Talanta 51 (2000) 281 – 289

www.elsevier.com/locate/talanta

Donnan dialysis of copper, gold and silver cyanides with

various anion exchange membranes
Djamal-Eddine Akretche *, Hacène Kerdjoudj
Laboratory of Ore Valorisation, Chemistry Institute, Uni6ersity of Science and Technology, ‘Houari Boumediene’, BP 32, El-Alia,
16111 Bab-Ezzouar, Algiers, Algeria
Received 30 December 1998; received in revised form 10 August 1999; accepted 25 August 1999

Abstract

Donnan dialysis is an ion exchange membrane process that can be used for the purification and concentration of
diluted solutions. In this work, the behaviour of gold, silver and copper in cyanide medium is examined. Flux of
cyanide complexes and corresponding free cyanide are determined using five commercial anion exchanger membranes
(AMV, ACS, RAI 5035, ADP and ADS). The results show that the rate transfer depends upon the nature of the
anion exchanger membrane. It is observed that the species number in the feed solution influences the transfer
selectivity of metal ion complex against free cyanide Thus, gold which forms only one stable species with cyanides is
transferred faster through an ACS membrane than copper which forms three species. However, this result is not
verified when an ADS membrane is used. A model of the complex transfer through anion exchange membranes based
on Donnan dialysis is proposed. A three compartment Donnan dialysis is performed to improve the separation
between the studied metals. Decyanidation is also examined and separation factors are calculated. It is shown that
Donnan dialysis can be an efficient technique for the separation of cyanides complexes of copper, gold and silver
when parameters such as anion exchange membrane and the number of compartments are optimised. An advantage
of this technique is also the possibility of recycling all reactants with a good impact on the environment. © 2000
Elsevier Science B.V. All rights reserved.

Keywords: Donnan dialysis; Cyanides; Metals; Separation

1. Introduction this field. Donnan dialysis is an ion exchange

membrane process that can be used when diluted
Interest in the development of efficient tech- solutions are treated. It is based on a chemical
niques for separations and recovery of ions has potential difference between two compartments
been a stimulant for testing various methods in separated by an ion exchange membrane polymer.
Moreover, Donnan dialysis is both an economi-
 cal, simple technological and energy saving pro-
Presented at the Sixth Symposium on Analytical Sciences,
Valencia (Spain), June 22–24, 1998. cess. However, Donnan dialysis is not applied in
* Corresponding author. Tel./fax: 213-2-247311. industry mainly because of its slow kinetic. Some

0039-9140/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 9 - 9 1 4 0 ( 9 9 ) 0 0 2 6 1 - 1
282 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289
potential applications have been studied at labo-
ratory scale [1 – 3], but only a pilot plant for the
recovery of Cs – 134 and Sr-90 and concentration
of uranyl ions is known [4]. G.Sionkowski et al.
[5] have studied the transport of sulphates of
divalent metals and have shown that Donnan
dialysis technique is efficient in preconcentrating
ions. Applications of these studies were performed
in the same laboratory by Wòdzki et al. [3].
Recently, Chérif et al. [6] have studied the separa-
tion of Ag+ from Zn2 + by means of the same
technique and reported that CMS Neosepta mem-
brane presents a higher permeability to univalent
than bivalent cations. Donnan dialysis is generally
applied in analytical fields where the preconcen-
tration of various cations and anions are
performed.
However, Donnan dialysis was often studied by
means of cation exchange with H+ ion as counter
ion [1–6]. Concerning the cyanide medium, few
studies were carried out. Law [7] has examined the
recycling of gold cyanide baths using a neutralisa-
tion of the excess of KOH by acid produced by
Donnan dialysis. Hamil [8] has studied the recov-
ery of some heavy metals from electroplating
baths using anion exchange membranes that are
not commercialised yet. Separation of noble
metals using anion exchange membrane was in-
vestigated in thiocyanate, EDTA and ammonia
media by Brajter et al. [2,9,10]. In the same labo-
ratory [11], the modification of an anion exchange
membrane was performed to improve the separa-
tion of copper, zinc, nickel and manganese in
iminodiacetate medium by means of Donnan
dialysis.
In this work, the behaviour of copper, gold and
silver in a cyanide medium is examined during
Table 1
Characteristics of the anion exchange membranes
Membrane Exchange capacity (meq g(1) Water contents (%)
AMV 1.25 16
ACS 1.40 22
ADS 0.54 22
RAI 5035 0.68 16
ADP 0.90 18
Donnan dialysis. Five anion exchange membranes
are tested to determinate the optima conditions
for the separation of the cyanides complexes of
copper, gold and silver. The decrease of the free
cyanide concentration in the feed compartment is
also investigated.
2. Experimental
2.1. Membranes and reagents
The characteristics of the chosen commercial
anion exchange membranes are reported on Table
1. Details of their internal structures cannot be
known but they probably contain alkylammo-
nium functional groups and divinylbenzene
crosslinks. Their properties can be deduced from
measurements of their characteristics. According
to Cohen [12], it is shown that:
“ a high exchange capacity, increases the selectiv-
ity between univalent and multivalent anions
because of the higher repulsion charge.
“ A high thickness decreases the diffusion giving
rise to lower ions flux.
“ A high water content can decrease the permse-
lectivity and favours the penetration of bulky
ions.
All the chemicals used were pure reagent grade.
2.2. Donnan dialysis experiments
All Donnan dialysis experiments were carried
out using a laboratory cell consisting in two (or
three) detachable square compartments made of
Plexiglas. Between these compartments, an anion
exchange membrane was inserted. Its working
Thickness ×106 (m) Furnisher
133 Asahi glass (Japan)
150 Tokuyama Soda (Japan)
130 Solvay (France)
200 Raipore (USA)
130 Solvay (France)
 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289 283

ter. The analyses have been carried out at the

respective wavelengths once the cyanides were
oxidised with H2O2 and a quantity of H2SO4 0.1
M was added.
The free cyanide ion was titrated by a volumet-
ric method using silver nitrate as precipitate
reagent and potassium iodide as indicator.

3. Results and discussion

3.1. Influence of the anion exchange membrane

Fig. 1. Evolution of both total copper and free cyanide
concentration versus time in the stripping compartment.
*Donnan Dialysis performed using a feed solution of only one
metal in cyanide medium.
area was 16 cm2 and the temperature 25°C. Both
the feed solution and the stripping solution of,
respectively 250 cm3 volume were circulated by
means of peristaltic pumps through each compart-
ment where the volume was 10 cm3. NaOH 6N
was used as stripping solution for the two-com-
partment cell. When three compartments were
used the concentration of the successive stripping
solutions were, respectively 4 and 6 N. The feed
and strip solutions were sampled at regular time
interval during 8 h.
2.3. Analysis
Total metal quantities were determined using a
Perkin Elmer atomic absorption spectrophotome-
Table 2
Flux of both metals cyanide complex and their corresponding free cyanide as a function of the anion exchange membrane
Donnan dialyses of copper, gold and silver in
cyanide medium were carried out using five differ-
ent anion exchange membranes. Experiments were
performed for single ion metal component in pres-
ence of KCN 0.2 M at pH 9 in the feed solution.
The stripping solution was composed by NaOH 6
N. The global flux is determined and the corre-
sponding free cyanide flux is also measured. It is
noticed that the free cyanide flux is always higher
than those of the metallic complex flux as it
shown on the Fig. 1. For each anion exchange
membrane, the results are reported on the Table
2. They are showing that an increase of the selec-
tivity between univalent and multivalent anions is
observed when both the ACS and the ADP mem-
branes were used. It can be owed to the high
exchange capacity of these membranes and a
probable layer of negative charges that has been
fixed on one side of the membrane [12]. This layer
can induce a strong electrostatic repulsion to mul-
tivalent (superiors complexes of both copper and
silver) than to univalent ions (free cyanides and
inferiors complexes of copper, silver and gold). It

Membrane Flux ×108 (mol m−2 s−1)

Au (I) Cu (I) Ag (I)

J (Au) J(CN) J(Cu) J(CN) J(Ag) J(CN)

RAI 5035 1.12 23 1.09 44 1.20 13

AMV 1.45 39 1.22 39 3.88 58
ACS 4.70 79 0.07 39 1.56 22
ADP 5.22 21 0.62 56 5.97 16
ADS 0.27 27 1.45 21 1.34 14
284 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289
appears that the RAI 5035 is the membrane that
have the lowest permeability because its high
thickness and low water content. Moreover, its
low exchange capacity does not allow attaining a
good selectivity in the transfer of the different
charged species across the membrane. The ADS
membrane shows a high permselectivity for the
bulky ions regarding to its high water content
combined to its low exchange capacity. Wide
pores that permit the penetration of voluminous
ions probably constitute its internal structure.
However, the AMV membrane has an irregular
behaviour that is not linked to the ion charge or
to both of the volume and the geometry of the
complexes. Its relative high exchange capacity
would induce a high selectivity between univalent
and multivalent ions, however, its internal struc-
ture gives rise to a versatile behaviour which
depends on the operating conditions. The metals
studied give rise to several species that are de-
pending upon the free cyanide concentration. Ac-
cording to the initial operating conditions, a
variable number of species that can have both
different charges and geometry may be obtained.
Each one of these parameters can influence the
transfer and needs a particular study for the
choice of the membrane.
3.2. Transfer modelling
The transfer modelling of ion species in the
course of Donnan dialysis can be interpreted ac-
cording to two concepts:
The first is based on the establishment of linear
relations between the phenomenon and its cause.
These relations are resulting from the Nernst
Planck electrodiffusion equation:
Ji = − Di
dCi
+Ci
dLngi Zi FCi dE
+ .
n (1)
dx dx RT dx
With: Ji, the flux of a species i; Di, diffusion
coefficient of the species i in the membrane; Ci,
the concentration of species i inside the mem-
brane; gI, the activity coefficient of the species i;
zi, the electrochemical valence of the species i; F,
the Faraday constant; R, the universal gas con-
stant; E, electric potential of the solution.
However, this equation is limited by the un-
knowing of the species modification inside the
membrane.
The second is based on the analysis of the
exchange by means of the thermodynamic of the
irreversible processes.
In comparison to electrodialysis, the electric
field is absent although the second term of the
equation of electrodiffusion is not cancelled be-
cause the presence of an electrochemical potential
difference between the two sides of the membrane.
In the literature, several models have been pro-
posed to the determination of the theoretical flux
when Donnan dialysis is performed [2,13–24].
Sudoh et al. [16] have applied the Nernst
Planck equation for the modelling of copper
transfer in chloride medium with assuming that
the electromigration term is negligible. They have
also shown that the water transport by means of
the osmosis phenomenon is proportional to the
ionic strength between the two compartments. Ng
and Snyder [2] have shown the influence of the
concentration on the transfer velocity and had
proposed a model based upon the membrane
resistance to the mass transfer. In this modelling,
the flux are proportional to a coefficient m which
is also proportional to another coefficient K
(defined as a mass transfer coefficient) obtained
from both the flow rate and the instantaneous
concentration of an ion i. Wen and Hamil [17]
have proposed a kinetically model for the trans-
port of the counter ions represented by an equa-
tion that considers the proportionality of metal
rate transfer to its concentration. It was also
shown that this rate transfer increases with the
exchange capacity of the membrane. Cherif [18]
and Salem [19] have applied the Nernst Planck
equation with taking into account of the electro-
chemical potential difference. They assimilated the
Donnan dialysis cell to an electrochemical chain.
Ktari and Auclair [20] have used a simplified
model based upon the Fick law that is available
only to the univalent ions.
This model was studied by Miyoshi [21] that
has derived the Fick equation which represents a
relation between the dialysis time and the concen-
tration and applied his model to bivalent ions.
However, he neglected the involvement of the
 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289 285

electric potential caused by the diffusion of ions. −m

[M(CN)n ](1) + mOHm− v [M(CN)n ]m− m + mOH(1) −

In our case, this last phenomenon can be impor- (3)

tant by means of the presence of various com-
plexes species in one part of the membrane and on
the other. In another study, Miyoshi [22] applied
the Nernst Planck equation and new equations
were derived when the valence of feed ions is
different from that the driving ions. By this man-
ner, some diffusions coefficients of ions were ob-
tained [23], but the different equations were
applied to systems which contain univalent and
bivalent free ions. The alone study related to
solutions that contain complexes species was per-
formed by Aulas et al. [24]. They studied the ion
transfer in the course of Donnan dialysis using a
kinetically model which is described by the fol-
lowing equation:
Where (1) and m are meaning, respectively the feed
compartment and the membrane.
(3) Exchange of [M(CN)n ] − m (or CN−) with
the OH− ions sited in the stripping solution ac-
cording to the equilibrium:
[M(CN)n ]m− m + mOH(2)
−
v [M(CN)n ](2)
−m
+ mOHm−
(4)
where (2) is the stripping compartment.
(4) Diffusion of the species [M(CN)n ] − m (or
CN−) in the stripping compartment through the
unstirred layer.
The ion exchange is occurring very fast, then,
the diffusion phase would be the limiting stage of
the transfer kinetic. In the course of a Donnan

Ji = J 
i
exp
Ab −Ab e
− exp
A − A e  (2) M(CN)n(2)
−m

=

OH−
(1) 
dialysis, the equilibrium is attained when:
m
(5)
RT RT
M(CN)n(1) 
−m
OH−
(2)

Where Ji is the instantaneous flux of the ion This condition of the equilibrium depends upon
i, J i the exchange flux at equilibrium (when
the stationary state is attained with a constant
flux).
Ab − Ab e is the difference of affinity that repre-
sents the deviation from the equilibrium in the
feed solution “ stripping solution way.
and A −A e is the difference of affinity,
which represents the deviation from the equi-
the ions activity in the part and in the other of the
membrane. Then, in Eq. (2), the affinity differ-
ences are meaning in fact the chemical potential
differences of a species i in the part and the other
of the membrane at an instant t and at those of it
correspond to the equilibrium state. This allows
writing:
Ab − Ab e = mi 1 − m  (6)
librium in the stripping solution, “ feed solution (1)
i1

way. A − A e = mi 2 − m 
(2)
i2 (7)
This model will be used attempting to follow
Where (1) and (2) are, respectively the feed and
the simultaneous transfer of the complex species,
the stripping compartment, mi is the chemical
which are varied in our system. Moreover, the
potential of an ion i.
Nernst Planck equation cannot be used because
The relation which exists between the activity
the difficult to determine the diffusion coefficients
and the chemical potential of an ion i involves the
of each species inside the membrane and the


following equation:
repartition of these species which can vary at any
time of the Donnan dialysis experiments.
Ji = J 
i1 i2
− 
n (8)
The transfer of a species from the feed solution i
i
1 i2
to the stripping solution is occurring by means of
Knowing that diluted solutions are used in Don-
the following stages:
nan dialysis, activities can be confounded with
(1) transport of [M(CN)n ] − m (or CN−) until to
concentrations, which are expressed by the
the diffusion layer sited in the feed solution.
equation:
(2) Exchange of [M(CN)n ] − m (or CN−) with
the OH− ions fixed inside the membrane accord- N
[i ] = (9)
ing to the equilibrium: V
286 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289

Table 3
Theoretical and experimental flux of both metal in cyanide medium and free cyanide in the course of Donnan dialysis during 8 ha

Membrane Flux ×108 (mol m(2 s(1)

J(Ag (I)) J(CN) J(Au (I)) J(CN) J(Cu (I)) J(CN)

RAI 5035 1.20 (1.81) 13 (32) 1.12 (1.30) 23 (40) 1.09 (3.47) 44 (60)
AMV 3.88 (14.36) 58 (115) 1.45 (1.60) 39 (61) 1.22 (6.91) 39 (142)
ACS 1.56 (1.75) 22 (8.69) 4.70 (5.90) 79 (83) 0.07 (0.30) 39 (142)
ADP 5.97 (10.53) 16 (4.52) 5.22 (6.10) 21 (38) 0.62 (3.05) 21 (41)
ADS 1.34 (3.21) 14 (6.10) 0.27 (0.34) 27 (51) 1.45 (3.50) 56 (220)

a
Experimental (theoretical).
where N is the mole number of the ion i and V the Then, the drawing of Ln(N  1 − N1)= f(t) or
volume of the compartment. Then, Eq. (8) Ln(N  2 − N 2 )= f(t) allows to attain theoretical
becomes: J . The results obtained for copper (I), gold (I)

i

Ji = J 
N1 N2
− 
n (10)
and silver (I) using different membranes in
cyanide medium are reported in Table 3.
N
i
1 N2 It is noticed that the model is in good agree-
where N1 and N2 are, respectively the mole num- ment with the experimental results when gold (I)
ber of the ion i in the feed and the stripping is the metal studied. This fact owes that it forms
compartment at an instant t, and N  
1 and N 2 are
only one species with cyanide ligand. In effect,
these numbers at the equilibrium state (stationary copper and silver are giving rise to many species,
state). which can be dissociated in the course of the
If N 01 is the initial mole number of the ion i, it experiments.
can be written: Theoretical flux of free cyanides are generally
higher than experimental flux. Two factors can
N 01 =N1 + N2 =N  
1 +N 2 (11) explain these results:
Then, knowing that S is the active area of the “ In the presence of bulky species a concentra-
membrane and Ji the mole number of the ion i tion polarisation phenomenon appears at the
which appears in the stripping compartment (and interface membrane – solution and disturbs the
which leave the feed compartment) by both area crossing of free cyanide through the anion
and time unity, the following relation can be exchange membrane.
established: “ The free cyanide which cross through the mem-
brane until the stripping compartment could be
dN1 dN linked to the metallic complexes, then their
S Ji = − =+ 2 (12)
dt dt concentration decrease during the experiments
The model used does not taking into account of
By means of this last equation, the replacement of
these factors, then it does not describe rigorously
Ji by its expression used in the Eq. (10) and
the transfer of the free cyanide. It is available for
integration, we get:

N1 = N  
1 + N 2 . exp
 
−S
1 1
+  J t
 n independent species.

N1 N 2
i
3.3. Application to the separation of metallic
(13) complexes

   n
and
3.3.1. Separation of Ag(I) –Au(I) –Cu(I) in
 
1 1 cyanide medium
N2 = N −N 2 2 . exp −S 
+  J
i t
N1 N2 Experiments of separation between copper,
(14) gold and silver in cyanide medium were carried
 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289 287

Table 4 last membrane gives rise to the best separation

Separation factors of the cyanometallic complex in the course
between the metals studied. However, these re-
of Donnan dialysis
sults are available only for the equimolar mixtures
Membrane F Cu
Au F Cu
Ag F Au
Ag
and the concentration can influence the separation
efficiency. The difference between the results ob-
ACS 0.26 0.47 1.83 tained for the AMV and the ACS shows also the
AMV 0.25 0.12 0.52 importance of the nature of the membrane in the
separation performing.
out using equimolar mixtures of Ag (I), Au (I)
and Cu (I) 10 − 3M in presence of KCN 0.2 M. 3.3.2. Separation Ag(I) –Au (I) –Cu (I) by means
The pH was fixed to 11 by adding of NaOH and of three compartments
the Donnan dialysis were performed during 6 h. The use of a third compartment as a second
Separations factors of the metals, two by two, stripping solution is examined to improve the
have been determined for three anion exchange separation. The experiments are applied on the
membranes. The separation factors were calcu- same mixtures used initially. NaOH 4 N com-
lated according to the following relation (consid- posed the first stripping compartment while the
ering two elements A and B): second contained NaOH 6 N. The initial compo-

FA
B=
,
[A] 0 [A] t
(15)
[B] 0 [B] t
where [A]0 and [B]0 are, respectively the initial
concentrations of A and B in the feed compart-
ment. [A]t and [B]t are the concentrations of the
same elements at an instant t of the dialysis. The
separation is efficient when the separation factor
is so far off the unity.
The results obtained are reported on the Table
4. They are showing that the more efficient sepa-
ration is occurring between copper and silver
when the AMV membrane is used. Moreover, this
sition of the feed solution can be detailed as
follows: Au (CN)−
10 −4
2 = 10
−3
M; Ag (CN)− 2 =5×
M; Ag (CN)3 = 5×10 − 4 M; Cu (CN)23 −
2−
= 3×10 − 5 M Cu (CN)34 − = 9.7× 10 − 4 M and
free cyanides equal to 0.192 M.
Tables 5 and 6 gives the percentage of each
metal present in each compartment after 6 h of
Donnan dialysis. The results are presented for
three anion exchangers membranes and they are
disposed by manner to do a comparison with the
experiments that use two compartments.
The results show again that the AMV is the
most appropriated membrane for the separation
of gold from the two others metals. All separa-

Table 5
Repartition of the metals inside compartments after 6 h of Donnan dialysisa

Membrane Metal Percentage in the feed solution (%) Percentage in the stripping solution (%)

AMV Silver 37.6 62.4

Copper 44.1 59.9
Gold 80.5 19.5
ADS Silver 51.5 48.5
Copper 41.8 58.2
Gold 25.9 74.1
ACS Silver 47.3 52.7
Copper 59.1 40.9
Gold 43.5 56.5

a
Donnan dialysis with two compartments.
288 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289

Table 6
Repartition of the metals inside compartments after 6 h of Donnan dialysisa

Membrane Metal Percentage in the feed Percentage in the first stripping Percentage in the second stripping
solution (%) olution (%) solution (%)

AMV Silver 39.6 43.2 17.2

Copper 44.1 44.7 11.2
Gold 91.1 8.2 0.7
ADS Silver 25.8 38.7 36.1
Copper 37.9 41.8 32.7
Gold 21.5 53.9 24.6
ACS Silver 33.1 38.1 28.8
Copper 20.6 53.1 26.3
Gold 35.7 29.3 35.0

a
Donnan dialysis with three compartments.

tions are improved when three compartments are
used, particularly for the ACS membrane. It is
also noticed, for the system using three compart-
ments and for this last membrane, that the pres-
ence of copper in the first stripping compartment
is very important. This result is unexpected since
the ACS membrane is supposed permselective to
the univalent ions regarding the multivalent. That
is to be owed to the fast transfer of free cyanides
from the feed compartment which provokes the
dissociation of Cu (CN)23 − and Cu (CN)34 − into Cu
(CN)− 2 which cross easily through the ACS
membrane.
Thus, an interesting perspective is offered for
the separation of cyanide complexes by means of
Donnan dialysis. This technique is therefore as-
similated to that the chromatography. It can
therefore be optimised by the research of the
number of compartments as it determines the
number of theoretical trays necessary for a
separation.
3.3.3. Application to the decyanidation
To estimate the removal of total cyanides from
solutions by means of Donnan dialysis, the fol-
lowing relation was used:
[CN−]Total initial −[CN−]Total at an instant t
t=
[CN−]Total initial
× 100
where the concentrations represented are from the
feed compartment initially and at an instant t of
the Donnan dialysis. t is the percentage of the
total cyanides ions that have cross through the
membrane to the stripping compartment. The re-
sults are reported in Table 7. They have been
calculated using two membranes for each metal in
cyanide medium. They are showing the poor
yields obtained (B5%) deducing that Donnan
dialysis is not an efficient technique for the de-
cyanidation. It is only adapted to the separation
of metallic complexes founded in diluted solutions
in presence of cyanides. This result could be ex-
plained by the high mobility of free cyanides that
can go back to the feed compartment with OH−
ions.
4. Conclusion
In this work, it is shown that Donnan dialysis
could be an efficient technique to separate copper,
gold and silver in cyanide medium. This result is
very interesting when it knows that the cyanide
Table 7
Decyanidation yield by means of Donnan dialysis as a func-
tion of both the metal present and the membrane
Membrane Decyanidation yield (%) in presence of
Ag (I) Au (I) Cu(I)
(16)
AMV 4.51 1.56 2.61
ACS 1.71 2.76 2.68
 D.-E. Akretche, H. Kerdjoudj / Talanta 51 (2000) 281–289
ion is the usual reagent for gold ore leachates.
Then, this membrane process can find an applica-
tion particularly when the solutions are very di-
luted. It can compete the conventional ion
exchange resin process.
After showing the role of the choice of the
anion exchanger membrane, the transfer mod-
elling has been examined and it shows the
difficulty to apply a mathematical model to a
system where several complexes species are in
solution. The difficulty is more high by the fact
that the composition of the solution can vary in
the course of the experiments.
The use of a third compartment to improve the
separation have shown that it is possible to apply
Donnan dialysis as chromatography when
parameters as the nature of the membrane and
the concentration inside the stripping compart-
ments are optimised.
The advantage of this technique is also the
possibility of the recycling of all reactants with a
good impact to the environment. However, Don-
nan dialysis depends greatly on the experimental
conditions, thus, before each application a simula-
tion study is necessary to fix the functioning
parameters
References
[1] R.M. Wallace, I&EC Process Design Dev. 6 (4) (1967)
423 – 431.
289
[2] K. Pyrzynska, Anal. Chim. Acta 255 (1991) 169 – 175.
[3] R. Wodzki, G. Sionkowski, T. Hudzik-Pieta, Polish J.
Environ. Studies 5 (4) (1996) 45 – 50.
[4] M.H. Kelly, D. Randall, R.M. Wallace, Bull. Union
Phys. 80 (1986) 1427 – 1447.
[5] G. Sionkowski, R. Wodzki, Sep. Sci. Technol. 30 (5)
(1995) 805 – 820.
[6] A.T. Chérif, C. Gavach, J. Molénat, A. Elmidaoui, Ta-
lanta 46 (1998) 1605 – 1611.
[7] H.H. Law, Plat Surf Finish 11 (1982) 72 – 73.
[8] Hamil HF, US Report No. EPA-600 12-82-098, 1982, pp.
31.
[9] Brajter K, Slonawska K. Chemical Separations. Litarvan
Literature, 1986, pp. 307 – 317.
[10] K. Brajter, K. Slonawska, J.A. Cox, Anal. Lett. 22 (1989)
779 – 792.
[11] B. Sawicka, K. Brajter, M. Trojanowicz, B. Kado, Sep.
Sci. Technol. 26 (5) (1991) 717 – 728.
[12] Cohen T. Doctorate Thesis. Paris XII, France, 1988.
[13] M.A. Lake, S.S. Melsheimer, Aiche J. 24 (1) (1978)
130 – 137.
[14] E.H. Cwirko, R.G. Carbonell, J. Memb. Sci. 48 (1990)
155 – 179.
[15] R.E. Meyer, P.M. Lantz, J. Electroanal. Chem. 61 (1975)
155 – 163.
[16] M. Sudoh, H. Kamei, S. Nakamura, J. Chem. Eng. Jpn.
20 (1) (1987) 34 – 40.
[17] C.P. Wen, F. Hamil, J. Memb. Sci. 8 (1981) 1714 –1725.
[18] Chérif AT. Doctorate Thesis, U.S.T.H.B, Algiers, Alge-
ria, 1990.
[19] Salem K. Doctorate Thesis. Montpellier II, France, 1993.
[20] T. Ktari, B. Auclair, J. Memb. Sci. 32 (1987) 251 –265.
[21] H. Miyoshi, Sep. Sci. Technol. 31 (1996) 2117 – 2130.
[22] H. Miyoshi, Chem. Eng. Sci. 52 (1997) 1087 – 2002.
[23] H. Miyoshi, J. Memb. Sci. 141 (1998) 101 – 110.
[24] F. Aulas, B. Tyburce, M. Rumeau, M. Renaud, Entropie
93 (1980) 15 – 21.