вход по аккаунту


Old and New Processes of Multiplicative Distribution (Liquid-Liquid Extraction). A Comparison of the Methods of aterial Flow and the Mode of Action

код для вставкиСкачать
As shown in Figs. 11 and 12, substance-specific curves,
reminiscent of neutralization curves, are obtained.
The use of a gradient in the flow direction (Figs. 9, 10B,
and C) offers considerable advantages, particularly in
the separation of substances of different polarities.
Whereas on a uniform layer of silica gel a carotene
mixture from Calendula flowers could only be separated
into seven zones, sixteen substances were recognized
after chromatography on a gradient layer of kieselguhrsilica gel.
Fig. 12. Separation of a mixture of alkaloids on a gradient layer
(diagonal divider box). Adsorbents: left: alumina, acidic, for thin-layer
chromatography (Woelm)
10 % gypsum; right: alumina, basic, for
thin-layer chromatography (Woelm)
10 % gypsum. Solvent: chlorofornilmethanol (95: 5 ) , chamber saturation. Flow distance: 15 cm. Below
the spots for atropine a minor alkaloid (?) is seen.
I am grateful to J. Fuchs for his conscientious and able
assistance with the experiments, to M . Kleinert of Messrs.
Desaga for the careful production of equipment required,
and to the Fonds der Chemischen Industrie for their supPort.
Received, June 29fh, 1964
[B l769/192 IE]
German version: Chemie-lng.-Techn. 36, 94 I (1964)
Translated by Express Translation Service, London
Old and New Processes of MultiplicativeDistribution (Liquid-LiquidExtraction) [1J
A Comparison of the Methods of Material Flow and the Mode ofAction
A new process is described Jor the separation of mixtures of materials by multiplicative
distribution, which is characterized by repeated addition of the mixture throughout the
whole process and by material exchange during the countercurrent flow of finite amounts
of one, and infinitesimally small amounts of the other phase. The intrinsic features of the
mode of action of the old distribution processes and the new one are developed, and their
relationship to one another is discussed. In this connection, an attempt is made to clarqy
the confused situation in the terminology of this field by introducing clearer definitions.
Reflux and “displacement distribution”, another new distribution process, are treated
briefly in two final sections.
1. Introduction
4. the apparatus used and other technical details.
In the present paper, we shall concentrate on item 2.
Multiplicative distribution consists of the separation of
dissolved materials by distribution between two coexisting liquid phases which flow in countercurrent to
one another, and which are thereby repeatedly brought
into contact with one another for the exchange of
materials. In the ideal case equilibrium is set up. For a
complete description of such a process, we must know:
1. the substances to be separated and the solvents;
2. the manner in which the mixture to be separated and
the solvents are introduced into the process, passed from
one material-exchange to the next, and finally removed ;
the total of these operations will be called the method or
scheme of material flow;
3. the amounts or volumes of substance for each of
these individual steps; and
[ l ] For a more detailed treatment, see W. Fischer, Chemie-Ing.Techn. 36, 85 (1964).
Angew. Chem. internat. Edit. 1 Val. 3 (1964) No. I 2
In any distribution process, definite quantities and concentrations of the dissolved materials are set up in all
parts of the apparatus. This pattern of quantities or
concentrations coming into being at a given moment
of time will be called the quantity-based or concentration
characteristic, or, for the sake of brevity, the characteristic.
2. Classification of the Methods of Material Flow
We classify the various methods of material flow somewhat differently from the proposals of Hecker and
Allemann [2] and Jiibermann [ 3 ] . We use as the main
[2] E. Hecker and K . Allemann, Angew. Chem. 66, 557 (1954).
[3] 0. Jiibermann in Houben- Weyl: Methode? der organischen
Chemie. Thieme-Verlag, Stuttgart 1958, Vol. I / l , pp. 227-339.
79 I
classification characteristics the method of adding the
mixture to be separated and the method by which the
two liquid phases are brought to the material-exchange.
passed in sequence through all the mixer-settlers,
without appreciable amounts of this phase being
retained in the individual units.
c) The Various Schemes of Material Flow
a) Method of Adding the Mixture
The mixture to be separated (subsequently called the
“mixture” for short) is added either only once at the
beginning of the process (discontinuous distributions)
or continuously throughout the whole process (continuous distributions).
With a single addition, the mixture is separated in the
course of the multistage multiplicative distribution into
numerous fractions, all of which differ from one another
in their composition. Under the continuous addition regime, on the other hand, the distribution system attains
a stationary condition after a certain time. Here the
mixture can be split only into two fractions differing in
In some types of processes with repeated addition of the
mixture, e.g. in which the mixture is added batchwise after
equal intervals but one or both solvents is fed continuously,
only quasi-stationary conditions are attained. The state of
the system is subjected to periodic variations, but it repeats
exactly at corresponding times of different periods. The
periods are also known as cycles.
b) Method of Bringing the Phases to
Material Exchange
Either finite or infinitesimal amounts of the two phases
arrive at the individual steps o f the material exchange
process or the process of establishing equilibrium. In
this, no account is taken of back-mixing, i. e. of transient
retrograde motion of parts of one or both phases, or of
the formation of finite working volumes (“hold up”,
see below) in the individual distributing elements.
In this classification, there $re obviously three possibilities :
1. Both phases are fed in finite amounta and brought
into equilibrium, and are subsequently separated again.
This applies, for example, to work with separating
funnels or decanting vessels.
2. lnfinitesimal amounts of both phases are caused to
undergo material exchange and are then separated. This
procedure is approached, for example, in a column when
both phases are fed in countercurrent to one another
with good intermixing. The same situation exists, however, when one of the two phases - subdivided almost
infinitesimally - is fixed and only the other phase moves,
as is the case in partition chromatography. This shows
that our classification principle is not identical with that
for the transport of the phases.
3. At any given moment an infinitesimal amount of one
phase is brought into equilibrium with a finite amount
of the second phase, is separated, and is then brought
into contact with the next finite portion of the second
phase. This case can be realized, e.g. with mixer-settlers,
I f one phase undergoes no movement but is present in
the individual mixer-settlers in finite amounts and
remains there, while the other phase is continuously
By combining the two classification characteristics.
therefore, six diferent schemes of material flow are
obtained (Table 1). For the processes arranged in
columns 1 and 2 the names proposed by Hecker and
Allemann [2] are used.
Table I . Classification of the methods of material flow in
multiplicative distribution processes
Arriving for material exchange or establishment of
Addition of
the mixture
finite amounts
of both phases
infinitely small
amounts of both
A : single
B: repeated
finite aniounts oE
one phase and
infinitely small
amounts of the
other phase
A l
Craig process
O’KeetTe proces
A 3
Partition chroniato- Johnson process
graphy (pure [4l)
and Cornish
van Dijck process
B 3
new process
Since the criterion of column 3 was not taken into
account in the older classifications [2,3], processes A 2
and A 3 have not previously been distinguished. This is
desirable however, since these processes exhibit very
different characteristics and very different separating
effects as a result of their different methods of material
flow [l]. It is therefore also desirable to give the two
processes different names. Pure [4]“distribution chromatography” and the “Cornish process” apply to class
A2, while for A 3 we propose the name of the “Johnson
process”, since, out of a number of authors who
developed this process independently of one another
(besides Johnson they are Kies and Davis [5], Weygand
[6], Fischer and Jiibermann [7]), Johnson was the first [8]
to give it an exhaustive description and mathematical
treatment [9].
Process B 3 in Table 1 differs from the Johnson process
only in that the addition of the mixture is repeated. We
have developed and described this previously unknown
process [lo]. (Exhaustive treatment in the next Section
and by F. Horj? [l I].)
-~ -
[4] By this is understood pure partition chromatography wlthout
complications due to adsorption etc.
[5] M . W . Kies and P . L . Davis, J. biol. Chemistry 289, 637 (1951).
[6] F. Weygand, A . Wacker, and H. Dellweg, Z . Naturforsch. 6h,
130 (1951).
[7] W . Fischer and 0. Jiibermann, Chemie-1ng.-Techn.23, 299
[S] J . D. A . Johnson, J. chem. SOC. (London) 1950, 1743.
[9] A short description of the principle of this process and the
derivation of the fundamental equation will be found in the
general review of distribution processes by S . Stene, Ark. Kemi,
Mineralog., Geol., Ser. A 18, no. 18, pp. 39 et seq. (1944).
[lo] W. Fischer, German Patent 1050731 (March 6th, 1956).
[ I I ] F. Horn, Chemie-1ng.-Techn.36, 99 (1964).
Arigew. Chem. internut. Edit.
1 Vol. 3 (1964) / No. 12
Other known distribution processes can be regarded as
variants of the six fundamental types of Table 1, in which
either the conditions are specially chosen (e.g. various
methods of withdrawing the substances in the Craig process,
addition of mixture at one end instead of to a middle stage
of the cascade in continuous processes) or the apparatus
brings about certain deviations from the selected and to
some extent idealized classification criteria (e.g. backmixing, only approximately infinitesimal subdividing of a
3. The New Process
roles of the previous nos. 1 to 4, and the newly added
one the function of no. 5. Five stages are therefore in
operation at any given time.
It is desirable to arrange the mixer-settlers in a circle or in
two parallel lines close to one another. When more units
are available than the number which must be operated
simultaneously to obtain the required separating effect, and
when all the units are provided with suitably arranged inlet
and outlet conduits, the transition from one cycle to the
next can be performed simply by operating suitable valves
The process can be carried out in a similar way if the
phases are treated in the reverse manner, with the light phase
in finite portions and the heavy phase in infinitesimal portions.
It is impossible to achieve in practice the situation in which
as presupposed above - only an infinitesimal amount of the
continuously flowing phase is present in each mixer-settler.
In actual Ftct, a finite “working volume” (“hold up”) of this
phase is necessary in the mixing and settling chambers.
However, this affects the separating effect only slightly when
the amount of material dissolved in this volume is small in
comparison with the total amount of dissolved material
present in the mixer-settler; cf. Horn [I 11. A similar situation
prevails in the Johnson process [12]. The following theoretical
considerations are idealized throughout by the assumption
of a vanishingly small working volume.
After a sufficiently large number of individual cycles
under constant conditions, a quasi-stationary state is
finally obtained. The concentration of the distributed
materials for each stage can be calculated for this case
[I I], assuming constant distribution coefficients and
the establishment of equilibrium at all stages; we also
assume here that the mixture to be separated is introduced in the solvent-free state. These three assumptions
always apply, in what follows, for the other processes as
well, unless the contrary is expressly stated,
a) Procedure
The new process is periodically quasi-stationary. The
operations during one period are shown schematically
in Figure 1 . The process is preferably carried out in
mixer-settler devices; in the example of Figure 1 , five
distribution elements are used.
Fig. I . Scheme of t h e new process.
1 t o 5 mixer-settlers; a = introduction of a n a m o u n t of mixture Z :
b = introduction of t h e light phase vl, i n very small a m o u n t s A y ;
c = removal of XAv1 = v1 with t h e a m o u n t of dissolved substance m5,1;
d = removal of a n element a t t h e e n d of t h e cycle with a n a m o u n t of
dissolved substance m l , h in a volume Vh; e = a d d i t i o n of a n element
a t the e n d of a cycle with volume Vh.
At the beginning of the cycle, each mixer-settler contains
a finite volume vh of the heavy phase, which remains in
it as long as it takes part in the exchange; the whole
cascade is free from the light phase. One of the distributing elements - the third in the example of Figure 1
- is supplied with an amount Z of the mixture to be
separated. This can be added once at the beginning of
the cycle, or gradually during the cycle. The light phase
is added to element no. 1 in small (in the ideal case,
infinitesimal) portions. Each of these portions is brought
into equilibrium with the volume vh, and is then immediately separated and transferred to the mixing chamber of the next unit, treated in a corresponding manner,
and finally removed from element no. 5. When in this
manner a certain volume v, of the light phase has
been added and withdrawn continuously, the addition
and removal of the finite volume vh of the heavy phase
as a whole is carried out all at once. This is preferably
performed by removing the mixer-settler no. 1 from the
process, together with the heavy phase vh which it
contains and the material dissolved in it, while a new
mixer-settler containing a volume vh of pure heavy
solvent is added at the other end of the distribution
cascade. This completes the cycle.
The next cycle takes place in an analogous manner. In
this, the previous mixer-settlers nos. 2 to 5 take over the
Angew. Chem. internat. Edit. 1 Vol. 3 (1964)
1 No. 12
b) Comparison with the O’Keeffe and
van Dijck Processes
The result calculated in this way for the fractions
removed from the two ends of a seven-stage distribution
by the new process, with the addition of the mixture to
be separated at the central (fourth) stage, are compared
in Figure 2 with the situations in the O’Keeffe and van
Dijck processes (which agree in their separating effects,
cf. p. 798). The amounts of these end fractions are
plotted logarithmically against the logarithm of the
extraction factor
. “I
K being the partition coefficient, c the concentrations.
While the curves for B 1 and B 2 intersect at G = 1 and
form mirror images of one another, this is not the case
with the curves for the new process, which express the
asymmetry of the flow of the two phases in this process.
Moreover, the curves of the new process are steeper in
their important sections. This means a better separating
effect (cf. the following paragraph). This difference is
particularly great at high values of G.
Figure 2 is also useful as a nomogram in the selection of the
operating conditions for a given separating problem. If t w o
substances a an d b are to be separated, and the solvents have
[I21 W. Fischer et al.,
Angew. Chern. 66, 319
but with variations of the absolute magnitude of G ;
associated values for Ga: Gb = 2 are always adjacent
to one another, while the absdute values of G decrease
in the five double columns from left to right. In addition,
Table 2 gives the so-called migration rafio or rejection
ratio [14].
= m7,1 : m 1 , h .
In the bottom line of Table 2 is given the overall separating e#ec?ct of the seven-stage cassade Ta,b = Ra: Rb.
A comparison of the figures of Table 2 shows the following facts (discussed exhaustively in [l]):
1. In processes B I and B 2, when the mixture is added
at the central stage, a material with G = 1 (column 5a)
is distributed equally in the two phases leaving the ends
of the system: m7,~= m1,h = 0.500. The migration ratio
R = 1.OOO. On the other hand, the new process operates
unsymmetrically: a material with G = 1 migrates preferentially with the light phase: m7,1 > ml,t, and R > 1.
2. If in processes B 1 and B 2 the values of G are chosen
in such a way that Ga = 1 : G b (reciprocity rule), i. e. for
our example Ga = k ' i = 1.414 and Gb = 1: 1/2 =
0.707, then m7,1,a = ml,h,b and m7,1,b = mi,h,a. It can
be seen from Table 2 (column 3, bottom) that material
a is obtained in 80 % purity with 80 % yield with the
light phase from stage 7, while material b is similarly
obtained in 80 % purity with an 80 % yield with the
heavy phase from stage 1. In order to arrive at the
corresponding situation in the new process, we must
make Ga = 1.334 and Gb = 0.667, i.e. the reciprocity
rule does not apply to this case. However, as Table 2
(column 3, top) shows, both purity and yield are substantially higher, amounting to 92.3 "/, for a and b (cf.
also R and T).
Fig. 2. Amounts of the fractions m1,h and m7,l in fractions of the
amount Z = I of material fed in as a function of the extraction factor G.
Seven-stage distribution, addition of the material to he distributed to
stage no. 4.
_ O'Keeffe
and van Dijck
processes (B 1 and B 2)
new process (B 3). light phase infinitesimally subdivided.
been chosen, the partition coefficients Ka and K b and the
G a : G b ratio are fixed. The absolute value of G can still be
varied by the selection of VI:Vh. If a rectangular aperture
with a horizontal width of (log Ga-log Gb) is cut out of
paper, its horizontal movement on Figure 2 corresponds to a
change in Vl:Vh- It is then possible to read off from the
ordinate the proportions of the two materials in the end
fractions corresponding to any value of v1: Vh and thereby
to determine the yield and purity of the separated products.
Table 2. Calculated figures for seven-stage distribution of two materials a and b under stationary conditions. Addition of the mixture in the
amounts Za = Zb = 1 to stage no. 4. The following relations hold for any two adjacent columns: 3 = Ka: Kb = Ga: Gb = 2.
1. Distribution according to the new process B 3; light phase infinitesimally subdivided
The advantage of the new process is shown still more
clearly by Table 2. In this Table are given some selected
figures from Figure 2 for m7,l and ml,h for two substances with the constant separating factor
[13] This follows, for example, from calculations given by
JidJermann PI PP. 250-253, and by Treybal[14],formula 7.70
f o r € ' = & a n d n' = n + 1.
I141 R . E. Treybal: Liquid Extraction. 2nd Edit., McGraw Hill,
Angew. Chem. internat. Edit.
Vol. 3 (1964)
1 No. I2
3. For separations by methods B 1 or B 2, the following
relation holds [13] when the mixture is fed to the central
stage (0 = number of stages):
apparatus is described by a horizontal row of the flowsheet, e.g. by the row E-F, in which three distribution
elements are in operation.
and the overall separating effect is therefore independent
of the absolute magnitudes of G (Table 2, 11). On the
other hand, in the new process the overall separating
effect Ta,b is much greater and depends not only on
the ratio G,:Gb but also to a large extent on the
absolute magnitudes of G (Table 2 , I). This shows that
the new process is jiindarnentally diferent Jrorn those
used hitherto.
4. When the conditions are selected in accordance with
the double column 3 of Table 2, the new process gives a
separating effect Ta,b = 144, i.e. a figure which is
somewhat greater than (Ga/Gb)7. If it is desired to
obtain the same separating effect by B 1 or B 2 , thirteen
or fourteen stages would have to be used, i.e. almost
twice as many as in the new process.
Tiiese differences, shown here on the basis of a few
examples, are of general nature, as Horn has pointed
out [I I]. This applies particularly to the finding that the
new process requires only half as many stages as the
older processes to obtain a given overall separating effect.
A further advantage of the new process is that it
requires smaller volumes of solvents than the older
processes to separate a given amcunt of a mixture
with the same separating effect; cf. [l].
These theoretical results have been confirmed by a
series of experiments [l, 101. In these, the same separation was carried out both by the O'Keeffe process and
by the new process in each case. An example is described
here :
A 1 : 1 mixture of succinic and oxafic acids was distributed
between water and n-britanol in seven stages with the
addition of the mixture to stage 4. The temperature was
1 "C and p = K s : K o = 1.03:0.57 = 1.81. After
stationary conditions had been achieved, figures for the
overall separating effect T in the O'Keeffe distribution
process varied between 8.6 and 9.4. They correspond to
p 3 6 to 3 3 8, while theoretically 9 4 is expected. In the new
process, the overall separating effect was between 30.6 =
(55 8 and 34.8 = P6.0, while according to calculation 76.')
would have been expected for idealized conditions. Thus,
for both processes, the separating effect obtained was
only slightly below the theoretical figures. However,
above all, the experiments confirmed the expected
improvement in the separation by the use of the new
4. Comparison of the Different Schemes of Material
Flow, and Their Modes of Action
a) Flow-Sheets
Prorcss A I . In the Craig process, which can be
presented by a triangular scheme (Figure 3, top),
mixture to be separated is added in one portion at
apex D. At a given moment of time, the state of
Atrgew. Cliem. interiztrt. Edit. / Vol. 3 (1964)
1 NG. I 2
Fig. 3. Interrelationships of the various schemes of material flow. Each
circle denotes one distribution operation. (For explanation, see text).
Broken arrows:
light phase.
Full arrows:
heavy phase.
Double line arrows: introduction of the mixture
Process A 3. The Johnson process is represented in the
triangular flow-sheet at the top of Figure 3 by a line
D-F or D-E, according to whether the heavy or the
light phase is used in a state of finite subdivision. Here,
as an example, we shall deal only with the first case. All
the distribution elements from D to F are provided from
the start with equal finite amounts of the heavy phase
which remain in them during the whole process. The
mixture is added to element D only once at the beginning of the process. The light phase then flows i n
infinitesimally small amounts successively through the
individual distribution elements of the row D-F, until
the desired degree of separation of the components has
been achieved.
Process B 1. The O'Keeffe process is described by the
central part of Figure 3. It is derived from the Craig
process if fresh amounts of mixture are added regularly
and the light and heavy phases are taken off from the
ends, so that the number of stages in operation (see p.
796) remains constant. The horizontal rows M -N and
P-Q together form one cycle, which is always repcated
in the same manner.
O'Keeffe [ I S ] worked with a doubled scheme, in such a b a y
t h a t two similar processes were carried out together with a
displacement lasting half a cycle. This is only the mechanical
[15] A . E. O'Keeffe, M . A . D d l i i w . and E. T. Sliller. J . Arner.
chem. SOC.71, 2452 (1949).
addition of two individual processes completely separate
from one another as regards materials, and consideration of
the simple flow-sheet of the center of Figure 3 is therefore
Process B 2. In the van Dijck process, the light and
heavy phases flow continuously and simultaneously in
opposite directions through all distribution elements of
the series, while the mixture is likewise fed in continuously at a suitable point. The process can therefore be
shown, for example, by the flow-sheet Y - Z at the
bottom of Figure 3 .
Process B 3. The new process is shown in Figure 3 by
the oblique rows M-N“, MI-N”’, etc., or by N-M“,
N’-M”’, etc. according to whether the heavy or the
light phase is used in the form of finite amounts, in a
similar manner to the Johnson process (A 3). In the case
where the heavy phase is divided into finite portions,
the row M-N” forms a cycle. When, after the addition
of mixture, the necessary amount of light phase has been
passed through the process in infinitesimal amounts,
M is removed, P-N” take the position of MI-Q“, and
N”’ is added, as described above. This completes the
b) Definitions
Before the separating effects of the individual methods
of material flow can be compared, some questions of
nomenclature must be discussed, since considerable
confusion exists in this respect. Here we wish to propose
some definitions, particularly in view of the current
efforts to arrive at an internationally uniform terminology in the field of distribution processes.
n = Number oj’ distribution steps. A distribution step
means an establishment of equilibrium with all the
distribution elements simultaneously in operation. The
term can only be used with respect to processes with
two finitely subdivided phases, i. e. processes A 1 and B 1.
elements. Consequently the concept of stages should not
be used in this case. In the O’Keeffe process, for the
example given in the center of Figure 3, cpr= 5 , while
throughout only three (row M-N) or two (row p-Q)
elements, alternately, are simultaneously filled and in
0 t h = Number of theoretical stages = number of practical
stages which, with complete establishment of equilibrium
in each individual distribution operation, give the same
overall separation effect. as a process without individual
distribution elements and/or without establishment of
We shall speak of the separating power of a process
when we mean the separation achieved only in a general
sense. As measurable magnitudes we propose the
1. The separation factor
p = K a .. K~
G . G~ =
: ‘l,b
%,a: c h , b
As a material constant, it is independent cf the type of
multiplicative scheme of material flow used.
All measurable magnitudes for the change in the ratio
of the amounts of two components of a mixture within
a multiplicative distribution cascade that depend on the
method of material flow, shall be called separating
e#ects. We distinguish:
2. The overall separating effect T of a distribution cascade. It has already been defined on p. 794 as the
quotient of the migration ratios R of two substances. If,
in accordance with p. 794, the corresponding ratio of
amounts is put for R and the last distribution element
of the cascade is denoted by w, we obtain the following
expression for T, which can also be formulated with the
Ta,b = R a : Rb =
my& : 3 . h . a
mw,l,b: ml,h,b
m d a: mw,l,h
m1,h.a: m l , h , b
cw,l,a : cw,l,b
cl,h,a : Cl,h,b
= Number of distribution cycles; cf. p. 792. This term
is meaningful only with continuous processes with at
least one finitely subdivided phase, i.e. processes B 1
and B 3.
z Number of distribution elements (also called elements,
for short) of an apparatus. This number can obviously
be given only €or apparatus which consist of individual
distribution elements, i. e. not for packed columns and
the like.
opr = Number of practical stages. Again, we can speak
of practical stages only in the case of an apparatus consisting of individual distribution elements. We define
opras the number of distribution elements traversed
by the mobile phase from its introduction to its removal.
If both phases are transported, the number of distribution elements traversed by each phase is the same. In prccesses A 2, A 3, B 2, and B 3, so far as apparatus
consisting of individual distribution elements are used,
upr = z. It must be particularly noted that this equality
does not apply to the O’Keeffe and the Craig processes.
In the latter case (i. e. in the Craig process), the number
of practical stages in the above sense is generally not
clear-cut, while there is no doubt about the number z of
Cl,b : Ch,b
Consequently, we may also regard Ta,b as the ratio of
the amounts or concentrations of the two materials a
and b in the light phase of the last distribution elexent
(= purity ratio of this fraction) divided by the corresponding purity ratio in the heavy phase of the first
3. We denote by the term partial separating effect U the
quotient formed in the same way from the concentration
ratio in any two distinct distribution elements i and j ,
now, however, within the same phase of the two elements :
ciAa: ci,h,b~
cj,h,a: Cj,h,b
The quotient relating to the light phase is identical with
that for the heavy phase under equilibrium cmditions
with a constant separation factor.
4. An important special case of the above, the separuting
effect S from stage to stage, is based on the definition of
a stage given above, i and i-1 being neighboring stages:
ci,h,a : Ci,h,b
Ci l,h,a:Ci l,h,b
Angew. Chem. internat. Edit. / Vol. 3 ( I 964)
Nc . I2
The separating effects defined in this way must be
supplemented for a complete characterization of the
separating power of a process or a plant by data on the
yield (i.e. the fraction of the amount of a component
added that is removed from the plant in enriched form),
and on the separating output, by which is understood the
amount of mixture passed through and subjected to a
given separating effect, referred to the time or to the volumes of the two solvents. These two differently based
forms of the separating output are connected with one
another and with the rate of throughput of the solvents.
The reciprocal 01 the separating output referred to
volumes is also called the solvent demand.
Comparison of the Processes
Process A 1. Figure 4a gives the quantity-based characteristics of three distributed materials with different
extraction factors G at a certain moment of the
process for a Craig distribution (fundamental process)
with a large number of distribution elements. It is
noteworthy that the curves for the two materials with
the reciprocal G values 0.1 and 10 are symmetrical to
one another, and are precisely the same with respect to
steepness and height [16].
It follows from the well-known calculations [17] of this
method of working that between neighboring distribution elements in simultaneous operation (in Figure 3,
top, these are the neighboring elements of a horizontal
row, e . g . the row E-F), the separating effect, Ua,b =
Pa,b. If G a = 1:Gb, it follows similarly for the distribution operations succeeding one another in time of the
oblique row D-F of Figure 3 that Ua,b =
Process A 2. A description of this scheme is not necessary in
order to understand the considerations predominating here.
A comparative discussion of its mode of action is to be found
in the German text [I].
Process A 3 . The characteristics of the Johnson process
differ in a quite definite manner from those of the Craig
process, as shown by Figure 4b on the basis of the
separation of a ternary mixture at a given moment of
time. With increasing -magnitude of the extraction
factor G, the curves become flatter and broader, and
reciprocal G figures do not yield curves of the same
steepness and height, as in the Craig process. For the
separating effect from stage to stage, e.g. in row D-F
of Figure 3, we have Sa,b = Pa,b [12].
Comparison of A 1 and A 3. We again consider a Johnson process in which the heavy phase is present in a
state of finite subdivision in stages D-F of Figure 3
and the light phase flows uniformly without the formation of a finite working volume. With this, we compare a
fundamental process of the Craig method, in which
decantation vessels may be used. In this case again each
portion of the heavy phase is associated with a definite
distribution element and the light solvent forms the
mobile phase, which however is transferred in batches.
1161 See, for example, 0. Jiibermann [3], p. 261.
[ 171 See, for example, E. Hecker: Verteilungsverfahren im La-
boratorium. Verlag Chemie, Weinheim 1955.
Airgew. Clrem. internat. Edit. Vol. 3 (1964) No. 12
,.o LL&
Fig. 4. Example showing the characteristics
a) of the Craig process (A 1); b) of the Johnson process (A 3).
Abscissa: Number of the distribution elements.
Ordinate: Content of the distribution elements, m.
We can now alter to some extent the temporal sequence
of the individual operations in the Craig process without
changing the principle, by first passing only one portion
of the light phase through all three distribution elements
from D to F. All three elements are then again free of
the light phase, and the state is comparable with that of
the Johnson process after the same amount of light
phase has been passed through. As is well known [18],
the extraction effect, and therefore the separation effect,
are increased when a given volume of the extracting
solvent is used in the form of separate small portions.
It therefore becomes understandable that in process A 1
the separating effect between two neighboring elements
of row D-F is only I/) while in process A 3 it is p. In
the normal Craig process, two further portions of light
phase have been brought into operation in the meantime, so that the three elements are then in the state
E-F. The separating effect between neighboring elements has then increased to P in the Craig process as
well, which, as compared with row D-F, arises from
the fact that the heavy phase which has been placed in D
has now been treated with three times the volume
corresponding to the Johnson process, the middle
element with twice the volume, and only the element F
with the same volume. The Craig process therefore
achieves the advantage of the Johnson process by
pursuing the separating operation partially, i. e. in the
flow-sheet of Figure 3 towards the left-hand end in
increasing degree. Mathematical treatment of the cases
shows that the disadvantage for the separating effect
between neighboring elements resulting from establishment of equilibrium with finite portions of the light
phase is just compensated by the partial pursuit of the
process. This result is not obvious.
Thus, using an equal number of distribution elements,
processes A 1 and A 3 give the same separating effect
but possess different characteristics. Furthermore,
[IS] See, for example, 0.Jiibermann [31, pp. 234-235.
process A 3 has a smaller solvent demand for the same
output [19].
Processes B 1 and B 2. The van Dijck process B 2 can
be carried out in columns or in mixer-settlers. We shall
first consider the second case In operation, mixers and
settlers contain certain working volumes (“hold up”)
of the two phases, their amounts and concentrations in
each distribution element being constant in time under
stationary conditions. The magnitudes of the working
volumes are obviously irrelevant to theoretical considerations with idealized conditions, and are to be
chosen only according to practical points of view (e .g.
the velocity of mass transfer and the rate of settling). For
these reasons, in the idealized calculation of the concentration characteristics it is possible to leave the
working volumes out of consideration. The van Dijck
process is then considered as if infiniresimal amounts
of the two phases were brought directly to mass exchange with one another - in the ideal case, brought
into equilibrium - and then separated, and passed to
the subsequent stages. On the one hand this justifies the
classification of the van Dijck process, when carried out
in mixer-settlers, in our category B 2. On the other
hand, these considerations show that the idealized van
Dijck process can be regarded as an O’Keeffe process
with the feed and transport of infinitesimal amounts
of all the participating materials, from which it follows
that the characteristics, the separating effects, the yields,
and generally the entire separating power of the two
processes are identical, provided all other experimental
conditions are equal.
If, on the other hand, the van Dijck process is carried
out in a columtz (without back-mixing), the scheme of
material flow corresponds completely to the conditions
of our category B 2. However, in general, the time of
contact of two infinitesimal portions of the two phases
is so short that their concentrations will remain further
from equilibrium than in mixer-settlers.
The calculation is most simply carried out by the use
of a fictitious “height equivalent to a theoretical stage”,
by analogy with rectification. Thus the calculation of
the van Dijck process in a column and its conception
within the framework of om considerations leads back
to the method of working with mixer-settlers sketched
above, and the same conclusions follow
When Ga is chosen as equal to 1/Gb in the van Dijck
process [20], the relation Sa,b = l/pa;b holds [21,22]
for the separating effect from (theoretical) stage to stage
in row Y - Z of Figure 3 The same applies in the
O’Keeffe process for neighboring stages, i. e. in Figure 3
for the row M-P-0-Q-N.
Process B 3. The new process has already been compared with processes B 1 and B 2 above. Figure 5 shows
in addition a comparison of the quantity-based charac[I91 0 . Jiibermann [3], pp. 276 et seq.
[20] In the German text [I], on p. 95 and in Table 5, this prerequisite was erroneously omitted.
[21] According to 0. Jiibermann [3], pp. 243-244, and 250, or
according to R . E. Treybai [14] using equation 7.84.
[22] With addition o f the mixture to the middle stage, the following expression holds for the average separating effect from stage
, even if Ga IjGt,.
to stage: S = T1Xo + I )
teristics for a seven-stage distribution with the addition
of the mixture to the middle stage for two materials
with G a = 2 and G b = 1/Ga = 0.5, i.e. p = 4. The
symmetrical character of the O’Keeffe procedure i s
reflected in the fact that the curve for the content of
material a in the light phase in the individual stages
forms a mirror image of that for material b in the heavy
phase. The unsymmetrical flow of the phases in the new
process leads to curves which deviate markedly from a
mirror-image relationship; the material with G b = 0.5
is somewhat held up in the first stages, while in particular
the material with G a = 2 is greatly displaced towards
higher-numbered stages as compared with the O’Keeffe
While in the O’Keeffe distribution in the case under
consideration the separating effect between neighboring
stages Sa,b = v p = 2 and is constant, with the new
process it is on average somewhat more than 4, i.e.
somewhat larger than F, and varies from one pair of
stages to another in a characteristic manner [l].
r E E F 5I
Fig. 5. Examples of the quantity-based characteristics at the end of a
cycle in seven-stage distribution.
_ _ process
(B 1).
New process ( B 3), light phase infinitesimally subdivided.
Addition of the material to be distributed in an amount of Z = 1 per
cycle to stage 4.
Abscissa: stage no. i.
Ordinate: amount of distributed material in the heavy phase mi,h and
in the light phase mi.1 of stage i.
Summarizing, it can be stated that the new process, at
the same yield and with a smaller consumption of solvent, requires only about half as many stages to obtain
a given overall separating effect as the OKeeffe and van
Dijck processes. Its characteristics distinguish it markedly from the other two processes. The separating
effect from stage to stage varies with the position of the
stages within the cascade, and the overall separating
effect rises with the absolute values of K at a constant
Angew. Chem. internat. Edit.
Vol. 3 (1964) / No. 12
value of pa,b = Ka :Kb. These phenomena also differentiate the new process from the Johnson process with
which it otherwise has much in common, particularly
in the method of the flow of phases.
5 . Reflux
If in processes B 1 or B 2 complete reflux of the dissolved
material at both ends of the cascade is used, the relation
Sa,b = Fa,b holds [23,24] for the separating effect from
stage to stage under stationary conditions; it is therefore
greater by a power of two than in the absence of reflux.
A corresponding situation exists [l] when the new
process B 3 is carried out with complete reflux. The
separating effect from stage to stage is thereby again
squared and, on average, becomes equal to 92. This
superior separating effect, which was previously shown
in the Johnson process and in the new process without
reflux, is obviously characteristic for the third method
of phase flow (Table 1).
6. Displacement Distribution 1251
When certain extreme deviations from the Nernst
distribution law exist (cf. Figure 6), peculiar phenomena
may arise in multiplicative distribution.
p F c -
Figure 6. Distribution isotherms.
Curve 1 : Nernst distribution law.
Curves 2 and 3: Extreme cases which can lead t o displacement
c = Concentration, w = aqueous phase, org = organic solvent phase.
In the first place, these cases often offer a convenient
possibility of working with reflux at both ends of the
distribution cascade. An example of this is given by
extractions of conlplexes by the scheme :
+ HXorg = M X o r g -t-
when the chloride ion is present only in the aqueous
solution and the complex-forming agent X only in the
organic solution. The marked dependence of these
equilibria on the hydrogen ion concentration of the
aqueous phase permits reflux to be carried out: if the
pH is kept high at that end of the cascade at which the
[23] If in [3] IA, = A and ,A R = a are p u t in Figure lob, p. 242,
the above result follows from the equations on pp. 243--244.
Similarly from 1141, equation 7.62.
[24] This relation also holds when Ga =
lb .
[ 2 5 ] Cf. U. Neirrel, Ph. D. Thesis, Technische Hochschule
Hannover. 1961.
Angew. Chem. internat. Edit. [ Vol. 3 (1964)
1 No. 12
organic solvent and the complex-forming agent are
added, the metal M can be transferred here completely
from the aqueous to the organic phase; if aqueous acid
is fed at the other end of the cascade, the reverse process
can be brought about there.
A further feature appears under these conditions. The
extraction from the aqueous phase follows the peculiar
Curve 2 in Figure 6 and leads to a saturation of the
organic solvent with the metal M, the height of which
is determined by the predetermined concentration of X
but is independent of the metal concentration in the
aqueous phase.
If, with a multistage distribution cascade, it is arranged
that the aqueous phase of all stages with the exception
of the first and the last contains the same concentration
of MCl, the organic phase flows past the aqueous
phase without a change occurring in the concentration
of the metal in the two phases. If a mixture of several
metals is present, the same applies for their total (equivalent) concentration; but in the middle stages there is
an interphase exchange of the individual metals with
one another, corresponding to the different stabilities
of their X-complexes. Thus, there is a rectangular concentration characteristic for the sum of the metals
within which the various metals separate from one
another in the course of the process as individual bands.
The picture is thus substantially analogous to displacement chromatography, and the reason is the same:
extreme positions of the equilibria at the front of the
mixture and at the front of the eluent, combined with
saturation phenomena. Because of this analogy, we
propose for multiplicative distribution processes with
rectangular concentration characteristics the name
displacement distribution.
When the extreme shape of the equilibrium isotherms is
given, this type of distribution can be carried out with all
three different methods of phase flow (p. 792). It differs from
a distribution in a system for which the Nernst distribution
law holds by the fact that in this latter case - even with
complete reflux - there are n o rectangular concentration
characteristics, the concentration varying from stage to
stage. In case the Nernst distribution law applies, discontinuous multiplicative distribution corresponds to elution
The 'occurrence of displacement distribution is not
restricted to the complex equilibria mentioned [l]. We
have also observed cases in which distribution isotherms
of the shape of Curve 3 in Figure 6 lead to displacement
distribution [I].
We have applied displacement distribution particularly
to the scheme of the new process B 3, and have obtained
very good separating effects under these conditions particularly with partial reflux. A theoretical treatment
of the problem on a digital computer with simplified
assumptions made by M. Kriiger on the basis of preliminary work by U . Neitzel showed that the separating
effects from stage to stage vary in a complex manner
with the position of the stages within the cascade, the
magnitude of the reflux, and the ratio of the amounts of
the components to be separated, but are sometimes even
greater than for the new process B 3 when the distribution law applies. This and the associated experiments
will be reported in detail elsewhere.
The present paper has been produced on the basis of the
experiments and results of many collaborators. The
following may be mentioned in particular: K.J.Bramekamp, J. Busemann, P. Eger, G . Fricke, H. Galster,
K . H. Grothe, M. Kruger and D. Schutte. - We thank
the Fonds der Chemischen Industrie, the Bundesmini-
sterium fur Wissenschaftliche Forschung, and the Deutsche Forschungsgerneinschaft for financial support of our
Received, September 5th, 1963
[B 1662/197 IE]
German version Chemie-lnp - T e c h 36, 85 (1964)
Translated by Express Translation Servlce, L o n d o n
Detection of a Heterocyclic
Nucleophilic Carbene I*]
By Dr. H. Quast and Prof. Dr. S. Hiinig
Chemisches Institut der Universitat Wiirzburg (Germany)
After it has not been proven so far that tetra-aminoethylenes
dissociate into nucleophilic carbenes [l-31, numerous reactions which have been ascribed to nucleophilic carbenes [4]
probably must be given a new interpretation. Since a yellow
crystalline substance recently described [5,6] is not ( I ) but
its “dimer” (2) [7], there are no reactions known - apart
from the rapid exchange of the protons on the a-carbon
atom of quaternary heterocycles [8] and the formation of
dimers of type ( 2 ) [3-6,9] - which definitely proceed via a
carbene and not via its dimer.
diisopropylethylamine [lo], and in differentiating it by means
of its reactions from its dimer (2) which is also formed under
comparable conditions [5,6]. Compound (2) is obtained in
70-75 % yield from (3) with sodium hydride in dioxan under
nitrogen and reacts with 2 moles of the azidinium salt
(4a,b) I l l ] or with 14 moles ofp-toluenesulfonylazide (4e) in
acetonitrile at 5 “ C to give 1 mole of nitrogen and m i x t u r e s
of compounds (5) and (6), e . g . 21 % (5u) + 41 % (6a), 14 X,
(5b) 50 % (bb), or 45 % (Se) 79 % (6e). The less reactive
compound (4d) gives only 1.3 moles of (6d) and 1.5 moles
of nitrogen.
In contrast, u n i f o r m p r o d u c t s are isolated when ( I ) is
prepared from (3) in the presence of the azides (4) : compounds f4a), (46), and (4c) undergo a very rapid reaction
to form almost exclusively triazacyanines [ l l ] , vir. 75-80 7;
(5a), 69 % (5b), and 32 % ( 5 c ) , whereas (4d), (4e), and (4f)
yield 70-80 % nitrogen and only the corresponding compounds containing only one N-atom in the chain [12], viz.
53 % (6d), 62 % (6e), and 44 % (6f).
The products of the reactions with the carbene in situ show
that the course of the reaction is decisively influenced by
the electrophilicity of the azide. It is only with the strongly
electrophilic azidinium salts (4a)-(4c) that the products
expected from a reaction which traps the carbene ( I ) are
actually obtained. The less reactive azides (4d), (4e), and
(4f) are apparently unable to capture ( I ) in the presence of
(3). Since the dimer (2) is readily formed under the reaction
conditions used [.5,6], it must be assumed that the reaction
with (4e) and (4d) proceeds via (2).
The reaction of (2) with azides is understandable if it is
assumed that an adduct (7’), or more probably (7), is formed,
and then decomposes by loss of nitrogen to form (6). Just
as in the decomposition of the adducts of electrophiles with
tetra-aminoethylene [3], a carbene must be formed here
too. The fate of this carbene is again determined by the
most reactive electrophile present. The yield of nitrogen, the
relative proportions of the products formed, and the observation that no iminodiazosulfone (Se) is formed from
(2) and (4e) when (2) is produced directly in the solution
from ( 3 ) prove that (7) cannot be transformed directly into
(5) here.
In the reaction of (2) with 2 moles of a benzenediazonium
fluoroborate @-NOz, p-CH3, p-N(CH3)2), only 1 mole of
nitrogen and the compound with two electrons less, i. e. (8)
1131, could be isolated (43 %, 26 %, 22 % yields). Although
(d) R = C H I - % D
(e) R = CHI#SOi-
(fl R
0 8 -
We have now succeeded in producing a short-lived nucleophilic carbene ( I ) by deprotonation of 3-methylbenzthiazolium salts (3), e.g. in acetonitrile with triethylamine or
i 91
( 9 ) [141 is expected by reaction with ( 1 ) [6], only traces are
formed. The red-brown color of the solutions is partially
Angew. Chem. internat. Edit. / Vol. 3 (1964)
I No. 12
Без категории
Размер файла
1 045 Кб
flow, distributions, mode, aterial, old, method, multiplicative, liquid, comparison, new, extraction, action, processes
Пожаловаться на содержимое документа