to a first conceptual process design (89). In this study, two process
alternatives have potential to separate the feed mixture within specifications. In
Figure 37, both processing strategies as well as the potential purification per-
formance are summarized.
In comparison to all existing alternatives, the number of required unit
operations is reduced and the applicability of chromatographic separations as
well as the liquid–liquid extraction in a column is demonstrated. Thus, liquid–
liquid extraction can be implemented for the specific separation of Eu2O3 and
Tb4O7 with residual cross-contamination at the low ppm level. This alternative to a
variety of leaching processes provides access to novel process alternatives, eg, in
combination with chromatography.
The economic analysis and evaluation of the processing alternatives consid-
ers a scale up to production scale. Scale up of chromatographic units is based on
experiments while scale up of liquid–liquid extraction processes is based on the
stage model. At constant column length, the corresponding column diameter is
calculated from the feed amount to be separated with column diameters techni-
cally limited to approx. 2m ID. If larger adsorbent amounts are necessary, multiple
parallel chromatography plants are considered. The loading volume is determined
by injection volume and concentration based on experimental runs. Elution time of
the target fraction is fixed to 10 column volumes by experimental results.
With the use of process modeling and algorithmic optimization for the liquid–
liquid extraction process, Helling and co-workers have found a number of
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 43

Fig. 38. Concentration profiles of liquid–liquid extraction and scrub. (Reproduced with
permission from Ref. (85). Copyright 2013, John Wiley & Sons.)

theoretical separation stages of about 18 for extraction. Seventeen steps are

necessary for washing and 15 further steps for the re-extraction of the target
component. The technical implementation can be carried out by three extraction
columns with a total height of 8–15 m. The height of the columns is strongly
dependent on the internal structure, ie, the use of rotating or static internals. The
phase ratio between the aqueous and organic phase is optimized to 1.1:1. For the
re-extraction, a phase ratio of 4:1 must be selected, resulting in approximately 15
theoretical separation stages. The raffinate stream from the extraction stage is
returned to the column head of the re-extraction. In Figure 38, concentration
profiles of extraction and wash/scrub are plotted.
The study by Helling (85) shows a significant economic profit potential for the
recycling of rare earths. Based on literature data as well as experimental results, a
processing concept was generated, which enables purification of waste to required
specifications for the different lighting phosphors. Two scenarios out of their study
show the highest economic potential based on cost analysis. This cost analysis also
serves to identify the process parameters with the highest impact on the economy
of the proposed processing concept. Essential criteria are the recycling amount of
salt buffer for the implemented chromatographic separations, and the setup of
three extraction columns for extraction, wash/scrub, and re-extraction. A combi-
nation of liquid–liquid extraction followed by each chromatographic separation for
organic and aqueous phase shows the highest potential for this separation task
(Fig. 39).
44 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Scenario 1
70

Scenario 2
Maintenance costs
Personal costs
60 Operating costs
Invest costs
50
Relative costs [%]

Scenario 2
40
Scenario 1

30

Scenario 1

Scenario 2

Scenario 2
Scenario 1
20

Scenario 1

Scenario 2
10

0
Chromatography 1 Chromatography 2 Liquid-Liquid extraction Drying Precipitation
Unit operation

Fig. 39. Cost distribution for both scenarios. (Reproduced with permission from Ref. (85).
Copyright 2013, John Wiley & Sons.)

6. Current Development and Trends

As the environmental impact of the chemical and energy industries has been
increasing for several decades, society expects scientists and engineers to design
sustainable chemical processes, produce fewer hazardous substances, and utilize
more environment-friendly resources (90). As an economical separation technol-
ogy, liquid–liquid extraction is an energy-efficient process and predestined for this
task. However, many of the organic solvents involved are toxic and highly
flammable (volatile organic compounds, VOCs). Various approaches can be taken
to improve the safety and environmental friendliness of the LLE processes. The
most important one are as follows:

1. The replacement of traditional organic solvents with green solvents such as

natural deep eutectic solvents (NADES), ionic liquids (ILs), or aqueous two-
phase systems (ATPS).
2. Reduction of solvent consumption in general through process intensification.

Both approaches are presented in the following sections.

6.1. Green Solvents. The increasing recognition of the advantageous
properties of deep eutectic solvents have led to a growing interest in using these
mixtures as an alternative to conventional organic solvents and to their predeces-
sors ionic liquids in numerous and diverse areas in recent years (91). They are
defined as systems consisting of a mixture of two or more components, but at least
one hydrogen-binding acceptor (HBA) and a hydrogen-binding donor (HBD). The
mixture forms a eutectic phase, characterized by a lower melting point below
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 45

100°C and lower than that of the individual components (92). Different NADES are
typically categorized according to their properties and their solution behavior into
four basic types (93):

• Type I (quaternary salt and metal halide)

• Type II (quaternary salt and hydrated metal halide)
• Type III (quaternary salt and hydrogen bond donor)
• Type IV (metal halide and hydrogen bond donor)

By now their application in extraction processes covers a wide field, including

extraction of proteins (94), botanicals (95), metal ions (96), organics (97), and
biofuel molecules (98) as well as in combination with other green processes such as
supercritical CO2 extraction (99). The main advantage of these alternative sol-
vents is their differentiation from classic organic solvents, such as sustainable
recovery, biodegradability, and nontoxicity (100). However, more and more posi-
tive properties are being discovered, such as biocompatibility and its stabilizing
properties for active pharmaceutical ingredients (APIs). The use of green solvents
also enables the development of completely new extraction strategies for by-
products or waste products, such as the extraction of glycerol (101) from biodiesel
or oryzanol by deacidification of palm oil (102).
6.2. Process Intensification. In addition to classical pinch analyses for
optimal recycling of various forms of energy in the process and energy and solvent
recovery as an overall concept, it is also necessary to consider the exchange or at
least the expansion of traditional apparatuses in favor of more energy-efficient
process concepts.
One concept to be understood in this context is process intensification, which
in a few words aims to drastically reduce the effort of a chemical process while
maintaining or even better, increasing the benefits. Efforts can be understood as
material and energy costs as well as the dimension of the plant itself. The benefits,
in turn, are defined by many aspects that have to be evaluated differently
depending on the case. Examples are increased mass transport of mass transport
limited reactions, but also gentler production conditions of sensitive active sub-
stances in the pharmaceutical industry (103). The different approaches toward
process intensification are summarized in Figure 40. However, one of the results of
this argumentation is the great research interest in micro- and millistructured
apparatuses (104,105).
Microprocess engineering can contribute in different ways to the overall
process of intensification and have a significant impact on the following points
(103):

• Shorter time to market due to faster transfer from laboratory to production

scale
• Thus, increased flexibility
• Lower operating costs
• Increased yield due to fewer by-products
• New and better product properties
46 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

Categories and working levels Examples of process intensification tools

• reactive mixer, jet injection reactor...

Chemical • microreactor, wall reactors...
reactions • Spaltrohr reactor, rotating disc reactors...
• nanoscale catalysts, multifunctional catalysts, biocatalysis...
Equipment
• static mixers
Physical and • splitting ring mills, ultra dispergators.....
thermal operations • compact heat exchangers
• adiabatic compression/expansion

• heat exchanger reactors…

Multifunctional
• reactive distillation…
methods • catalytic wall reactor…
• ultrasound
Process Alternative • plasma, microwaves and laser technologies
methodes energies • fuel cells
• electric fields

• supercritical media
Special methods • solvent antisolvent processes
• submicro emulsion/dispersion.

• partial substitution by process‐intensive units

Reengineering of
• partial modularization, miniaturization...
existing plants • IP protection and benchmark control
Plant
design • adaptation to subregional raw material and energy supply
Replacement and • modularization, miniaturization...
green investion • consideration of regional boundary conditions
• (use of national funding programmes)

Fig. 40. Examples of process-intensive tools. (Reproduced with permission from Ref. (106).
Copyright 2005, John Wiley & Sons.)

• Inherent security
• Reduced recycling and waste streams
• Process-integrated environmental protection

While devices based on microeffects are increasingly accepted in reaction

engineering, there is still no breakthrough for separation applications like liquid–
liquid extraction. Additionally, the limited throughput potential, mostly caused by
complex fluid dynamics and difficult phase separation, is problematic (107,108).
The implementation of milliscale devices in engineering processes is a new
and steadily growing field. In contrast to conventional industrial techniques, a
milliprocess is characterized by improved mass and heat transfer and small
equipment size. The majority of current applications are established in the field
of reaction technology. However, the transfer of knowledge from these scientific
focuses to separation technology is performed hesitantly. A concept in this context
is membrane-supported extraction. Wellsandt published a milliextractor for the
separation of organic/organic systems as wells as for ATPS that operated on a scale
of 1–2 L/h and also showed how the concept of developing the process and the
crucial parameters for equipment design (Fig. 41).
6.3. Process Models for Multicomponent Reactive Extraction. The
trend in the industry is toward biotechnological applications such as the purification
of cultivation broths. This industrial sector is characterized by large-scale annual
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 47

Fig. 41. Process development strategy for new separation devices. (Reproduced with
permission from Ref. (109). Copyright 2015, John Wiley & Sons.)

production and complex separation tasks. Due to the need for gentle processing,
mainly aqueous systems are used, which are characterized by low differences in
density and interfacial tension (110,111). In addition, these systems are often more
viscous than the ECFE standard test systems. The combined occurrence of these
characteristics makes the design even more difficult, making more pilot plant tests
necessary. Due to the low feed quantities (a few litres instead of 100–1000 L) in the
early development phase of biotechnological products, there is a serious limitation in
process development. Scaling from the laboratory to the industrial column on the
basis of process models has not yet been researched for these systems. However, the
use of reliable process models is indispensable for competitive processes based on
green technology due to the problems already described (112).
In addition to the equilibrium data for thermodynamic design (number of
stages and washing agent/feed quantities) and hydrodynamics, information on
mass transfer kinetics is required for the design of extraction apparatus. Due to the
increasing use of dissociation and reactive extraction in addition to classical
physical extraction, it must also be possible to describe reaction equilibrium
and kinetics. In the project “From single drop to extraction column”, a knowl-
edge-based approach based on the results of laboratory tests at standardized test
systems was demonstrated for the design of extractors. Hydrodynamics and mass
transfer studies on the three EFCE test systems toluene/water/acetone, n-butyl
acetate/water/acetone, and isotridecanol/water/1-propanol provided the basis for a
model concept for column design. Investigations were carried out on individual
drops in a standard apparatus with subsequent validation of the modeling results
obtained from the parameters in tray and packing columns as well as stirred
columns. As a result, the simulation programs ReDrop and LLECMOD were
further developed (16,113,114) (Fig. 42).
The project provided an efficient design method for physical extraction
columns. There is therefore still a general need for research into more complex
48 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

State of the art

Industry Academics
(Pilot plant based) (Model based)

Shaking Flask Shaking Flask (Equilibrium)

Laboratory (Equilibrium) (Single-) Drop Experiments
(Hydrodynamics & Mass Transfer)
Model Paramter Determination

Hydrodynamics & Mass Transfer Feasibility Test up to DN200

Pilot DN 50/60, evtl. DN150
Feed: Min. 800 L, up to 4000 L
for ECFE Standard Test
Systems shown

Design and Construction of Design and Construction of

> DN1000 Finale Plant Production Scale Plant Based
on Lab Scale Model Parameter
Only Not Possible

Fig. 42. State-of-the-art extraction column design.

dissociative and reactive extractions and multicomponent systems. Some working

groups are currently working on transfers of the basic concept for column design
based on single drop measurements using the example Zn-DEHPA as a test system
for metal salt extractions (115). However, future work must focus on the
separation of real complex multicomponent mixtures, eg, from fermentation
processes and phytochemicals following solid extraction. It makes sense to
extend the conceptual approach to multicomponent systems in the fine chemical
and pharmaceutical industries in order to meet the increased requirements for
efficient process development and optimized production processes. This objective
does not primarily require an apparatus design as previously developed, but
above all the design and conception of complete reinforcement extraction pro-
cesses consisting of extraction, scrub, and re-extraction as well as solvent/
auxiliary recycling. Complex mixtures are defined as a mixture of many compo-
nents of different substance classes with partly unknown structure and sub-
stance data. The task is to isolate a target component from very similar
molecular secondary components with sufficient purity and yield under economic
conditions. The processes already shown for monoclonal antibodies, artemisinin,
and rare earths provide a good insight into the research topics of the coming
years.
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 49

BIBLIOGRAPHY

CITED PUBLICATIONS

1. R. Goedecke, Grundlagen, Methodik, Technik, Praxis, Wiley-VCH Verlag GmbH,

Weinheim, Germany2006.
2. A. Mersmann, M. Kind, and J. Stichlmair Grundlagen und Methoden. Springer, Berlin,
2005. (2., wesentlich erw. und aktualisierte Aufl.)
3. J. Gmehling, B. Kolbe, M. Kleiber, and J. Rarey, Chemical Thermodynamics for Process
Simulation, Wiley-VCH Verlag GmbH, Weinheim, Germany, 2015.
4. R. E. Treybal, Liquid Extraction, 2nd. ed., McGraw-Hill, New York, 1963.
5. K. Sattler and H. J. Feindt, Principles and Design, Wiley-VCH Verlag GmbH, 2010.
6. M. Baerns, Technische Chemie, 2nd ed., John Wiley & Sons, Inc., Hoboken, 2013.
7. A. Einstein,Über die von der molekularkinetischen Theorie der Wärme geforderte
Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen, 1905.
8. W. R. A. Vauck and H. A. Müller, Grundoperationen chemischer Verfahrenstechnik. Dt.
Verl. für Grundstoffindustrie; Wiley-VCH Verlag GmbH, Stuttgart, Weinheim, 2000.
9. M. Kraume, Grundlagen und apparative Umsetzungen. Springer Vieweg, Berlin, 2012.
10. W. K. Lewis and W. G. Whitman, Ind. Eng. Chem. 16(12), 1215–20 (1924).
11. Lohrengel B. Einführung in die thermischen Trennverfahren. Trennung von Gas-,
Dampf- und Flüssigkeitsgemischen. München: Oldenbourg. 2., überarb. Aufl. 2012.
12. A. S. Kertes and F. Grauer, J. Inorg. Nucl. Chem. 40(10), 1781–82 (1978).
13. J. Leistner, Modellierung und Simulation physikalischer, dissoziativer und reaktiver
Extraktionsprozesse. Zugl., Dortmund University, Dissertation, 2005, Shaker, Aachen.
14. W. S. Wise and D. E. Williams, The Less Common Means of Separation, 1964.
15. K. Schügerl and co-workers, Chemie Ingenieur Technik. 58(4), 308–317 (1986).
16. H.-J. Bart and co-workers, Chemie Ingenieur Technik 78(5), 543–47 (2006).
17. K.-H. Reissinger, J. Schröter, and W. Bäcker, Chemie Ingenieur Technik 53(8), 607–14 (1981).
18. Godfrey J. C., ed., Liquid-Liquid Extraction Equipment, John Wiley & Sons Ltd,
Chichester, 1994.
19. R. Gayler, N. W. Roberts, and H. R. C. Pratt, Trans. Inst. Chem. Eng. (31), 57–68 (1953).
20. C. Sternling and L. Scriven, AIChE J. 5(4), 514–23 (1959).
21. M. Modigell, Untersuchung der Stoffübertragung zwischen zwei Flüssigkeiten unter
Berücksichtigung von Grenzflächenphänomenen. Dissertation, Technische Hoch-
schule, Aachen, 1981.
22. M. Qi, Untersuchungen zum Stoffaustausch am Einzeltropfen in flüssigkeitspul-
sierten Siebboden-Extraktionskolonnen. Clausthal, Techn. Univ., Diss., 1992.
23. H. W. Brandt, K.-H. Reissinger, and J. Schröter, Chemie Ingenieur Technik 50(5), 345–
54 (1978).
24. S. Weiss, K.-E. Militzer, and K. Gramlich, Mit 54 Tabellen. Dt. Verl. für Grundstof-
findustrie; Wiley-VCH Verlag GmbH, Leipzig, Stuttgart, Weinheim, 1993.
25. B. Hoting, Untersuchungen zur Fluiddynamik und Stoffübertragung in Extraktios-
kolonnen mit strukturierten Packungen. Zugl.: Clausthal-Zellerfeld, Techn. Univ.,
Diss., 1995. Düsseldorf: VDI-Verl. Als Ms. gedr 1996.
26. H. Haverland and co-workers, Chem. Eng. Technol. 10(1), 151–57 (1987).
27. Y. Marcus, A. K. Sengupta, and J. A. Marinsky, Ion Exchange and Solvent Extraction,
Marcel Dekker, New York, 2002.
28. R. Marr, Chemie Ingenieur Technik 50(5), 337–44 (1978).
29. R. A. Leonard, Sep. Sci. Technol. 23(12–13), 1473–87 (1988).
50 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

30. M. J. Hampe, Chemie Ingenieur Technik 57(8), 669–81 (1985).

31. A. Pfennig, W. Bäcker, and P. Deckert, Vom Einzeltropfen zu Extraktionskolonne -
Anforderungen und neue Entwicklungen. 589. DECHEMA-Kolloquium, Fran-
kfurt am Main. 2005. Available at http://processnet.org/processnet_media/
VomEinzeltropfenzurExtraktionskolonne.pdf.
32. J. Schröter, W. Bäcker, and M. J. Hampe, Chemie Ingenieur Technik 70(3), 279–83 (1998).
33. J. K. Eggersgluess, S. Both, and J. Strube, Chim. Oggi/Chem. Today 30(4), 32–36 (2012).
34. J. Eggersglüß, 2014. Entwicklung und Auslegung von Flüssig-Flüssig Bioextraktion-
sprozessen. Antikörperaufreinigung mit wässrigen Zweiphasensystemen. Zugl., Claus-
thal, Techn. Univ., Diss., 2013 u.d.T.: Prozessentwicklung und Auslegung von Flüssig-
Flüssig Bioextraktionsprozessen: Antikörperaufreinigung mit wässrigen Zweiphasen-
systemen. Aachen: Shaker.
35. M. Henschke, Chem. Eng. J. 85(2–3), 369–378 (2002).
36. S. A. K. Jeelani and S. Hartland. Ind. Eng. Chem. Res. 37(2), 547–554 (1998).
37. C. Nadiv and R. Semiat, Ind. Eng. Chem. Res. 34(7), 2427–2435 (1995).
38. M. Henschke and A. Pfennig, AIChE J. 45(10), 2079–2086 (1999).
39. J. Ayesterán and co-workers, Chem. Eng. Technol. 38(10), 1894–1900 (2015).
40. M. Henschke, Dimensionierung liegender Flüssig-flüssig-Abscheider anhand diskonti-
nuierlicher Absetzversuche. Aachen, Techn. Hochsch., Diss., Zugl., 1995. 1994. Düssel-
dorf: VDI-Verl. Als Ms. gedr.
41. A. Schmidt and co-workers, Antibodies 6(4), 21 (2017).
42. R. Bauer, Die Längsvermischung beider Phasen in einer gerührten Fest-Flüssig
Extraktionskolonne. Disseration, RTH, Zürich, 1976.
43. S. Stemerding and F. J. Zuiderweg, Chem. Eng. 168, 156–160 (1963).
44. P. V. Danckwerts, Chem. Eng. Sci. 2(1), 1–13 (1953).
45. K. J. Valentas and N. R. Amundson, Ind. Eng. Chem. Fund. 5(4), 533–542 (1966).
46. K. J. Valentas, O. Bilousn and N. R. Amundson, Ind. Eng. Chem. Fund. 5(2), 271–279 (1966).
47. C. A. Coulaloglou and L. L. Tavlarides, Chem. Eng. Sci. 32(11), 1289–1297 (1977).
48. R. B. Olney, AIChE J. 10(6), 827–835 (1964).
49. G. Casamatta, Comportement de la population des gouttes dans une colonne d’ex-
traction. Transport, rupture, coalescence, transfert de matire. Dissertation, Institut
National Polytechnique de Toulouse, Toulouse 1981.
50. G. Casamatta and A. Vogelpohl, Chemie Ingenieur Technik 56(3), 230–231 (1984).
51. S. D. Al Khani, C. Gourdon, and G. Casamatta, Ind. Eng. Chem. Res. 27(2), 329–333 (1988).
52. B. Kelley, Biotechnol. Prog. 23(5), 995–1008 (2007).
53. A. Jungbauer, Trends Biotechnol. 31(8), 479–492 (2013).
54. A. L. Zydney, Biotechnol. Bioeng. 113(3), 465–475 (2016).
55. H. Chmiel, Bioprozesstechnik, 3rd ed., Spektrum Akademischer Verlag, Heidelberg, 2011.
56. S. Sommerfeld and J. Strube, Chem. Eng. Process. Process Intensification. 44(10),
1123–1137 (2005).
57. D. C. Andersen and D. E. Reilly, Curr. Opin. Biotechnol. 15(5), 456–462 (2004).
58. P. Gronemeyer, R. Ditz, and J. Strube, Bioengineering 1(4), 188–212 (2014).
59. P. Gronemeyer, R. Ditz, and J. Strube, Biochem. Eng. J. 113, 158–166 (2016).
60. H.-J. Henzler, Chemie Ingenieur Technik 84(9), 1482–1496 (2012).
61. P. Gagnon, J. Chromatogr. A 1221, 57–70 (2012).
62. FDA, Code of Federal Regulations Title 21. Section 210.3. 2015. Available at https://
www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfCFR/CFRSearch.cfm?fr=210.3
63. J. H. Vogel and co-workers, Biotechnol. Bioeng. 109(12), 3049–3058 (2012).
64. J.-L. Prugnaud and J.-H. Trouvin, A New Generation of Biologics, Springer, New York,
2013.
65. D. Litscher, Akupunktur & Aurikulomedizin. 1, 28–31 (2016). https://www.
springermedizin.de/artemisinin/10021954
APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION 51

66. V. Dhingra, K. Vishweshwar Rao, and M. Lakshmi Narasu, Biochem. Educ. 27, 105–
109 (1999).
67. H. J. Bart, and co-workers Phytoextrakte - Produkte und Prozesse. Vorschlag für einen
neuen, fachübergreifenden Forschungsschwerpunkt 2012.
68. Schicketanz P. Phytopharmaka. 2016. Available at www.das-pta-magazin.de.
69. WHO, World Malaria Report 2017. Switzerland, 2017.
70. G. Wernsdorfer and W. H. Wernsdorfer, Malariamittel aus Pflanzen. 1991. Available
at www.zobodat.at/pdf/MOGTP_13_0287-0300.pdf
71. U.S. Pat. 4,952,603 (1990) H. N. ElSohly and F. S. ElFeraly.
72. D. K. Klayman and co-workers, J. Nat. Prod. 47, 715–717 (1984).
73. A. F. Rodrigues and co-workers, Quim. Nova. 29, (2006). doi: http://dx.doi.org/10.1590/
S0100-40422006000200030
74. U.S. Pat. US 6,685,972 B2 (2004) S. Kumar, K. Gupta, D. Singh, M. Gupta, and A. P.
Kahol.
75. A. A. Lapkin, P. K. Plucinski, and M. Cutler, J. Nat. Prod. 69(11), 1653–1664 (2006).
76. N. Acton and D. K. Klayman, Planta Med. 51, 441–442 (1985).
77. H. N. ElSohly, E. M. Croom, and M. A. ElSohly, Pharm. Res. 4(3), 258–260 (1987).
78. J. C. Laughlin, Trans. R. Soc. Trop. Med. Hyg. 88, 21–22 (1994).
79. H. Woerdenbag and co-workers, Planta Med. 57(S 2), A93–A93 (1991).
80. I. McGill, Ullmann’s Encyclopedia of Industrial Chemistry. Wiley-VCH Verlag GmbH,
Weinheim, Germany, 2000.
81. F. Xie and co-workers, Minerals Eng. 56, 10–28 (2014).
82. Zang ZB,. and co-workers, Hydrometallurgy 9(2), 205–210 (1982).
83. C. Persiani and co-workers, Sep. Sci. 2(6), 789–796 (1967).
84. H. Schmidt-Traub,M. Schulte, and A. Seidel-Morgenstern (eds.), Preparative Chroma-
tography. Wiley-VCH Verlag GmbH, Weinheim, Germany, 2012. (vol. 2., completely
revised and updated ed.)
85. C. Helling and J. Strube, Chemie Ingenieur Technik 85(8), 1272–1281 (2013).
86. J. Alstad, J. H. Augustson, and L. Farbu, J. Inorg. Nucl. Chem. 36(4), 899–903 (1974).
87. T. Sato, Hydrometallurgy 22(1–2), 121–140 (1989).
88. U.S. Pat. US7976798B2 (2009) R. Otto and A. Wojtalewicz-Kasprzak.
89. Josch, J. and Strube, J. (2012), Characterization of Feed Properties for Conceptual
Process Design Involving Complex Mixtures. Chemie Ingenieur Technik, 84: 918-931.
doi:10.1002/cite.201200030
90. K. Ghandi, GSC. 04(01), 44–53 (2014).
91. L. I. N. Tomé; and co-workers, Appl. Mater. Today 10, 30–50 (2018).
92. E. L. Smith, A. P. Abbott, and K. S. Ryder, Chem. Rev. 114(21), 11060–11082 (2014).
93. X. Li and K. H. Row, J. Sep. Sci. 39(18), 3505–3520 (2016).
94. N. Li and co-workers, Talanta 152, 23–32 (2016).
95. J. Cao and co-workers, ACS Sustain. Chem. Eng. 5(4), 3270–3278 (2017).
96. H. Zhao, S. Xia, and P. Ma, J. Chem. Technol. Biotechnol. 80(10), 1089–1096 (2005).
97. J. G. Huddleston and co-workers, Chem. Commun. 68(16), 1765–1766 (1998).
98. A. G. Fadeev and M. M. Meagher, Chem. Commun. (3), 295–296 (2001).
99. A. M. Scurto, S. N. V. K. Aki, and J. F. Brennecke, J. Am. Chem. Soc. 124(35), 10276–
10277 (2002).
100. H. Zhao and G. A. Baker, J. Chem. Technol. Biotechnol. 88(1), 3–12 (2013).
101. M. Hayyan and co-workers, Fuel Process. Technol. 91(1), 116–120 (2010).
102. Z. I. Israyandi and K. Mulia, AIP Conf. Proc. (Proceedings from the 14th International
Symposium on Therapeutic Ultrasound, Las Vegas, Nevada, USA, April 2–5, 2014),
1823, 20107 (2017).
103. V. Hessel, Chemical Micro Process Engineering, 2005.
104. C. Helling and co-workers, Chemie Ingenieur Technik 84(6), 892–904 (2012).
52 APPLICATION AND FUNDAMENTALS OF LIQUID–LIQUID EXTRACTION

105. Strube J. and co-workers, Chemie Ingenieur Technik 83(7), 1044–1065 (2011).
106. A. Hüther, A. Geißelmann, and H. Hahn, Chemie Ingenieur Technik 77(11), 1829–1837
(2005).
107. T. Wellsandt, B. Stanisch, and J. Strube, Chemie Ingenieur Technik 87(1–2), 150–158
(2015).
108. B. Stanisch, T. Wellsandt, and J. Strube, Chemie Ingenieur Technik 87(9), 1207–1214
(2015).
109. T. Wellsandt, B. Stanisch, and J. Strube, Chemie Ingenieur Technik 87(9), 1198–1206
(2015).
110. J. Eggersgluess, T. Wellsandt, and J. Strube, Chem. Eng. Technol. 37(10), 1686–1696
(2014).
111. J. K. Eggersgluess and co-workers, Chem. Eng. Technol. 37(4), 675–682 (2014).
112. J. Strube, F. Grote, and R. Ditz, In Biopharmaceutical Production Technology, Wiley-
VCH Verlag GmbH, 2012. pp. 657–705.
113. M. Kalem, F. Buchbender, and A. Pfennig, Chem. Eng. Res. Des. 89(1), 1–9 (2011).
114. H. Jildeh, M. Attarakih, and H.-J. Bart, Chemie Ingenieur Technik 84(8), 1282 (2012).
115. H. Jildeh, M. Attarakih, and H.-J. Bart, Chemie Ingenieur Technik 85(9), 1399 (2013).
116. D. Garthe, Fluiddynamics and Mass Transfer of Single Particles and Swarms of
Particles in Extraction Columns, Dissertation, Technische Universität München,
2006.

AXEL SCHMIDT
JOCHEN STRUBE
Clausthal University of Technology, Clausthal-Zellerfeld, Germany