Removal of valeric acid from wastewaters by membrane contactors

Removal of valeric acid from wastewaters by membrane contactors

journal of MEMBRANE SCIENCE ELSEVIER Journal of Membrane Science 137 (1997) 45-53 Removal of valeric acid from wastewaters by membrane contactors M ...

552KB Sizes 0 Downloads 6 Views

journal of MEMBRANE SCIENCE ELSEVIER

Journal of Membrane Science 137 (1997) 45-53

Removal of valeric acid from wastewaters by membrane contactors M . R o d r i g u e z a, R . M . C . V i e g a s b, S. L u q u e a, I . M . C o e l h o s o b, J . P . S . G . C r e s p o b, J . R . A l v a r e z ~'* ~'Department of Chemical Engineering and Environmental Technology, Univer~ity of Oviedo. 33071 Oviedo. 5"pain hDepartment of Chemistry. Faculdade de Ciencias e Tecnologfa, Universidade Nova d~' Lisboa. 2825 Monte de Caparica. Portu.~a/

Received 10 March 1997; received in revised form 14 July 1997: accepted 16 July 1997

Abstract Membrane contactors, providing a non-dispersive extraction technique, were used for the removal of valeric 01-pentanoic) acid from synthetic aqueous solutions simulating an industrial wastewater from polymer manufacturing. Amberlite LA-2 (secondary amine) in toluene was chosen as the extraction system. Equilibrium conditions were determined and mechanistically modelled for different extractant concentrations allowing the further calculation of mass transfer coefficients. The influence of the hydrodynamics of both the aqueous and organic phases on the overall mass transfer coefficient, calculated through two proposed methods, was studied. The integration of extraction and backextraction was also carried out, allowing a further acid removal with lower extractant concentrations. ~(, 1997 Elsevier Science B.V. Kevwords: Valeric acid; Membrane contactors; Hollow fibres; Solvent extraction; Mass transfer coefficient

1. I n t r o d u c t i o n Low molecular weight carboxylic acids appear in several aqueous industrial waste waters (petrochemical, polymer, pulp and paper and pharmaceutical) for which the removal and recovery of the acid is necessary due to both economical and environmental aspects [ 1-3]. Several processes have been proposed for the removal of these acids, i.e. precipitation, adsorption, membrane processes and solvent extraction. Precipitation shows a lack of selectivity and adsorption is often not efficient enough because of the small amount of adsorbent. Membrane processes have the disadvanrages of low selectivity and fouling [1-4]. Solvent extraction represents a suitable choice for intermedi*Corresponding author. Fax: +34 485 103 434. 0376-7388/97/$17.00 1997 Elsevier Science B.V. All rights reserved. P l l S0376 7 3 8 8 ( 9 7 ) 0 0 1 8 2 - 8

ate concentrations. The use of reactive extractants enhances the selectivity of the extraction due to a specific chemical reaction between the solute and the extractant 13,4]. Membrane contactors represent a new way to carry out separation processes like gas absorption and solvent extraction. They are commonly hollow tibre modules used as substitutes for packed towers. Extraction with hollow fibre modules is fast due to the large inteffacial area per volume. The interface is stabilised at the membrane pores, avoiding emulsification. Dispersion-free solvent extraction has been recently studied and proved to have several advantages over conventional extractors: high surface area per volume, no need for density difference between phases to achieve phase separation, no limitations of loading or flooding, ability to handle particles and systems that emulsify readily, no need of agitation or moving parts,

46

M. Rodriguez et al./Journal of Membrane Science 137 (1997) 45-53

and ability to provide several extraction stages in a single piece of equipment. The main disadvantage is the lower mass transfer coefficients, due to the resistance to mass transfer in the pores of the membrane. This disadvantage is overcome by the high membrane contact area [5-8]. The aim of this study is the assessment of the viability of a continuous process of valeric acid removal using an extraction-backextraction process. For this purpose, the effect of hydrodynamics on the overall mass transfer coefficient and on the operation stability were studied for both the aqueous and organic phase using an aqueous valeric acid solution simulating a wastewater stream from polymer manufacturing,

2. Experimental 2.1. Materials Synthetic solutions of valeric (n-pentanoic) acid in demineralized water (5 g/l), simulating a wastewater stream from polymer manufacturing, were tested, Amberlite LA-2 (Fluka) (10 vol%) in Toluene (Merck, > 99%) (90 vol%) was chosen as extraction system according to previous research on the topic [4]. Liqui-Cel Laboratory hollow fibre modules with [email protected] membranes from Hoechst Celanese Corporation have been used in this study. Each module contains 2100 hydrophobic polypropylene fibres with a nominal internal diameter of 240 pm, nominal thickness of 30 p~m, 16 cm long and with an average pore size of 0.05 ~tm. Each module provides an effective area of 0.23 m 2. 2.2. Methods 2.2.1. Determination of extraction isotherms Extraction equilibrium isotherms were determined for several extractant concentrations, up to l0 vol%, using the extractant with toluene as diluent, or just toluene, by contacting known volumes of the organic phase with known volumes of the aqueous phase containing 0.06 M valeric acid. The total volume of the sample, adding up to 30 cm 3, was placed in screwcapped flasks and mechanically mixed and thermostated in a water bath at 25°C. Data at higher acid concentration were obtained by contacting known

volumes of an aqueous phase containing 0.30 M valeric acid (solubility limit concentration) with known volumes of the organic phase (toluene or toluene + Amberlite LA-2, 10 vol%) which were either fresh or preloaded with valeric acid using the aforementioned method. Once the two phases had settled, the concentration of valeric acid was measured in both phases. 2.2.2. Membrane contactor experiments The aqueous phase was pumped through the lumen of the fibres while the organic phase flowed on the shell side of the module. Due to the hydrophobic nature of the fibre a slight overpressure (0.2 bar) was applied to the aqueous phase in order to stabilise the interface within the membrane. All the experiments were carried out in co-current flow. The system was maintained at 25°C using feed jacketed glass beakers for both the aqueous and organic phases. Connecting tubing was made of Viton 'R due to problems of compatibility with toluene. Because of the extremely fast kinetics of extraction, the aqueous solution volume had to be ten times larger than the organic volume, The experiments were designed to analyse the influence of the flow rate of both organic and aqueous phases on the overall mass transfer coefficient. The evolution of the concentration of the acid in the aqueous phase with time was measured under different hydrodynamic conditions. The Reynolds number of the aqueous stream was set at a determined value and experiments were carried out varying the Reynolds number of the organic phase. Afterwards, the organic phase hydrodynamics were fixed, and the aqueous phase flow rate was varied. Integrated continuous extraction-backextraction experiments using different extractant concentrations were also carried out to evaluate their influence on the degree of removal achieved. The integrated system set-up is shown in Fig. I. The studies on the hydrodynamic effect were carried out with the same equipment, using only one hollow fibre module. 2.3. Analytical methods 2.3.1. Determination of extraction isotherms The concentration of valeric acid in the aqueousphase was determined by acid-base titration with a

M. Rodr{guez et al./Journal of Membrane Science I37 (1997) 45-53

....~:'~":::~"~?

:z:~::'~:~: ........

brane, and organic phases, as shown in Eq. (2):

xa - ka + ~

I ~,~r~r q

t

Fig. 1 Extraction-backextractionexperimental set-up,

standard sodium hydroxide solution (0.01 or 0.1 M), using phenolphthalein as indicator. The organic-phase acid concentration was determined by two-phase titration (1 cm 3 of the organic solvent mixed with 7 cm 3 of ethanol and 3 cm 3 of distilled water) with a NaOH solution (0.01 or 0.1 M), using phenolphthalein as indicator [4]. The mass balance closure with respect to the organic acid concentration was around ±5%~ and in most cases within 2%. 2.3.2. M e m b r a n e contactor experiments The concentration of valeric acid in the aqueous phase was determined by HPLC using a refractive index detector (Merck Hitachi, Japan). The column used was a Shodex SH 1011 (Showa Denko K.K., Japan) and the eluent was 0.01 N sulphuric acid. In order to determine the Reynolds number of the organic phases their viscosity was measured using a Brookfield viscometer (model LVTD-II digital). The density was gravimetrically determined, 2.4. Mass transfer coefficient calculation

The concentration of the acid in the aqueous phase was measured as a function of time to determine the overall mass transfer coefficient, K~, defined by: J~ol = K~ "Am' (c/~ /

c'~/)

47

~1)

where J~,,~ is the solute flux, A m is membrane area, C(a~ is the concentration in the aqueous solution at the time t, and ~'*(a,,is the equilibrium acid concentration in the aqueons phase [9,10]. The overall mass transfer coefficient can be considered as the contribution of three terms. Since I/K~ is the overall resistance for mass transfer, it can be expressed as the contribution of the aqueous, mere-

+koD

~2)

If all the individual coefficients km and k,, are referred to the aqueous concentration, they have to be multiplied by the distribution coefficient D. In our case, D is considered to be constant in the range of valeric acid concentrations studied. In the case of taking into account non-constant D over the process, another approach should be considered [12]. By making mass balances to the system, assuming a constant distribution coefficient, and combining the obtained equations, an exponential dependency of the concentration of the solute in the aqueous phase with time can be established. The key equations resulting for the co-current flow are: Vc~.~lo ÷ c~alo e a~ c(~,3 1+ V 1+ V

V,~ 1 + Q

Q~,

~3) V~ V - VoD"

Qa Q - QoD

c(~)0 being the initial acid concentration in the aqueous phase, Q~, and Qo the aqueous feed and organic phase flowrates respectively. V~ and V~, are the volumes of the aqueous and organic phases. Am is the effective area provided by the module and D is the distribution coefficient determined by the equilibrium data, and assumed constant throughout the experiment [10,111.

3. Results and discussion 3.1. Equilibrium isotherms

Before the membrane contactor experiments were carried out, equilibrium conditions had to be determined as a basic study for mass transfer coefficient calculations. The most suitable extraction system for the removal of valeric acid was found to be formed by Amberlite LA-2 (secondary amine)diluted 10 vol% in toluene [4].

48

M. Rodrfguez et al./Journal of Membrane Science 137 (1997) 45-53

2.0

Table I Calculated values for the parameters of the model

1.5

Extractant system

Parameter

Value

Toluene (diluent alone)

P KD

2.0 15.4

Amberlite LA-2/toluene

K~I K~2 Ks3

1123_8% 4253_8% 49.5±3%

2, ~

1.0

0.5 0

v-

0

0.05

0.1

0.15

IHVl ~=~(tool L -l) Fig. 2. Experimental data and model prediction for the equilibrium of valeric acid with Amberlite LA-2 in toluene at 25"C. Extractant concentration: Q): 0, &: 2.5, 1~: 5, 0 : 1 0 vol%. Solid lines: model

prediction. Fig. 2 shows the equilibrium data obtained when this extraction system is contacted with valeric acid solutions. A remarkable degree of extraction is observed when only toluene is used, regarded as physical extraction by the diluent. The contributions of physical and reactive processes are of the same order at an acid concentration of about 0.06 M, and above that concentration physical extraction becomes more important, A mechanistic model has been developed to predict the equilibrium relationship, taking into account the addition of physical extraction by the diluent (toluene) followed by a dimerization of the acid in the organic phase, and reactive extraction of the amine. Overloading of the amine, forming complexes with 1, 2 and 3 molecules of acid per amine molecule is also considered. Eq. (4) gives the relation between the concentration of the acid in the aqueous phase and in the organic phase: Cnv~o, = ~a(P[HV](~)+2p2Ko[HV]~a))+CR~o! 2 3 Ks l [HV] (a) ÷ 2Ksl Ks2 [HV] (a) + 3Ksl Ks2Ks3 [HV] (a) x 2 3 I + K ~ [HV](a)+Ks1K~2[HV](~)+K~IK~2K~3[HV](a) (4) where (~d represents the volumetric fraction of the diluent in the extraction system, P the partition coefficient between the aqueous phase and the diluent, and /i'D the dimerization constant of the acid in the organic phase. Cg(o~is the molar concentration of the extractant in the organic phase and K~, Ks2 and Ks3 the formation

constants of the three considered complexes respectively. Further explanation of the model can be found elsewhere [4]. This model proves to predict satisfactorily the equilibrium isotherms under different extractant concentrations. Fig. 2 shows the equilibrium isotherms for three different extractant concentration and the prediction of the model. The values of the different parameters of the model have been determined using the LevenbergMarquardt algorithm, and are shown in Table 1. The fitted parameters for different extractant concentrations are in the range of the values shown in the table.

3.2. Mass transfer coefficient calculation Two methods have been used to determine Ka, from the experimental data. Both methods assume the exponential expression given by Eq. (3), and thus, a constant distribution coefficient. There is a maximum value of the distribution coefficient at very low concentrations, but it remains constant at higher aqueous concentrations. In this work the distribution coefficient is assumed constant with a value corresponding to the lowest valeric acid aqueous concentration reached (D=50). This means that at very low concentrations the overall mass transfer coefficient should be more sensitive to the hydrodynamics of the system and the resistances of the membrane and the organic phase would be lower. As it is shown afterwards, the agreement of the experimental values with the model will validate the assumption made.

3.2.1. Method 1 According to Eq. (3), the concentration of the acid in the aqueous phase varies with time in the form of an

+<

M. Rodr[guez et al./Journal o~ Membrane Science 137 (1997) 45 53

49

-'-" -0 00(i4 ---" 004

:,a 1 ) 0 2 -i)<1() /

II II

i

~

~

100

2(X)

300

00f)16

400

,

tl

t (rnin)

i

,

0.01

i

0.02

~

,

I)0;

,

~

IJli4

,

I)~) c,

c~,~ -cm( ( tool L ~)

Fig. 3. Valeric acid concentration as a function of time. Method I. Re~, - 5.41: Re,, 1.42. Q : Experimental data; Solid line: model. K~, ~4.72 10 ~ ' m s t.

Fig. 4. Mass transfer coefficient calculation through method [I. Re,~ -- 5.41; Re,, - 1.42. O : Experimental data: Solid line: model. K~,-- 4.72 >< 10 ~ ' m s t.

exponential equation ci~u = a + h e x p ( - c t )

driving force (cla > c~,>).Plotting the calculated data ~" - a straight line is obtained. The dcla~/dt vs. (c/,,:, _ C~:~l/) slope of this line accounts for K.,. The value of c t+a ) • varies throughout the experiment due to the valeric acid depletion in the aqueous phase with time. therefore c*
(5)

where a and b depend only on known parameters, and thus. their values can be determined in advance. By fitting the experimental data to this exponential equation best values for c are obtained. Fig. 3 shows the model and experimental data for the evolution of acid concentration in the aqueous phase with time. Ka is then determined by substituting the value of the other parameters in Eq. (3): Ku

Q~ In [1 - c ( V a ) (1 + Q~ ] (1 ~ Q)Am Qa \ ~ J

(6)

3.2.2. Method H In this method, experimental data are fitted to an exponential like equation without any kind of restriction. Best values for a, b and c are so obtained, By deriving Eq. (5) the following expression is obtained:

dc'{~,~ _

bcexp(-ct) (7) dt By using these values of b and c, it is possible to calculate pairs [dc(al/dt, t] according to Eq. (7). Considering the mass transfer equation, V,,

de(,, I

Am

dt

K~,(c(~,I

c~l )

(8)

dc~,~/dt is linearly dependent on the concentration

3.3.1nfluenceof coe~cientthehydrodynamicson • The hydrodynamics of both phases have been found to have a remarkable influence on the mass transfer coefficient. Increasing flowrates on any of the two phases yield higher coefficients until a maximum is reached as shown in Figs. 5 and 6. The two proposed methods for the calculation of the overall mass transfer coefficient provide values of the same order, the values obtained with the first method being slightly higher. The two methods differ mostly at the highest Reynolds numbers. Taking into account that the total mass tranfer resistance is the addition of three terms. as shown in Eq. (2), and that the overall mass transfer coefficient reaches an asymptotic value when increasing Reynolds number, the contribution of each resistance could be evaluated. For both phases the asymptotic value of K~, is 0.9×10 5 m s 1 This means that fixing a given Reynolds number for the organic phase, increasing

M. Rodr{guez et al./Journal of Membrane Science 137 (1997) 45-53

50

t • o

0.8

=" ~ 0.6

• •o

% "~ 04

changes in turbulence would change the overall mass transfer coefficient. If the membrane resistance was higher than that corresponding to the phases, hydrodynamics would not play any role in the improvement of Ka. In the case of valeric acid, the membrane resistance keeps a low value, or similar to the resistance of both phases due to the high distribution coefficient in the extraction process.



o



°

~o o.2 ~ 0 0

n

n

n

i

10

20

30

40

Rea

50

Fig. 5. Overall m a s s transfer coefficient as a function o f Re~; Reo = 1.42. , : M e t h o d I; © : M e t h o d II.

• •



0.8

4"

0.6

o

o



o

% •o "~ 0.4 . o ~ o.2 0

0

n

a

2

4

,

n

6

~

~

8

i

10

Reo Fig. 6. Overall m a s s transfer coefficient as a function of Reo; Rea = 11.9. , : M e t h o d I; © : M e t h o d II.

3.4. Integrated extraction-backextraction process The integration of the extraction and backextraction processes in a simultaneous operation has been studied. NaOH aqueous solutions, 100% in excess, were used as backextractant. NaOH reacts with the acid in the organic phase forming the subsequent sodium salt, which is transferred to the aqueous backextractant phase. It has been found that the integrated process allows us to almost completely exhaust the acid from the aqueous phase, as shown in Fig. 7. In this integrated process, it was possible to remove valeric acid from the aqueous feed to a higher extent than in previous extraction experiments, since it was not accumulated in the organic phase, but subsequently transferred to the backextractant solution. Due to the large excess of NaOH, all of the acid initially transferred to the organic phase is reextracted, and thus, the organic phase is continuously regenerated.

turbulence in the aqueous phase would give zero resistance in the aqueous phase, and therefore

0.00

0.6

1/Ka = l/(Dkm) + l/(Dko). On the other hand, working at a fixed Reynolds number and increasing turbulence of the organic phase, zero resistance would be obtained in the organic phase, and then, 1/Ka = 1/(ka) + l/(Dkm). The values of fixed Reynolds numbers were Rea = 11.9 for the aqueous phase and Reo ---- 1.42 for the organic phase. An experiment carried out at these conditions would have the three terms, and Ka = 0.662 x 1 0 - S m s -1. This leads to k a = Dko = 2.44 x 10 -5 m s ~ and Dkm = 1.45 × 10 -5 m s ~. These results indicate that in the system used, hydrodynamics has a strong influence on membrane performance since resistances of both phases are of the same order as the membrane resistance, and any

.



o.s

• 0.04



o.4 0.3 -"~

_'~

~ ~" o.02 °o• o • o I? °, o ,o , ~,

0.2 ~ 0.1

0

0

100

200

o 300

400

t0,1.) Fig. 7. Valeric acid concentration in the feed O and in the backextractant • as a function of time. E x t r a c t i o n - b a c k e x t r a c t i o n continuous experiment. Rea 20.9; Re,, = 3.17; Reb = 20.96, A m b e r l i t e L A - 2 10 vol%. Volume ratio: a q u e o u s : o r g a n i c :

backextractant=10:1: 1.

M. Rodr{guez et al./Journal qf Membrane Science 137 (1997) 45-53

A concentration factor of 1 • 10 was achieved, and therefore, the concentration of valeric acid in the backextracting phase is 0.5 mol 1-1. The acid is in the fon'n of its sodium salt, acidifying this phase with a mineral acid it is possible to reestablish the ionic form of the acid, but due to its limited solubility in water (0.36 mol I 1) two phases are obtained, the one on the top being pure valeric acid. The influence of the concentration of the extractant (Amberlite LA-2) in the organic phase on the degree of removal of valeric acid has also been studied, Figs. 8 and 9 show the evolution of the concentration

006

- ~ " 0.04

=~ 0.02

% ~e o •• 0 • •

od w

• aJ o o Io

0

~

0

o g n

1O0

200

300

t (min) Fig. 8. Influence of extractant concentration (Amberlite L A - 2 ) o n the degree of removal of valeric acid in extraction-backextraction continuous experiments. Re~, -- 12; Re,, = 3.3; Reb = 12. 0 : 1 0 vol%: 0 : 5 vol%; i : 2 vol°A:. Volume ratio: aqueous : organic : backextractant=l() : I : 1.67.

51

of valeric acid in the feed and backextractant with time for different extractant concentrations under the same hydrodynamic conditions. It was found that the concentration of extractant does not influence the degree of removal of valeric acid. The continuous regeneration of the organic phase allows to use lower extractant concentrations yielding the same degree of removal. This fact represents a remarkable economical advantage. In these experiments the concentration factor achieved was 1 ' 6 and, therefore, the solubility limit was not reached. This was due to the higher aqueous phase volume used for the backextraction process. The use of an integrated continuous system, such as the one presented in this work, would allow the treatment of a waste stream using a small amount of organic phase corresponding to the hold-up of the membrane contactor. This system provides a way, to remove and concentrate pollutants for disposal or further treatment. Since there is a physical barrier for the organic phase, pollution due to the loss of organic phase is minimized. The low' flowrates used for the treatment indicate a low energy' consumption and therefore low operating costs. Membrane area would be the main factor for a scale-up design, but

due

to

the

high

surface/volume

ratio

of

the

hollow fibre contactors, the installation would be very, compact.

4. Conclusions (1.4

0~ "~

• •

0:

• • • • "o

g " 0.~

0

--|o .

¢



o

o

o

O

o

. . . ~00 2oo t (~.)

3oo

The following conclusions can be drawn from the present study: 1. The proposed mechanistic model predicts satisfactorily the equilibrium conditions under different extractant concentrations. 2. Overall nmss transfer coefficients are influenced by the hydrodynamics of both aqueous and organic phases. Higher coefficients are obtained for higher Reynolds numbers, until an asymptotic value is reached. 3. The two proposed methods for the estimation of

Fig. 9. Influence of extractant concentration on the backextraction of valeric acid during extraction-back extraction continuous experiments. Re~ = 12; Reo = 3.3; Reb = 12. O: 10 vol%; O: 5 vol%: i : 2 vol%. Volume ratio: aqueous : organic : backextractant

the mass transfer coefficient provide

= 1 0 : I :1.67.

coefficients.

similar values for

small Reynolds numbers. For the higher Reynolds numbers, the first method yields higher mass transfer

52

M. Rodr{guez et al./Journal of Membrane Science 137 (1997) 45-53

4. It is possible to carry out continuous extractionbackextraction of valeric acid, with almost complete removal of the acid from the feed, 5. In a continuous extraction-backextraction process the concentration of extractant in the organic phase does not influence the amount of acid removed, being possible to reduce the concentration of Amberlite LA-2 down to 2 vol%. The high distribution coefficient for this extraction system allows the use of lower extractant concentration reducing the operation costs. 6. The continuous extraction-backextraction process used is suitable for the purpose of environmental treatment of wastewater, providing a continuous regeneration of the organic phase and a concentrated stream that at proper conditions allows the recovery of valeric acid.

ka

Ks1, Ks2, K~3 P Q Qa Qo Rea Reb Re,, t V E~ Vo Od

Aqueous phase individual mass transfer coefficient, m s-1 Valeric acid-extractant complex formation constants Partition coefficient (physical extraction) As defined in Eq. (3) Aqueous phase flow, m 3 s -1 Organic phase flow, m 3 s-I Aqueous phase Reynolds number B a c k e x t r a c t i o n phase Reynolds number Organic phase Reynolds number Time, s As defined in Eq. (3) Aqueous phase volume, m 3 Organic phase volume, m ~ Volumetric fraction of diluent in organic phase

5. Notation

a, b, c Am C(a) C~b) C~a) C(a)o CHV(o) CR~o~ D

Parameters for exponential curve fitting Membrane area, m 2 Valeric acid aqueous phase concentration, mol 1-t Valeric acid backextraction phase concentration, tool 1-t Equilibrium valeric acid aqueous phase concentration, mol 1-~ Initial valeric acid aqueous phase concentration, mol 1Total valeric acid organic phase concentration, tool 1 ~ Extractant concentration in the organic phase, mol 1 t

Jsol

Distribution coefficient (global extraction) Non dissociated valeric acid aqueous phase concentration, mol 1-1 Solute flux, mol s i

Ka

Overall mass transfer coefficient,

[HV](a)

m s

Kd km

ko

I

Valeric acid dimerization constant Membrane individual mass transfer coefficient, m s 1 Organic phase individual mass transfer coefficient, m s -~

Acknowledgements The authors wish to thank the Direccidn General de Investigacidn Cientffica y T~cnica (Spain) and Conselho de Reitores das Universidades Portuguesas (Portugal) for the financial support given to this project through the programmes Acciones Integradas entre Espafia y Portugal, and Acc6es Integradas LusoEspanholas respectively.

References [1] A.S. Kertes, C.J. King, Extraction chemistry of fermentation product carboxylic acids, Biotechnol. Bioeng. 28 (1986)

269-282. [2] K. Abbasian, W. Degener, K. Schiigerl, Chances of reactive extraction of primary and secondary metabolites, Ber. Bunsenges. Phys. Chem. 93 (1989) 976-980. [3] K. Schtigerl, W. Degener, Recovery of low-molecular weight compounds from complex aqueous mixtures by extraction, Int. Chem. Eng. 32 (1992) 29-40. 141 S. Luque, J.R. Alvarez, C. Pazos, J. Coca, Recovery of valeric acid from aqueous solutions by solvent extraction, Solv. Extr. Ion Excb. 13(5) (1995) 923-940. [5] B.W. Reed, M.J. Semmens, E.L. Cussler, Membrane Contactors, in R.D. Noble and S.A. Stern (Eds), Membrane Separations Technology. Principles and Applications, Elsevier, Amsterdam, 1995.

M, Rodriguez et al./Journal of Membrane Science 137 (1997) 45 53 [61 E. L. Cussler, Hollow fibre contactors, in J.G. Crespo and K.W. B6ddeker (Eds), Membrane Processes in Separation and Purification, Kluwer Academic Publishers, Dordrecht, 1994. [71 J. Rodriguez, Contactores de membrana para separaciones bif~isicas fluido/fluido: 1 Generalidades. Propiedades fisicas mils significativas, lng. Qu/m, 9 (1994) 169-178. [8] R. Prasad, K.K. Sirkar. Hollow fibre solvent extraction: Performances and design, J. Membr. Sci. 50 (1990) 153-175. 191 N.A. D, Ella, L. Dahuron, E.L. Cussler, Liquid-liquid extractions with microporous hollow fibers, J. Membr. Sci. 29 (1986) 309-319.

53

[10] C.J. Tompkins, A.S. Michaels, S.W. Peretti, Removal of p-nitrophenol from aqueous solution by membrane-supported solvent extraction, J. Membr. Sci. 75 (1992) 277-292. [11] L. Dahuron, E.L. Cussler, Protein extractions with hollow fibres. AIChE J. 34 (1988) 130-136. [12] I.M. Coelhost~, J.RS.G. Crespo and M.J,T. Carrondo. Kinetics of liquid membrane extraction in systems with variable distribution coefficient, J. Membr. Sci.. 127 t 1997t 141152.