319

Emulsion Liquid Membranes Part, I : Phenomenon and Industrial Application

Fhssig - Membran - Permeation Teil I : Verfahrensprinzip und industrielle Anwendung
J. DRAXLER and R. MARR
Institutfiir Verfahrenstechnik, Technische Universittit Graz, Inffeldgasse25,8010 Graz (Austria)

(Received December 16,1985;

in final form July 1,1986)

Abstract Emulsion liquid membrane processes are discussed. The characteristics of the processes, research activities and commercial applications are surveyed and the advantages and disadvantages are outlined. The first large-scale plant for the separation of zinc is described as well as several pilot plants for the separation of various metals from aqueous solutions.

Kurzfassung Es wird ein iiberblick iiber das Gebiet der fliissigen Membranen gegeben. Die Besonderheiten dieses neuen Prozesses werden ausfiihrlich diskutiert sowie die Forschungsaktivit%ten und die industriellen Anwendungen aufgezeigt. Die erste grosstechnische Anlage wird beschrieben, welche zur Riickgewinnung von Zink aus Viskoseabwtissern gebaut wird, sowie verschiedene Pilotanlagen fiir die Riickgewinnung von Metallen aus verschiedenen industriellen Prozess- und Abwlssern.

Synapse
Mit dem Bau der ersten grosstechnischen Anlage mit jliissigen Membranen nnch der Methode der multiplen Emulsionen ist der Fliissig-Membran-Permeation der Durchbruch zu einem industriellen Verfahren gelungen. In dieser Aniage, welche zur Riickgewinnung von Zink aus Viskoseabwtissern gebaut wird, wird erstmals ein Extraktionsmittel eingesetzt, welches auf Grund einer extrem langsamen Stripreaktion bisher in der SolventExtraktion keinen Eingang gefunden hat. In der Permeation mit multiplen Emulsionen ist dieses Extraktionsmittel hingegen sehr gut gee&net, da auf Grund der ausserordentlich grossen StoffaustauschjLYche der inneren Phasengrenze (ca. lo6 m21m3) die Stripreaktion ausreichend schnell abtiuft. Besonders deutlich wird dies bei der Nickel-Permeation. Obwohl das Gleichgewicht fiir die Extraktion von Zink und Nickel mit diesem Extraktionsmittel nahezu ident ist (Abb. 7) verlriirft die Stripreaktion fiir Nickel noch bedeutend kmgsamer, wodurch die Vorteile der Fliissig-Membran-Permeation gegeniiber der Solvent-Extraktion noch deutlicher sichtbar werden. Entscheidend fiir den erfolgreichen Einsatz der Fltissig-Membran-Permeation war die Entwicklung einer einfachen und kostengiinstigen Spaltung der prirmiren Emulsion. Obwohl hierfiir zahlreiche Miiglichkeiten in 0255-2101/86/%3.50

Betracht kommen, hat sich nur die Spaltung im elektrostatischen Feld durchsetzen Mnnen, welche bei geringem Energieverbrauch eine schnelle und vollst$ndige Spaltung der Emulsion ermiiglicht. Abbildungen 4 und 8 zeigen schematisch zwei verschiedene Anordnungen zur Emulsionsspaltung im elektrischen Feld mit isolierten und nicht isolierten Elektroden. Obwohl mit der elektrischen Spaltung der letzte Schritt in der Entwicklung der jliissigen Membranen nuch der Methode der multiplen Emulsionen entscheiden verbessert werden konnte, kann die Fliissig-Mem branPermeation bisher nur fiir die Abtrennung von Zink und Kupfer als optimiert und reif fiir eine industrielle Anwendung betrachtet werden. Bei allen anderen Elementen ergeben sich immer wieder Schwierigkeiten, welche hauptstichlich auf das Tensid, das zur Stabilisierung der prim&en Emulsion verwendet wird, zuriickgefiihrt werden. Tenside fiir die Fliissig-Membran-Permeation wurden bisher ausnuhmslos nach ihrer Fiihigkeit ausgewrihlt, stabile Emulsionen zu bidden Es zeigt sich aber immer mehr, dass dies allein zu wenig ist und andere Kriterien wie Wassertransport, Stofftransportwiderstand, Wasserltislichkeit, Widerstandsftihkkeit gegeniiber Bakterien und der katalytische Effekt auf die Zersetzung des Extraktionsmittels, beriicksichtigt werden miissen. Bei der Apparateauswahl ist zu beticksichtigen, dass man doch Gegenstromapparate verwenden ~011, obwohl
GElsevier Sequoia/Printed in The Netherlands Ty

Chem. Eng. Process., 20 (1986) 319-329

320 die Permeation im aligemeinen als ein einstufiger transport- und nicht gleichgewichtslimitierter Prozess angesehen wird. Fiir den untersten Konzentrationsbereich ist dies aber nicht mehr zutreffend, so dass zur Erreichung der in den meisten Lrindern vorgeschriebenen niedrigen Grenzwerte fiir die Einleitung von Metallen in die Vorfluter mehrstufige Apparate notwendig sind. present) the diluents and extractants are washed out and the membranes become ineffective within fourteen days. Recently, attempts have been made to repiace the lost organic phase continuously in the pores [4]. This technique seems to be very promising, but is just in the beginning stages and up to now no economic data have been published. In a similar device - the creeping film pertractor, developed by Boyadzievh [5] the two aqueous phases are sucked in and pumped continuously through two porous rigid supports of high hydrophility, and are separated by the liquid organic membrane phase, which flows counter-currently to the two aqueous phases. Emulsion liquid membranes could be developed as an industrial process, especially since the splitting of the emulsion can be done effectively and cheaply [6]. The fist plants on an industrial scale will be built for the recovery of metals from diluted aqueous solutions, but there are a lot of other possibilities for the use of emulsion liquid membranes. There seem to be many applications in biotechnology and medicine fol the controlled release of active substances, and also in waste water treatment for the recovery of ammonia and phenols [7,81.

1. Introduction
Many membrane based separation processes suffer from low specific transport rates resulting from low diffusivities in non-porous solid membranes. To overcome this problem liquid membranes have been developed which offer much higher transport rates because of the low viscosity and large mass transfer area. There have been many attempts to realize liquid membranes, but only two different methods have succeeded: emulsion liquid membranes or liquid surfactant membranes, which were first developed by Li [ 11, and supported liquid membranes [2,3]. Figure 1 shows these two types of liquid membranes. Emulsion liquid membranes are prepared by dispersing a primary emulsion - usually of the water-in-oil type - in a second aqueous phase. The organic phase, which consists of a low viscosity diluent, a surfactant to stabilize the primary emulsion and sometimes an extractant, separates the two aqueous phases and acts as a liquid membrane. The emulsified aqueous phase is usually the receiving phase and the continuous aqueous phase the feed phase. Supported liquid membranes can be obtained when a rigid porous substrate is filled with an organic solution of the extractant. The substrate can be in the form of a flat sheet forming the divider in a multi-compartment cell or as hollow fibres with porous walls, which, made into modules, allow reactors with high surface to volume ratios. With supported liquid membranes there is no need to prepare and split an emulsion, so they seem to have the advantage over emulsion liquid membranes, but they have one decisive shortcoming, which has not been overcome up to now, and this is the short lifetime of the membranes. By solubility or by intrusion of water in the pores of the substrate (especially when surfactants are

2. Mass

transfer mechanism

Liquid membranes are homogeneous, non-porous membranes, where the solute is dissolved at one side of the membrane and released at the other side. The driving force is given by the concentration difference of the solute at both interfaces. By adding an extractant (carrier). which reacts with the solute at the two interfaces, this concentration difference can be enhanced and the mass transfer accelerated (facilitated transport). For the separation of metals, coupled transport is the most important (Fig. 2). In this case the carrier reacts not only with the solute but also with an auxiliary component and thus a high concentration difference can be maintained for a long time. By this auxiliary component the concentration difference can be maintained even if the concentration of the solute in the receiving phase becomes higher- than in the feed phase. A detailed description of the various mass transfer models describing the mass transfer in emulsion drops is given in Part II of this paper [9].

Liquid

membrane

Reaction: Phase III

CuZ*+2HXSCuX, Phase II

+2H+ Phase I

supported liquid membrane Fig.

emulsion Liquid membrane Fig. 2 Coupled transport.

1, Types of liquid membranes.

321 3. Membrane composition For many applications the membrane consists only of a diluent and a surfactant to stabilize the primary emulsion. For the separation of metals from aqueous solutions additional extractants are needed. Normally the extractants are the same as those used in solvent extraction, but there may also be some differences. The extractant concentration in the membrane phase is about l%&%, and because of this low concentration no modifiers are needed. Only if a quaternary ammonium salt like Aliquat 336 is used as a carrier (e.g. for vanadium permeation) it might be necessary to use small amounts of a modifier to prevent third-phase formation after splitting of the emulsion. Diluen t Although experiments have been performed with aromatic diluents [lo], aliphatic diluents are generally preferred because of the lower solubility in water and better emulsion stability. The mass transfer is in many cases diffusion controlled, and therefore the viscosity of the diluent is an important parameter. Suitable diluents are kerosene fractions like Shellsol T with a viscosity of l-2 CP. Surfactant The surfactant is the most important component in the emulsion liquid membrane process. It determines not only the stability of the emulsion but also influences a lot of other parameters like osmosis and water solubility and often represents a high mass transfer resistance. It is interesting to note that in the development of the liquid membrane process relatively little attention had been paid to the search for a suitable surfactant. With few exceptions only two surfactants have been used so far: SPAN 80 and ECA 4360, or mixtures thereof. Both are good emulsifying agents but both suffer from some major disadvantages. SPAN 80, sorbitan mono-oleate, is a good carrier for water molecules and therefore favours swelling of the emulsion by osmosis. Furthermore, it has a high solubility in water and is not suitable for emulsions with NaOH as the dispersed aqueous phase. ECA 4360, a polyamine, does not suffer from these shortcomings, the water transport is minimized, losses by water solubility are very low and stable emulsions with NaOH as dispersed phase are possible. But as it is alkaline it may react with organic and inorganic acids and may represent a high mass transfer resistance. As is shown in the section on copper permeation, ECA 4360 is also an accelerator for decomposition of the extractant by acids. These disadvantages make it necessary to look for other surfactants. This has not been done so far, because the stability of the emulsion was originally the only parameter used in the choice of a surfactant and both SPAN 80 and ECA 4360 were able to satisfy this demand. The shortcomings of these surfactants only became apparent when attempts were made to prepare the emulsion liquid membranes for industrial application. Extractants When emulsion liquid membranes are used for the separation and recovery of metal ions, this process may be compared with solvent extraction, where extraction takes place at the outer interphase and stripping at the inner interphase. It may be regarded as a combined extraction/stripping process and therefore it is surprising that only in some cases may the same extractants be used. This is due to the different mechanisms of the extraction and permeation processes. Solvating agents like tributyl phosphate, tri-octylphosphine oxide, etc., which are widely used in solvent extraction, cannot be used in the liquid membrane process because with these agents the coupled transport mechanism cannot be realized. An example of this is the extraction of germanium from high acidic solutions. An extractant for this is LIX 63 which acts under these high acidic conditions as solvating agent. Stripping is done with NaOH solution. This process cannot be done by emulsion liquid membranes because in the preparation of the primary emulsion - NaOH solution in organic membrane phase containing LIX 63 - LIX 63 is converted into its sodium salt and can no longer act as a solvating agent. For a similar reason the low acidic organophosphorous extractants (phosphonic and phosphinic acids) have only a restricted applicability in emulsion liquid membranes. In solvent extraction these extractants are used as sodium or calcium salts to maintain a constant pH. By preparing an emulsion with a mineral acid as stripping agent, the free acid would be released from its salt and by the ion exchange of the metals with the extractants the pH in the feed phase rapidly drops under the efficacy of the extractants. But there also exist extractants which may only be used in the emulsion liquid membrane process. This surprising fact results from the very large inner interfacial area of the emulsion (more than lo6 m2/m3 emulsion). When the rate controlling step is the stripping reaction at the interphase the transport rate in emulsion liquid membranes is much higher than in solvent extraction, when the extraction is carried out in a conventional extraction apparatus. Bis(2-ethylhexyl)dithiophosphoric acid is a very strong extractant for metal ions but nevertheless it is nearly unknown in solvent extraction, because the metals cannot be recovered from the loaded organic phase. At least for zinc and nickel extraction this is not due to the bad equilibrium conditions, but to the low stripping kinetics. Figure 7 shows the equilibrium isotherms for the extraction reaction of zinc and nickel with sulphuric acid, which were obtained after 5 min of extensive mixing in a homogenizer. If stripping is done in a shaker it takes 4 hours to reach equilibrium. This would be much too long in a conventional extraction apparatus. 4. Emulsion properties An emulsion drop, as shown in Fig. 1, has an ideal form. The drop diameter varies between 0.1 and 2 mm

322 and the diameter of the dispersed droplets between 0.5 and 10~m. Unfortunately, the emulsion drop will not be able to keep its form; it will undergo some changes during the mass transfer [ 111. The two most important changes are a decrease of the emulsion volume due to break-up of the dipsersed droplets and an increase of the emulsion volume caused by water transfer from the outer to the inner aqueous phase. Break-up of the internal droplets By increasing the concentration of the surfactant and the dissipated energy it is possible to prepare emulsions with very high stability, where no break-up occurs within several months. But this is unfavourable because the emulsion has to be broken to recover the concentrated metals in the internal phase. A compromise has to be found between sufficient stability of the emulsion and easy breakage. Emulsions which satisfy this demand normally have a break-up in the range O.Ol%-0.1% of the internal phase depending on the applied shear force. By use of anion exchangers and NaOH as stripping agent, the break-up is slightly higher and can reach I%-2%. But in general the emulsions are so stable that break-up of the internal droplets may be neglected. Osmosis The water transport through the membrane phase is also a very unfavourable effect, because the concentrated inner solution is diluted again and the driving force is diminished. This often prevents highly concentrated solutions being obtained. Moreover, the rheological properties of the emulsions are changed, which might cause problems with pumps and pipes. This water transport is made possible by the extractants and surfactants which both act as water carriers. ECA 4360 transports much less water than SPAN 80 and for this reason it should be preferred. But the extractants can also transport a lot of water and therefore other liquid membrane systems such as supported liquid membranes will not be much better. There are not many problems with cation exchange systems - sometimes it is even desirable to have a small amount of osmosis to have an outlet in a closed cycle ~ but with anion exchange systems problems might occur. Especially in the system using tertiary amine as extractant and NaOH solution as stripping agent, a lot of osmosis is possible, depending on the extractant and NaOH concentrations and the ionic strength in the outer aqueous phase (Fig. 3). Applicability of this system on an industrial scale will mainly depend on finding a solution to this problem.
Phase III : 700 ml WC$,- soiu
tion

Volume 100,

of inner phase [ml1

so-

Phase

25ml

NaOH

5n

Fig.

3. Osmosis in the system

tertiary

amine-NaOH.

3-8 ppm carbon depending on the components used, the concentration of inert salts and the pH of the aqueous phases. This is much less than is usual in solvent extraction, because even with the low soluble aliphatic diluents no modifier is necessary. To reach these low solubilities, ECA 4360 must be used as the surfactant and not SPAN 80, which has a much higher solubility. These low values allow emulsion liquid membranes to be used even if no biological treatment of the waste water follows prior to passing it into the draining canal.

6. Emulsion splitting
Although there have been some interesting attempts to apply the emulsion liquid membrane technique without splitting the emulsion [12, 131, it seems impossible in most problems to avoid splitting to recover the concentrated solutes. Only the development of an effective and cheap method to split W/O emulsions made it possible to consider an industrial application. Many methods have been tried 114, 1.51, among them thermal and chemical splitting, centrifugation and ultrasonics. Chemical splitting by demulsifiers was found to be very effective, but the recovery of the demulsifier in a continuous process by distillation was too expensive for industrial application. Only adoption of the wellknown electrostatic splitting of water from crude oil in the emulsion liquid membrane technique added decisively to the feasibility of the whole process. Many devices have been suggested [ 161 with which to apply direct and alternating current. The mechanism of the splitting is not very clear and it seems as if not only the applied voltage but also the frequency has an important influence [ 171. Figure 4 shows a possible device with insulated electrodes. Insulated electrodes are to be preferred because of the lower energy consumption and no .possibility of electrolysis, but it must be ensured that the insulation is safe against short-circuiting. The energy consumption in this apparatus lies between 0.5 and 5 kW h/m3 emulsion.

5. Solubility of the membrane phase

Especially when emulsion liquid membranes are used in waste water treatment, the solubility of the membrane components in the aqueous phases is of great importance. In general, the solubility lies in the range

323
emulsion electrodes

Equilibrium:

feed phase receiving phase

0.01 g/l Zn, pH = 1.5 100 g/l Zn, pH = -0.5

insulation 3 kV 0,s kWhl&emulsion

organic

phase

>

-+?q~eous

phas;

Fig. 4. Electrostatic

splitter with insulated electrodes.

7. Apparatus If the primary emulsion is considered to be a homogeneous organic phase, the emulsion liquid membrane process looks identical to a conventional solvent extraction process. Therefore the apparatus for the membrane process should be the same as in solvent extraction. But the similarity of these two processes is only superficial. It is restricted to the hydrodynamic behaviour, although there exist some differences in that too. The mechanism of the mass transfer is quite different and therefore suitable equipment has to be chosen very carefully. Mass transfer by emulsion liquid membranes is, like all other membrane separation processes, commonly regarded as a rate process. The separation is accomplished by a driving force, not by equilibrium between phases. This would mean that the emulsion liquid membrane process is synonymous with a reaction process, where only the residence time is decisive for a certain yield of products. In this case a stirring vessel would be a suitable apparatus, where the feed and emulsion phases flow co- or cross-currently through the apparatus and are separated in a settler afterwards. Indeed, for many applications this is a suitable piece of equipment and, as can be seen from many publications, such an apparatus is used in most cases. But if the metal species in the feed phase is to be separated down to concentrations which meet the requirements of the environmental protection agencies, such a co- or crosscurrent mixer-settler is not sufficient; a counter-current mixer-settler would be needed in this case. To understand this, we shall examine the equilibrium for coupled, carrier-mediated transport. condition When the metal ion and auxiliary component are transported, counter-currently through the membrane phase, equilibrium is given by [ 181 V
cA.I _= CA, III CB.1

Under these conditions the maximum zinc separation in the feed phase is 99.0%, which is equivalent to 10 mg/l Zn. Even if this equilibrium concentration could eventually be reached, it would be higher than the limits set by the environmental protection agencies in most countries. Using a counter-current process, where the 10 ms/l Zn are in contact with a fresh receiving phase and not with a receiving phase already containing 100 g/l Zn, as in a co-current process, the zinc concentration can easily be reduced to less than 1 mg/l without affecting the enrichment in the receiving phase. When a counter-current mixer-settler battery must be applied, the settler between each mixer should be designed specially. In contrast to mixer-settlers in solvent extraction, it is necessary that the interface in the settler is renewed continually. This can be achieved when the two phases leaving the mixer enter the settler at the bottom, so the dispersed emulsion phase may coalesce at the organic/aqueous interphase. Of course, counter-current columns may be used instead of mixer-settlers. In this case such a permeation column exhibits two essential differences when compared to conventional extraction columns: (1) As can be deduced from the advancing front model [ 191, renewal of the surface of the emulsion drop has considerable influence on the permeation kinetics, because the solute need not diffuse to the inner core of the emulsion drop, but only to the outer layer of the dispersed inner droplets. Thus, extraction columns like the Kiihni or SHE extractors [20] utilizing the mixer-settler principle with repeated coalescence and redispersion of the emulsion drops, should be especially suitable for emulsion liquid membrane processes. Figure 5 shows a comparison of a SHE extractor and a permeation column having only stirrers, and no internals for the separation of zinc, in a pilot plant at a
Zn tglll

0 0

SHE Permeator

( cB,III i As an example, the following data are assumed the separation of zinc out of waste water: A = Zn*+, Start: B = H*, V= 2 1 m3 h-l, 1 g/l Zn, pH Y 4

for

feed phase receiving phase

0.01 m3 h-, Zn = 0, pH <-OS

Fig. 5. Comparison column.

of a SHE extractor

with

a permeation

324
Yield !%I

TABLE 1. Compari~~on between liquid membrane permeation (LMP) and solvent extraction Advantages of LMP Lower investment costs Lower operating costs Disadvantages of LMP No scrubbing stages Non-ideal bebaviour of the emulsion Restricted applicability of sobating agents

loo)cYT==
50
0+ 0

Little affected by solid particles and various modifiers

Lower solubility of organic compounds

V
5 i

Hlml

Fig. 6. Influence of backmixing in a permeation column.

comparable specific throughput. It can be seen that the same degree of separation can be achieved in the SHE extractor after 2 m instead of 2.7 m in the permeation column. But despite this evident advantage, it is probably more economic to build large-scale permeation columns than those which operate according to the mixer-settler principle. A two or three metre higher permeation column with no internals is surely cheaper than a Kiihni or SHE extractor. (2) Since the emulsion liquid membrane process is mainly a rate process, backmixing is of no great importance. Assuming a rate controlling reaction at the interface, the difference between plug flow and backmixing has been calculated for a column with a diameter of 1.6 m and an active height of 10 m, and plotted in Fig. 6. Unlike the situation in extraction columns, this difference is very small, so that the scale-up of permcation columns is much easier than for extraction columns.

Operating costs have always been regarded to be about the same 1211, but the calculations were based on the high costs of splitting the emulsions, which can now be considerably reduced, thus bringing the operating costs lower than in solvent extraction. The main advantage of the emulsion liquid membrane process, combination of the extraction and stripping in because no one step, is also the main disadvantage; scrubbing stages and p1-l regulations are possible. This is a major drawback, because the possibilities of application are restricted. For instance, the separation of Co and Ni causes no problems in solvent extraction. but up to now it has been impossible with liquid membranes. The same is true, as mentioned above. with the solvating extractants. In general, it can be said that solvent extraction has more possible applications, because scrubbing stages and pH regulation are very often necessary, and thus liquid membranes will never be able to replace solvent extraction completely. But when it is possible to use emulsion liquid membranes, this technique will surely be, as has been confirmed in many pilot plants, the cheaper and more economical process. In the following sections, pilot plants for the separation of- various metal ions out of industrial waste waters or process streams are described; these are operated in Austrian industry and at the Institute of Chemical Engineering, TU Graz.
Zinc permeation

8. Economics
The profitability of an emulsion liquid membrane system depends on many parameters, so it can only be calculated for specific problems. Especially for metal ion separation, it can be compared with solvent extraction, and from this comparison an evaluation can be made as to which process is more economical, but in general nothing is said about the profitability, which will depend to a great extent on the current metal prices. Table 1 shows a qualitative comparison between solvent extraction and liquid membrane permeation. The lower investment costs in liquid membrane permeation are due to the srnaller amount and size of the equipment compared with solvent extraction apparatus because extraction and stripping is combined in one step. Furthermore, the different mass transfer mechanism requires a much smaller organic extractant input, which is a considerable factor in the economy of the process. Many applications exist for the separation of zinc from different aqueous solutions. Although the price of zinc has fluctuated a lot in recent years, there is increasing interest in the recovery of zinc, which is also enhanced by the tightened laws for environmental protection in many countries. In sulphate media zinc can be removed by cation exchangers such as some organophosphorous acids. Which of these organophosphorous acids is applied depends primarily on the concentration of zinc in the feed, but also on the co-extracted elements. From the reaction equation of zinc with an organophosphorous acid: Zn2+ + ZHDEHP e Zn(DEHP)2

where HDEHP = di(2-ethylhexyl)phosphoric acid it can be seen that for every molecule of zinc extracted two

325 protons are released, and that the pH of the feed solution is therefore lowered. According to the acidity of the various organophosphorous compounds, they extract in different pH ranges. Organophosphinic acids do not extract zinc below a pH of about 3, i.e. when only 50 mg/l zinc are extracted the pH drops under 3, and no more zinc can be extracted with organophosphinic acids. The acidity of the organophosphorous compounds is increased in the sequence: phosphinic < phosphonic < phosphoric < dithiophosphinic < dithiophosphonic < dithiophosphoric acid. With organophosphoric acids about 300-400 mgjl zinc can be recovered. Without intermediate neutralization higher concentrations of zinc can only be recovered by dithiophosphorous acids. Figure 7 shows the equilibrium conditions for the extraction of zinc and nickel with bis(2-ethylhexyl) dithiophosphoric acid. It can be seen that zinc can be extracted even at a pH of about 0.5, which is equivalent to an extraction of about 10 g/l zinc. This means that about 10 g/l zinc can be recovered from the feed phase without having an intermediate neutralization or, as is often done in solvent extraction, using the sodium salt of the extractant to maintain the pH. It is obvious that the acidic strength for the stripping reaction increases in the same order. In the case of dithiophosphoric acid, at least 4n HsS04 is necessary to re-extract zinc from the organic phase. The extractant concentration is usually 2%--4%; higher concentrations do not enhance the transfer rate. As surfactant, ECA 4360 is preferred to SPAN 80, mostly due to lower osmosis; concentrations of 1%-2% are sufficient to obtain a stable emulsion, even if the viscosity of the diluent is as low as l-2 cP. In the last few years there have been several pilot plants, which have all proved to be very successful [22]. Now the first large-scale plant is under construction and should go into operation in Spring 1986. This plant is built by Lenzing AG in Austria for the recovery of zinc from spin baths in viscose production. The problem there was to separate 400-600 ppm zinc selectively from calcium in a waste water stream which also contains 5-8 g/l HzS04, 25 g/l Na2S04, many modifiers like long-chain amines, and some solid many modifiers like long-chain amines, and some solid_ particles. Several processes, such as solvent extraction, ion-exchange resins and precipitation were tested, but the emulsion liquid membrane technique proved to be the most economical. During the pilot plant experiments, which lasted more than two years, several extractants and surfactants were tested and all other parameters optimized. Owing to the high acid content in the waste water only dithiophosphinic or dithiophosphoric acid can be used, but dithiophosphoric acid is much cheaper (l-2 U.S.$/kg). SPAN 80 was first used as the surfactant (no osmosis because of high ionic strength in feed), but there was trouble with bacteria, which destroyed the surfactant and extractant. This was not observed with ECA 4360 but now the problem might be the inverse: the extractant bis(2-ethylhexyl)dithiophosphoric acid is only very slowly decomposable by bacteria and if high amounts of this extractant are dissolved in the feed phase, waste water problems could result. But since, under the given conditions, only 3 ppm of the extractant are soluble in the waste water, no problems should occur. In the pilot plant an average separation of more than 99.5% could be achieved and a concentration in the receiving phase of about 60 g/I. This phase is further concentrated by evaporation and can be reused directly in the process stream. In a first stage the plant was built for a throughput of 75 m3 h-r of waste water, the organic membrane phase having a throughput of 5 m3 h- and the receiving phase 0.5 m3 h- If this plant works efficiently, . an extension to a throughput of 300 m3 h- is planned. Figure 8 shows a schema of the electrostatic splitter which will be used in this plant. It is an emulsion splitter with uninsulated electrodes and consists essentially of a settler, where one electrode is immersed in the organic phase and the second electrode is represented by the earthed aqueous phase. The distance between electrode and interphase is l-5 cm and the applied voltage is 15-20 kV.
20 5 kV kWh/m> emulsion

% Extraction 100

emulsion

_I

organic

phase

interface

-II,

aqueous phase

Fig. 8. Electrostatic

splitter with uninsulated

electrodes.

0.5

lo

H&

hl

Fig. 7. Equilibrium for Zn and Ni extraction. Aqueous phase: 1 g/l Zn, Ni; organic phase: 5 wt.% di(2-ethylhexyl)dithiophosphoric acid, 95 wt.74Shellsol T; phase ratio: 1 :l.

In Fig. 9 a flow sheet of the plant is shown with the throughputs and concentrations. As the mass transfer apparatus, a counter-current column with a diameter of 1.6 m and an active height of 10 m will be used. As was demonstrated by the various pilot plants, zinc separation with emulsion liquid membranes works very efficiently and more plants will certainly be built in the viscose industry. Whether there will be more plants in other industries, like in the hydrometallurgical industry, will mainly depend on the impurities, because

326

Fig. 9. Flow sheet of a plant for Zn recovery: 1, waste water, 75 m3 h-l, 500 mg/l 2x1; 2, waste water, 75 m3 h-l, 1-3 mg/l Zn; 3, receiving phase, 0.5 m3 h-, 250 g/l HzS04; 4, receiving phase, 0.5 m3 h-, 60 g/l 2x1; 5, organic phase, 5 m3 h-l; 6, column; I-1, homogenizer; emulsion, 5.5 m3 h- l; P, permeation S, electrostatic splitter; T, tank for organic phase.

sheet of a pilot plant for the recovery of Cu: 1, feed, 800 mg/l Cu; 2, raffinate. 5 mg/l Cu; 3, receiving phase, 50 g/l Cu; 4, spent electrolyte, 30 g/l Cu; 5, organic phase; 6, emulsion; P, permeation column; S. electrostatic splitter; H, homogenizer; E. electrolysis.

Fig. 10. Flow

the organophosphorous extractants are not very selective. Problems may be caused when Ni, Co, Fe, Cu, Cd are present, especially when the zinc in the concentrated solution should be taken by electrolysis. Copper permeation Copper is the element which has been investigated most in solvent extraction as well as in liquid membrane permeation. Many extractants have been developed, which cover a wide pH range. In liquid membrane permeation the more acidic extractants, like LIX 70 or the ACORGA reagents, are the most important in the separation of copper from acidic solutions. The features the mechanism, selectivity, of copper permeation, modelling and economics are described elsewhere [10,21, 131; here, a specific case, the recovery of copper from the flue dusts of a copper recovery plant, is described [23]. These flue dusts contain mainly zinc and copper; of the impurities, only Sn is of some value. Leaching is done with sulphuric acid and two different solutions are obtained for the copper recovery with emulsion liquid membranes: (a) 100 g/l Zn, (bj 10 g/l Zn, 8 g/l Cu 0.8 g/l Cu

From these solutions, Cu should be separated selectively and enriched to concentrations where electrolysis is possible. A flow sheet of the pilot plant is shown in Fig. 10. The data given therein are valid for solution (a). The compositions and throughputs of the phases were as follows. Feed 200 l/h: 100 g/l Zn 8 g/l cu pH 2.0 2 wt.% ECA 4360 5% ACORGA PT 5050 93% Shellsol T 30 g/l cu 2.50 g/l sulphuric

Membrane

phase

120 I/h:

Receiving

phase

60 l/h:

acid

The spent electrolyte was used as receiving phase. In this phase Cu was concentrated from 30 g/l to 50 g/l and recovered by electrolysis. The zinc concentration in the receiving phase remained constant at 20 mg/l, so the selectivity was sufficient. In the feed phase the copper concentration could be lowered from 8 g/l to 0.040 g/l, The zinc and all the impurities were not transported and remained constant. During the pilot plant experiments, which lasted several months, two major problems had to be faced. The first was that the separation of 99.5% of the copper down to 40 mg/l was not sufficient, since 5 rng/l were required for the following treatment of the zinc electrolyte. After 2 m of the permeation column, which had an active height of 6 m, 40 mg/l had already been obtained. This meant that the maximum separation had been achieved and no better separations were possible. When a copper-free 250 g/l sulphuric acid solution was used as the receiving phase, separations down to less than 3 mg/l could be obtained, but with the preloaded solution the equilibrium condition allows no better separation. When solution (b) with 0.8 g/l Cu was used as the feed phase, separations down to less than 5 mg/l could be achieved, although the same preloaded receiving phase was used. This is due to the different proton concentration in equilibrium, because the pH drop is less than in the solution with 8 g/l copper. The second problem was the decomposition of the ACORGA PT 5050 reagent, which was much more dramatic than in solvent extraction plants. It was proven that the surfactant ECA 4360 acted as an accelerator for the decomposition of the extractant. The chemistry of this acceleration is not quite clear, but many surfactants show the same bchaviour. The following investigations therefore centred on the search for a suitable surfactant. Table 2 shows the effects of some surfactants on stability, osmosis, mass transfer resistance and on the decomposition of the extractants. It can be seen that the surfactant PA 18 satisfies all these demands. The decomposition of the extractant is in this case of the same order of magnitude as is usual in solvent extraction, where no surfactant is used.

327
TABLE 2. Influence Surfactant of surfactants on copper permeation Mass transfer resistance ++ t _ Osmosis Break-up (stability) +* + t+ ++ ++ f+ ---_ + + _ + + f+ -, not suitable; --, absoluteIy Decomposition of extractant (ACORGA PT 5050) ---++ _ +

F.CA 4360 SPAN 20 SPAN 40 SPAN 60 SPAN 65 SPAN 80 SPAN 85 TWEEN 20 TWEEN 40 TWEEN 60 TWEEN 65 TWEEN 80 TWEEN 85 ANTAROX CO210 ANTAROX CA420 ANTAROX CO430 PA 18 ++, very suitable for application

+ -+ ++ _

_ _ ++ in copper permeation;

_ -+ --+ +, suitable;

_ ++ ++ + + not suitable.

Nickel The application of nickel permeation to the hydrometallurgical industry is restricted, since the selectivity of the nickel extractants to other metals is very poor. But in solutions where no disturbing impurities are present, like in the electroplating industry, the recovery of nickel by liquid surfactant membranes might gain importance. Pilot plant experiments with synthetic solutions brought very promising results [24]. In a two-stage mixer-settler the nickel concentration could be reduced from an initial concentration of 2.2 g/l to about 200 mg/l in the first stage and to 3 mgjl in the second stage. The maximum concentration which was obtained in the receiving phase was 52 g/l. This is less than the expected value of more than 80 g/l and is due to a problem which occurred during the experiments: unlike in zinc permeation, the surfactant ECA 4360 represents a considerable mass transfer resistance and cannot be used alone. Addition of SPAN 80 decreases this mass transfer resistance at a constant emulsion stability, but this addition causes swelling of the emulsion by water transport from the outer to the inner aqueous phase and thus the concentrated inner phase is diluted again. In Fig. 11 these connections are shown. It can be seen that the mass transfer resistance is decreased with increasing SPAN 80 concentration. On the other hand, osmosis is increased in the same order. A maximum enrichment in the inner phase can be obtained at a SPAN 80 concentration of 0.2%, where the mass transfer resistance is already substantially lowered and the water transport has not yet increased too much. As can be seen from Fig. 7, the extraction equilibria of zinc and nickel are about the same. But the kinetics, especially the kinetics of the stripping reaction, is much lower for the nickel transport. Therefore multistage mixer-settlers with sufficient residence times are

Ni ,,(

Ni, [g/II

% Dsmosir

0.2

o.5

SPAN8OW%l l, Nil;

11. Influence of SPAN 80 on nickel 0, NiIII; 0, osmosis; t = 10 min.

Fig.

permeation:

necessary for a complete separation of nickel. Comparative studies in a counter-current column showed that under given conditions the residence time is not sufficient for a complete separation of nickel, although zinc can be separated completely even for a smaller column length (see Fig. 12). This also illustrates the importance of the mass transfer resistance of the surfactant. The process can only be optimized if a surfactant is found which causes no water transport and provides a stable emulsion without having a considerable mass transfer resistance. Anion permeation There have been many promising reports on the permeation of anionic 0x0-complexes of various metals,

328 A major drawback of the permeation of anions is the poor selectivity of the amines to different anions. It is true that sulphate ions are less transported than a lot of other anions, but if the concentration of the anion to bc transported (e.g. WO,* , MoOd2-, Crz0T2-, ReOg-, . ..) becomes very low and the sulphatc concentration is high, the sulphate transport may become dominating. This is why some of the above-mentioned anion complexes cannot be separated completely from sulphate-containing industrial solutions. For instance. in sulphate (chloride) milieu chromium can easily be separated down to 3-5 ppm, but in most countries this value is too high to allow the solution to pass to the draining canal. Less than 1 ppm (preferably 0.1 ppm) is required and this value can only be obtained when most of the sulphate (and/or chloride) ions have been transported. Since every sulphate (chloride) ion consumes an equivalent amount of hydroxide ions in the inner phase, the separation of chromium to less than 1 ppm is mainly an economic and not a technical problem.

Fig 12. Separation

dvarious

metals in a pilot plant.

especially of chromium [25, 261, but for industrial use a lot of: problems have to be overcome. Normally tertiary amines like Alamir 336 or Hostarex A327 are used for the permeation of anions; primary and secondary amines show lower stability of the emulsion, and quaternary ammonium salts are only used when tertiary amincs are not applicable. An example of the last case is the permeation of vanadate, which is best done in the pH range 6-8, where tertiary amines are not effective. The stripping agents are usually the same as in solvent extraction with some restrictions. Ammoniacal solutions with a pH of more than I1 cannot be used because of the quick transport of ammonia across the membrane. Instead of ammonia, NaOH can be used. The best stripping agents in emulsion liquid membranes ale probably chloride solutions like NaCl or NH4CI, but they are seldom desirable for industrial application. Nitrates and carbonates may also he used, but sulphates are not applicable. Corresponding to the following equations the permeation of anions A- can be according to the co- or counter-transport mechanism: Co-transport: RSN + H+ + A- -r R,NHA + HZ0 Counter-transport: RSNHOH + A- e RANHA + OH+ OII- -

9. Conclusion Tn many successful pilot plants it can be proved that the emulsion liquid membrane technique has developed into an industrial process. In particular, the separation of zinc and copper has been investigated to such an extent that no major problems can be expected in the operation of a large-scale plant. The first large-scale plant with a throughput of 75 m3 h- has been built for the Austrian viscose industry fol- the recovery of Lint from spin baths. For the separation of other metals, some problems still exist which have to be solved prior to industrial application. The surface active agent has, up to now, only been chosen according to the stability of the emulsion. But this is only one factor in the choice of a suitable surfactant; others are the capability of transporting water, the mass transfer resistance, solubility in water, the catalytic effect on the decomposition of the extractant, and the resistance to an attack of bacteria. If surfactants can he found which meet the demands, several more large-scale plants for the separation of various metals can be expected in the near future.

__-

R3NHA

RBN + A

Nomenclature
A. B C H n N substance A, B concentration height of column normality stirrer speed time volume carrier

It is not quite clear which mechanism takes place, but there is no influence on the effectiveness of the process; the difference between these two mechanisms is given only by a small amount of water. Usually 1% of amine in the membrane phase is sufficient to ensure a quick and complete separation of the anion. Higher amounts of extractants in the membrane phase are unfavourable. because the water transport through the membrane is accelerated, whereas the transfer rate of the anion is only slightly enhanced.

t
V X

Superscripl
V

stoichiometric

coefficient

329 Subscripts I II
III
9 D. Lorbach and R. Man, Chem. Eng. Process., in press. 10 F. Nakashio and K. Kondo, Sep. Sci. Tech., 15 (4) (1980) 1171-1191. 11 .I. Draxler and R. Marx, Ber. Bunsenges Phys. Chem., 86 (1982) 64-68. 12 M. B. Dines, U.S. Patent 4337225, 1982. 13 N. N. Li, R. P. Cahn, D. Naden and R. W. Lai, HydrometaL lurgy, 9 (1983) 277-305. 14 N. N. Li, U.S. Patent 4125461, 1978. 15 N. N. L& T. Hucal and R. P. Cahn, U.S. Patent 4001 109, 1971. 16 G. A. Davies, U.S. Patent 4283290, 1981. 17 H. B. Hauertmann and K. Schtlgerl, Paper presented at ACHEMA 85. DECHEMA, Frankfurt, 1985. 18 A. Kopp, Thesis, TU Graz, 1979. 19 T. A. Hatton, E. N. Liihtfood, R. P. Cahn and N. N. Li, lnd. Eng. Chem., Fundam.. 22 (1983) 27 -35. 20 W. Gaubinger, G. Husung and R. Marr, Chem. Ing. Tech., 54 (1982) 850. 21 J. W. Frankenfeld, R. P. Chan and N. N. Li, Sep. Sci. Tech., 14 (4) (1981) 385-402. 22 M. Protsch and R. Marr, Proc. Int. Solvent Extraction Cant, Denver, CO. 1983, Am. Inst. Chem. Eng., 1983, pp. 66-67. 23 R. Wachter, Thesis, TU Graz, 1986. 24 H. Lackner, Thesis, TI; Graz, 1986. 25 A. M. Hochhauser and E. L. Cussler, AIChE Symp. Ser., 71 (152) (1975) 136-142. 26 X. Li and Y. Zhang, fioc. JT Meet. Chem. Eng.. Vol. 2, Chem. Ind. Eng. Sot. China and Am. Inst. Chem. Eng., 1982, pp. 571-582.

th bm

phase I = receiving phase phase II = membrane phase phase III = feed phase theoretical backmixing

References
N. N. Li, U.S. Patent 3410 794, 1968. P. R. Danesi, Sep. Sci. Tech., 19 (11 & 12) (1984/85) 857-894. M. Teramoto and H. Tanimoto, Sep. Sci. Tech., I8 (10) (1983) 871-892. K. Schneider and T. H. Ringelen, Membranverfahren in der Umwelttechnik, G VC. VDI-Gesellschaft Verfahrenstechnik und Chemieingenieurwesen, Postfach 1139, 4000 Dusseldorf, 1985, pp. 119-131. L. Boyadzhiev, Proc. 3rd Austrinn-Italian-Yugoslav Chemical Engineering Conf (CEC), Graz, 1982, Vol. II, Inst. Chem. Eng., TU Graz, 1982, pp. 349 -357. A. Kriechbaumer and R. Marr, ACS Symp. Ser., 272 (1985) 381-398. H. H. Downs and N. N. Li, J. Sep. Process. Technol., 2 (4) (1981) 19-24. W. Halwachs, E. Flaschel and K. SchUgerl, J. Membr. Sci., 6 (1980) 33-44.